SDSM:Historical Changes: Difference between revisions

From SMUSwiki
Jump to navigation Jump to search
(Reformat text)
(Add item about decoupling the main menu builder and the report writer screen.)
 
(18 intermediate revisions by the same user not shown)
Line 21: Line 21:
[[#top|RETURN]]
[[#top|RETURN]]


== SDS Laravel: Fixes to Broken Behavior ==
== Fixes to Broken Behavior ==


=== 2024 Apr 18: User Preferences Screen Fails to Display ===
=== 2024 Apr 18: SDS Laravel: User Preferences Screen Fails to Display ===


'''Context:''' User <code>darren.duncan</code> attempted to visit their
'''Context:''' User <code>darren.duncan</code> attempted to visit their
User Preferences screen, example urls
''SDS Laravel'' User Preferences screen, example urls
<code>https://api.smus.ca/User/Preferences</code> (production) and
<code>https://api.smus.ca/User/Preferences</code> (production) and
<code>https://api-latest.smus.ca/User/Preferences</code> (preview).
<code>https://api-latest.smus.ca/User/Preferences</code> (preview).
Line 34: Line 34:
Preferences screen. In preview, the revealed underlying symptom was that
Preferences screen. In preview, the revealed underlying symptom was that
the code line in <code>PreferencesController</code> of
the code line in <code>PreferencesController</code> of
<code>$addresses = $user->guardian->addresses;</code> was dying with <code>Attempt to read property addresses on null</code>. The problem occurred for user <code>darren.duncan</code> but not for other users.
<code>$addresses = $user->guardian->addresses;</code> was dying with
<code>Attempt to read property addresses on null</code>. The problem
occurred for user <code>darren.duncan</code> but not for other users.


'''Expected Behavior:''' The User Preferences screen should always display,
'''Expected Behavior:''' The User Preferences screen should always display,
Line 53: Line 55:
placeholder message <code>Shows address info</code> instead of each
placeholder message <code>Shows address info</code> instead of each
individual address.
individual address.
[[#top|RETURN]]
=== 2024 May 21: SDS Laravel: Timeout Feature Broken In Sub-Directory Hosted Apps ===
''SDS Laravel'' has a security feature such that if a user is inactive for
a period of time, 15 minutes typically, they will be automatically logged
out of the app. The app displays a countdown timer on the top right hand
corner of the screen that ticks for every second of activity. Javascript
code running in the client web browser will normally reset the countdown to
the full amount whenever a user interface (UI) event occurs that it
considers continued user activity, such as mouse movement over the screen.
This feature is implemented partly by the web client invoking the
<code>/session</code> endpoint on the server.
That invocation fails on any ''SDS Laravel'' app instance that is hosted at
a web address which is a sub-directory of the base url (meaning it is based
at an address like <code>https://foo.com/bar</code> rather than at an
address like just <code>https://foo.com</code>). This is because the web
client is trying to unconditionally treat every app instance as if it is
hosted directly at the base url with respect to its attempts to invoke the
<code>/session</code> endpoint, and so it is invoking the wrong web address
for instances that are at sub-directory urls.
As a result, for broken app instances, while the first UI activity
indicating continued user activity will reset the countdown timer displayed
on screen to the user, all subsequent UI activity will have no effect on
the timer, and it will not reset, and the server will not be aware of this
activity. Only a full page load like clicking a link to a new page will
register as activity and reset the timer.
This breakage affects all ''SDS Laravel'' instances at
https://sdsdev.smus.ca which are in sub-directories.
The primary change of this task fixes the problem by making the web client
respect the actual location of the app when invoking <code>/session</code>.
Mainly it is a 1-line change in the Laravel Blade template file
<code>resources/views/layouts/main.blade.php</code> to use
<code><nowiki>url: '{{route('session')}}'</nowiki></code> rather than
<code>url: '/session'</code>.
An additional change of this task is to fix a problem where the
app's background image doesn't display for the same underlying reason. The
problem is that the web client is trying to load the image
<code>public/images/body.png</code> from the wrong location. The fix
updates 1 line in the CSS file <code>public/css/smus_custom.css</code> to
<code>background: url('../images/body.png')</code> from
<code>background: url('/images/body.png')</code>; the newly-relative url is
relative to the location of the CSS file itself.
There is still additional broken behavior related to static asset loading
like the background image example, affecting custom fonts for example, but
these references are in the generated file <code>public/css/app.css</code>
as parts of third-party dependencies, and so these were left alone.
[[#top|RETURN]]
=== 2024 Jun 12: SDS Laravel: Form Meister Screen Fails to Display ===
This task made 2 distinct changes.
The first change...
The ''SDS Laravel'' screen Admin Menu -> Online Forms -> Form Meister fails
to display when visited. In production it displays a generic 500 internal
server error message. In preview, the revealed underlying symptom was an
attempt to invoke the non-existent PHP class
<code>App\Models\System\Form</code>.
This task fixed the offending PHP class reference to instead refer to the
existing PHP class <code>App\Models\Abstract\Form</code>. The result was
that the Form Meister screen then displayed when visited.
(Note that the ''SDS Gavintech'' version of this screen was disabled in
favor of the new version, so it was not functional for anyone.)
The second change...
For an idiomatic PHP Laravel project, each of its web controller classes
has an unqualified PHP class name in the form <code>FooController</code>.
While nearly all of the 161 ''SDS Laravel'' controller classes followed
that naming convention, these 3 were lone exceptions:
* Admin/OnlineForms/EditFormCampaigns.php
* Admin/OnlineForms/EditFormMeister.php
* Admin/OnlineForms/MergeFormCampaign.php
This task renamed each of those 3 to add <code>Controller</code> to their
unqualified names.
[[#top|RETURN]]
=== 2024 Jul 4: SDS Laravel: Functional Screens Not Distinguished in Main Menu Bar ===
The main navigation menus of ''SDS Laravel'' declare a hierarchical list of
named screens, but a design decision was made such that, while the app is
still incomplete and under development, the menus list many screens that do
not yet exist in ''SDS Laravel'' as placeholders along with many which do.
There was a serious usability problem with how this was first implemented,
such that each menu item did not distinguish in any way whether it was for
a screen that existed or whether it was a placeholder; both kinds looked
identical visually and both kinds were active hyperlinks that went
somewhere, and a user could only tell which kind it was by choosing it and
seeing where they went as a result. In the case of placeholders, every one
un-helpfully went to the home screen.
This was a problem because users exploring ''SDS Laravel'', whether to use
it normally or to test it, would waste a lot of time searching for needles
of functional screens in haystacks, with much trial and error, as about
half of the total menu item count was placeholders.
This task made 2 related changes for usability without losing anything.
The first change is that now for each menu item which is a placeholder, its
name is prefixed by "(N/A)" so it clearly stands out visually.
The second change is that now for each menu item which is a placeholder,
its hyperlink is no longer functional, and a user isn't taken anywhere when
clicking on it; they stay on their current screen.
Note that common established practice in desktop GUI apps is to use
alternate fonts or colors for visual distinction, such as non-active menu
items having grey text versus black text for active ones, but this task
eschewed that solution given its greater complexity. A future enhancement
can do this, however.
Note that an alternate solution could have been to just not display
placeholders anymore, so all menu items went to screens, but it was deemed
more useful to retain the placeholders.
Note that for the purposes of this menu logic, the determination of whether
a screen was implemented or a placeholder was just whether or not a Laravel
route was defined for it, in the source file <code>/routes/web.php</code>.
Implementing this change was a single line change in the source file
<code>/app/Models/Sub/Page.php</code>.
[[#top|RETURN]]
=== 2024 Jul 23: SDS Laravel: Form Campaign Add Target Students Uncheck All Button Fails ===
The ''SDS Laravel'' screen Admin Menu -> Online Forms -> Form Campaign
-> Consolidated Health Form -> Manage has broken behavior.
The "Add New Targets" screen section presents a list of students with a
checkbox for each student. When visiting the screen, the checkboxes all
default to a checked state. Often this is what the user wants but sometimes
they want to just pick a few students instead, and to assist with this is
an "Uncheck All" button.
Clicking the "Uncheck All" button as the user's first action on this screen
succeeds. But if any boxes are manually checked afterwards, clicking
"Uncheck All" again does nothing, and those boxes are still checked.
Furthermore, if ''SDS Laravel'' is updated to use Vue 3 from its current
Vue 2, the same functionality manifests additional broken behavior such
that the "Uncheck All" button has no effect even on initially visiting the
page, so it doesn't even work a single time to uncheck all the boxes.
This task fixes the broken behaviour by just making a few small updates in
the 1 PHP source file
<code>resources/views/admin/online_forms/form_campaigns/show.blade.php</code>.
Primarily there was this 1 substitution in the "Uncheck All" button definition:
    onclick="$('#student_list input:checkbox').attr('checked', false);"
    onclick="Array.from(document.getElementsByClassName('checkbox_student_list_add_target')).forEach(function(checkbox) { checkbox.checked = false; });"
In support of that, there was also this 1 addition to each student checkbox:
    class="checkbox_student_list_add_target"
Those 2 line changes are all that was needed to fix the broken behavior.
The underlying problem appears to be that jQuery's mechanism to enumerate
the set of checkbox DOM nodes fails to find them when any environmental
change occurs such as when users click the checkboxes, so the JavaScript
code to uncheck those boxes acts like the checkboxes don't exist.
Note that built-ins like <code>querySelectorAll()</code> have the same problem.
Whereas, <code>getElementsByClassName()</code> is resilient to such
environmental changes and can still find the checkboxes after user actions.
This task also makes a user interface change to enhance users' experience,
which is to add a complementing "Check All" button.
[[#top|RETURN]]
=== 2024 Aug 15: SDS Gavintech: Input Absences Screen Campus Filter Not Sticky and Other Bugs ===
The ''SDS Gavintech'' screen Staff Menu -> Attendance -> Input Absences has
several bugs or otherwise sub-optimal behaviors, and this task makes
7+ distinct changes to respectively fix or improve those.
All of the changes are isolated to the single PHP source file
<code>pages/attendance/input_absence.php</code> and so should not impact
any part of SDS besides this one screen.
The first/primary change...
The filter toggle "Show all campuses"/"Restrict to campus only" had a bug
such that the setting of this toggle was not sticky; as soon as the user
took any action which reloaded the screen, such as adding/editing/deleting
an input absence record, or such as flipping the filter toggle "Day
students only"/"All students", the campus toggle would reset to its default
of showing the current campus only. Besides this going against user
expectations and making more work for them, it is also inconsistent with
the behavior of the day students toggle which generally did persist through
any corresponding user actions.
This task normalized the behavior of the screen so that both of these
filter toggles behave the same and persist through any actions the user
takes on the screen; they would only reset to defaults if the user leaves
the screen entirely and returns.
In particular, this task's change makes things better because if one uses
the screen to add an absence for a student in another campus, then they can
actually see the record they just added on the screen after they submitted
the form, while otherwise it would look more like their action had no
effect, besides the transient status message of "Success: Added entries".
The second change...
This task also made a behavior change such that the current setting of the
campus filter toggle is now reflected in the title of the screen with a
"(campus only)" or "(all campuses)" suffix, so its more clear to users what
the setting is. This also makes the screen more consistent with the other
filters, such that the day students filter had a corresponding title suffix
already, "(day students)" and "(all students)" respectively, and the date
filter which shows the date its filtered to.
The third change...
This task also made a behavior change such that the value of the "Set date"
form/button now defaults to the same date the screen is currently
displaying, rather than the real-world current date, which is effectively
like the screen not persisting their choice to an extent.
I feel that behavior is more useful and in line with user expectations or
less surprising to them. It also means that this form is more thoroughly a
localized version of the application global "Date" setting. That is,
changing the date globally also affects this screen, but changing it on
this screen only affects this screen, until the user leaves this screen and
it resets to match the global setting.
The fourth change...
This task also made a behavior change such that the link to "Print list of
absent day students" has been renamed to "Print list of absent students",
and it now displays unconditionally instead of only when the screen is
filtered to just day students.
This new behavior was deemed more useful such as if we have an emergency on
campus and we have to evacuate, a full list would indicate day students
with absences entered by Ally/Bonita and boarding student absences entered
by the Health Centre.
However, there was no behavior change with respect to what actually
happened when that link was clicked on; because of a prior bug, the link
always led to a list of all students anyway, rather than filtering to show
just day students.
The fifth change...
This task also fixed a bug in the "Existing absences" -> "Entered by"
column such that the HTML for the link to email the person was malformed
and the url did not go anywhere / got a 404, but with the fix it correctly
goes to the email screen.
The sixth change...
This task also removed the link "Email Advisors & Students", which was
never implemented and did nothing but reload the page when clicked on.
The seventh+ changes...
This task also further normalizes how urls are generated that point back to
the current screen, reducing duplication of repeated details, and also
normalizing that arguments for filters only appear in urls if they did
before or are not defaults.
This task also normalizes some other implementation code, such as
specifying that all the "Dropdown" calls return their results rather than
printing them.
[[#top|RETURN]]
== Fixes to Slow Performance ==
=== 2024 May 14: SDS Laravel: Main Menu Bar Heavy Database Use Very Slow Page Loads ===
''SDS Laravel'' currently relies on the "pages" database table to
canonically define its app main navigation menus and it currently has 201
records where each defines a menu item.
The main menus are hierarchical, and the "order" field of each "pages"
record serves double duty to define the parent/child aka super/sub
relationships of menu items as well as their display order. The "order" of
a record is used as a unique key for the record for this purpose, as well
as a foreign key from child to parent.
Each "order" is a positive integer N such that:
* The result of log-100(N) is the item's tree level.
* The result of div-by-100(N) is the item's parent item's "order" in the tree.
* N itself is the item's display order relative to its siblings.
For example, a root level menu item has an "order" in 1..99, and for the
root item with an "order" of 3, all of its children would have "order" in
301..399, and the child of 303 would have "order" in 30301..30399 etc.
(Very notably, the actual primary key field of "pages" is NOT used for
describing menu parent/child relationships in conjunction with a second
field, as is the more typical practice in other databases.)
''SDS Laravel'' reads the entire "pages" database table and constructs a
hierarchical menu in 2 different circumstances. The primary and most
important circumstance is for nearly every single screen, those where the
user is logged in, and it is a normal screen, and not an API, a menu for
navigation appears at the top of each screen. The secondary circumstance is
on the "System Settings"->"Pages" screen that is used to edit this database
table, where the hierarchy is rendered as a left column.
Prior to the changes of this task, the hierarchical menu assembly was
performed in a very inefficient manner such that the application made a
separate database query for every single tree node to fetch its children,
meaning that it was making 138 queries for nearly every screen displayed to
the user just for making the navigation menu alone. And on the "Pages"
screen it was making 427 additional queries (total 565) just to display the
menu hierarchy on that one screen. For context, the sum total of all other
database queries for all other purposes on a single screen was 20-30 or so.
Compounding the above query-per-node inefficiency was another kind of
inefficiency where, for each menu node, the same query for its children was
often done up to 3 times (or up to 5 times), where only the last query
actually enumerated the list of child nodes, and prior times simply tested
if there were any children, and if there were children, the query was run
again to actually use them.
Following the changes of this task, the hierarchical menu assembly
only involved 2 SQL queries as overhead for nearly every screen, and 2 more
on the "Pages" screen, because it now selects all the "pages" records at
once rather than a handful or zero at a time, and uses the same fetched
records when building every menu node. The records are grouped when fetched
in a PHP array of arrays, the root array keyed by parent "order" values,
and the array value for each being an array of its child "pages" records.
This change had a massive impact on the user-visible screen loading speed
of ''SDS Laravel'', particularly when a developer runs the application and
its database on opposite ends of a VPN tunnel where the added latency of
each database call magnified the overhead time of each database query.
For example, before the changes, running the app on a local machine at
home, where it talked to a database on a school server, took about 23..25
seconds (155 queries) on the fastest screen with a menu to display on that
screen, and after the changes it took about 3.4-3.7 seconds (19 queries).
And on the "Pages" screen it took about 1.5 minutes (581 queries) to
display the screen before the changes, and about 4.1-4.2 seconds (20
queries) after the changes.
For a typical non-developer user scenario which didn't have that kind of
added VPN latency, the time to load a faster screen before the changes was
around 0.46-0.49 seconds, and after the changes it was about 0.12-0.13
seconds. For the "Pages" screen it was about 1.8-1.9 seconds before and
about 0.1-0.3 seconds after.
This task kept its changes about as minimal as possible while making this
change. There was no significant change to how screens or menus were
defined, and anything else related to the application design, aside from
the minimal needed to consolidate the database queries. While there is a
lot of room to make other improvements to how menus are implemented, other
such changes are out of scope for this task.
Here are specific numbers for before/after, school/home-VPN, on 4 screens.
SDS Laravel/API Screen: /
* Blade Views Rendered: 1
* Database Queries Run Before: 0
* Database Queries Run After: 0
* Eloquent Models Used Before: 0
* Eloquent Models Used After: 0
* HTTP Page Load Time at School Before: about same
* HTTP Page Load Time at School After: 58.8ms,59.9ms,55.3ms,59.3ms,63.1ms
* HTTP Page Load Time at Home/VPN Before: about same
* HTTP Page Load Time at Home/VPN After: about same
SDS Laravel/API Screen: /home
* Blade Views Rendered: 3
* Database Queries Run Before: 192
* Database Queries Run After: 56 (136 fewer)
* Eloquent Models Used Before: 455
* Eloquent Models Used After: 475
* HTTP Page Load Time at School Before: 672ms,671ms,734ms,725ms,617ms
* HTTP Page Load Time at School After: 362ms,416ms,364ms,271ms,256ms
* HTTP Page Load Time at Home/VPN Before: 30.22s,30.78s,29.54s,29.90,29.83
* HTTP Page Load Time at Home/VPN After: 9.30s,9.07s,9.21s,9.24s,9.35s
SDS Laravel/API Screen: /Event/Calendar/33297
* Blade Views Rendered: 4
* Database Queries Run Before: 155
* Database Queries Run After: 19 (136 fewer)
* Eloquent Models Used Before: 200
* Eloquent Models Used After: 220
* HTTP Page Load Time at School Before: 462ms,487ms,589ms,493ms,491ms
* HTTP Page Load Time at School After: 126ms,131ms,125ms,121ms,125ms
* HTTP Page Load Time at Home/VPN Before: 24.06s,25.20s,24.06s,24.67s,23.72s
* HTTP Page Load Time at Home/VPN After: 3.38s,3.41s,3.39s,3.72s,4.73s
SDS Laravel/API Screen: /System/Page
* Blade Views Rendered: 34
* Database Queries Run Before: 581
* Database Queries Run After: 20 (561=136+425 fewer)
* Eloquent Models Used Before: 779
* Eloquent Models Used After: 436
* HTTP Page Load Time at School Before: 1.86s,1.85s,1.79s,1.88s,1.80s
* HTTP Page Load Time at School After: 170ms,207ms,173ms,316ms,208ms
* HTTP Page Load Time at Home/VPN Before: 1.5min,1.5min,1.5min
* HTTP Page Load Time at Home/VPN After: 4.16s,4.17s,4.09s,4.12s,4.12s
[[#top|RETURN]]
=== 2024 Aug 20: SDS Gavintech: Input Absences Screen Heavy Database Use Very Slow ===
The ''SDS Gavintech'' app has widespread structural inefficiencies in the
way that it uses the database, and this manifests as the primary cause of
very slow app performance for users. The app may make 1000s or 10000s of
distinct database calls every time a user visits any screen or performs any
action on the screen that involves reloading the page, when it should only
be making on the order of 1-20 database calls total each such time. As a
result, each user action that loads/reloads a screen may carry a delay of
1-2 seconds at the bare minimum and often 10-20 seconds or several minutes,
which cumulatively is a very poor user experience.
To be specific, that was the experience of the production SDS instance in
early June of 2024. A pending migration as of this writing to newer
production hardware would yield performance improvements of 2-5X alone,
which helps significantly, but is still very noticeably slow. In contrast,
changes to the app source code to correct the structural inefficiencies
would instead yield further performance improvements in the 15X+ range, such
that user actions then actually may seem instant. Meanwhile, any specific
metrics given below for this task are all using the same newer hardware, to
be apples to apples comparisons, and also each version is refreshed several
times and the shortest generation time is used.
This task makes several distinct changes specific/isolated to the 1 screen
Staff Menu -> Attendance -> Input Absences such that the logic specific to
that screen is optimized in how it uses the database.
As a result of this task's changes, in order to display the screen, a total
of exactly 8-9 database calls are made the screen-specific elements (4
trivial ones plus 4-5 main ones), rather than about 9594 calls.
Further, as a result of this task's changes, in order to bulk-add new
absence records, a total in typical cases of exactly 4 database calls are
made for the database updating logic (2 trivial ones plus 2 main ones), or
a typical extreme maximum of about 24 database calls if adding around 1100
students, rather than a minimum of 6 calls and a maximum of about 4000-5000
respectively.
As a result of this task's changes, all user actions on the Input Absences
screen that involve a screen load/reload now take about 0.1 seconds, which
is essentially immediately, including displaying and bulk adding, as far as
the work specific to the screen itself is concerned. Unlike the old logic,
the new logic should scale and have essentially this same speed no matter
how many students there are, rather than slowing down with more students.
This should be very noticeable to users, as every single individual action
should feel a lot faster, which is even more significant given a typical
user workflow of performing dozens to a hundred actions here each day.
As a result of this task's changes, the amount of server memory used also
goes down by orders of magnitude, from about 168MB to about 4MB usually
(2MB in some cases), a 42X savings, as far as the work specific to the
screen itself is concerned. If widely applied, such changes can also mean
SDS can scale to handle orders of magnitude more concurrent users on the
same hardware without users seeing further slowdowns due to competition for
resources with other concurrent users.
All of the changes are isolated to the single PHP source file
<code>pages/attendance/input_absence.php</code> and so should not impact
any part of SDS besides this one screen.
Prior to the changes of this task...
This is the overhead for every screen minus that specific to the main menus;
it comprises 16 generally distinct simple database calls:
    Page generation time: 0.011
    Total SQL Queries: 16
    Total SQL Cache Hits: 15
    Total Local Cache Hits: 2
    Total Memcache Hits:
    Memory Usage: 2048K
    Realpath Usage: 87K
This is the overhead for every screen inclusive of the main menus,
when using SDS with the effective date of 2024 May 6 and
as a test/staff user "Darren Duncan" that is not a teacher/parent/student;
the main menus alone add 1567 database calls:
    Page generation time: 0.231
    Total SQL Queries: 1583
    Total SQL Cache Hits: 7078
    Total Local Cache Hits: 10
    Total Memcache Hits:
    Memory Usage: 32768K
    Realpath Usage: 125K
This is the overhead for every screen inclusive of the main menus,
when using SDS with the effective date of 2024 May 6 and
as a teacher user "Carol Adamson",
who also has a "My Courses" main menu but lacks some other staff menu items;
the main menus alone add 2049 database calls:
    Page generation time: 0.27
    Total SQL Queries: 2065
    Total SQL Cache Hits: 9188
    Total Local Cache Hits: 65
    Total Memcache Hits:
    Memory Usage: 57344K
    Realpath Usage: 87K
This is Input Absences screen on effective date 2024 May 6 by user Darren
Duncan including all the above overhead (user Carol Adamson doesn't have
privileges to see this screen);
the initial filter settings of current/Senior campus only plus all
day plus boarding students (about 600 students)
there are 10131 database calls total:
    Page generation time: 1.471
    Total SQL Queries: 10131
    Total SQL Cache Hits: 31468
    Total Local Cache Hits: 5351
    Total Memcache Hits:
    Memory Usage: 180224K
    Realpath Usage: 125K
This is the same Input Absences screen initial load but with the main menu
generation disabled, so it is more clear what this screen alone is doing;
there are 9594 database calls that are specific to this screen;
many of those calls are the same as ones in the main menu generator, and
those same calls are cached / only made once when both the main menu
generator and this screen's logic are enabled:
    Page generation time: 1.406
    Total SQL Queries: 9610
    Total SQL Cache Hits: 23359
    Total Local Cache Hits: 5343
    Total Memcache Hits:
    Memory Usage: 167936K
    Realpath Usage: 126K
Note that all further numbers below likewise have the main menu generation
disabled, to just highlight what Input Absences alone is doing.
This is the same thing but with the campus filter set to all campuses
and showing both day and boarding students (about 1000 students);
there are 3514 database calls specific to this screen;
there are actually 2/3 fewer database calls despite showing more students:
    Page generation time: 0.499
    Total SQL Queries: 3530
    Total SQL Cache Hits: 17071
    Total Local Cache Hits: 289
    Total Memcache Hits:
    Memory Usage: 65536K
    Realpath Usage: 88K
This is the same thing but with the filter set to day students only,
and campus filter still main only;
there are 9541 database calls specific to this screen;
unlike the campus filter, the day student filter only has a small effect:
    Page generation time: 1.462
    Total SQL Queries: 9557
    Total SQL Cache Hits: 20482
    Total Local Cache Hits: 5280
    Total Memcache Hits:
    Memory Usage: 165888K
    Realpath Usage: 126K
This is the same thing but with both filters toggled such that it shows
day students only plus all campuses;
there are 2942 database calls specific to this screen:
    Page generation time: 0.469
    Total SQL Queries: 2958
    Total SQL Cache Hits: 17827
    Total Local Cache Hits: 217
    Total Memcache Hits:
    Memory Usage: 55296K
    Realpath Usage: 89K
About 80-90% of the database calls specific to this screen are about
fetching a list of currently enrolled students for the effective date, with
or without campus/day filters, and using that list both to populate a
picklist of students and also to filter a display of existing absense
records, because it is making a handful of separate database calls for each
of about 1000 students (both when all 1000 are used and when some aren't).
It also makes a handful of separate database calls for each existing
absense record displayed, but these tend to be on the order of dozens of
records instead of 1000.
When the user performs an action like adding/editing/deleting an absence
record on this screen, the actual database modification task is typically
very fast, assuming just a single student updated, usually around 2-6
database calls for that, but then the page reloads carrying the large
aforementioned overheads. As an exception to this, when the user performs
a bulk addition of absence records, by selecting multiple students, each
individual student involves 4-5 database calls, of which 2 are reads and
2-3 are updates, and so in the unusual case of adding all thousand at once,
it can make around 4000-5000 database calls, and take around 15 seconds on
the newer hardware to do so; and then the page reloads.
Following the changes of this task...
This is the same thing, on just displaying the Input Absences screen,
for effective date 2024 May 6, for user Darren Duncan;
there are 8-9 database calls specific to this screen;
it is 9 when filtering to day students only, and 8 otherwise;
it is unchanged by filtering on one or all campuses:
    Page generation time: 0.093
    Total SQL Queries: 24
    Total SQL Cache Hits: 25
    Total Local Cache Hits: 5
    Total Memcache Hits:
    Memory Usage: 4096K
    Realpath Usage: 54K
    Page generation time: 0.108
    Total SQL Queries: 25
    Total SQL Cache Hits: 26
    Total Local Cache Hits: 5
    Total Memcache Hits:ƒ
    Memory Usage: 2048K
    Realpath Usage: 92K
In addition, the logic for batch-adding adds almost no time to that in
typical cases of 1 or few students, and otherwise a small fraction of 1
second for the extreme case of bulk-adding all thousand students at once.
The main general change made by this task is to clone several call chains
worth of source code for several dozen routines across a handful of classes
into <code>input_absence.php</code> and iteratively refactoring that whole
code mass to its most concise specific-to-this-screen's-use-case version.
No code shared with other screens was altered, allowing a guarantee that
no other screens are affected by these changes, making it much lower risk,
and a much simpler task to accomplish.
The most important effect of the main change is the consolidation of
database calls. Typically how the logic worked before this task's changes
was that in order to produce a list of something, first you had a query
that just produced a set of record ids for the list we wanted, and then a
separate query was made for each of those ids to get other fields we want
for each record, sometimes in one call and sometimes in a handful, some for
display purposes, and others for filtering purposes. Hence you may have
around 3000-8000 calls to deal with a list of around 1000 students. This
task changed things so that exactly 1 SQL query was done instead in place
of that first id-fetching query that also gathered all the related details.
The 4-5 display-only main queries then were for these purposes:
* One to derive the input time range for the effective date, eg 8:30am-3:20pm, from a period list.
* One to display the picklist of students to add/edit absences for.
* One to display the picklist of attendence codes.
* One to display the list of existing absence records.
* Iff filtering students by day, one to fetch a list of fee statuses to filter with.
(While in theory the last one could also have been inlined, I chose not to
because it was relating fee statuses with "like" rather than equality.)
There was also these 3 add-only SQL statements per each batch of 100 students:
* One to test that the user-supplied student ids match student records.
* One to insert up to 100 absence records.
* An indirectly called one to insert a log entry for the prior insert.
In the interest of security and scalability I compromised my adding logic
between the pre-changes version of a separate database call per single
student and using a single database call for the entire list of 1000+
students, into batches of 100, which is essentially the same performance as
all-in-one but shouldn't be as likely to fail in pathological cases.
A related general change is that the main screen logic no longer turns
database records into cached PHP objects that it then calls methods on, and
instead just works with the records as regular PHP arrays which typically
don't correspond to single database tables but joins of several. The logic
goes from interacting heavily with the app's record object cache to
generally not interacting with it at all, either to add or remove items.
In particular, the logic no longer has up to a handful of repeatedly
converting back and forth between record ids and record objects per record.
Between this, and the greatly reduced use of the app's SQL query cache due
to fewer database calls, likely is the main reason why so much less memory
is used by PHP in the new version.
Note that while some database queries or other logic could possibly be
streamlined or simplified further, I tried to maintain the exact effective
logic as possible from before, in case there might be subtle behavior
changes from such further streamlining. But an actual future rewrite of
this screen such as under ''SDS Laravel'' could likely revisit this.
Generally speaking, the screen logic I overhauled was that where many
similar database queries could be consolidated, while logic that only had
individual distinct queries, such as related to getting the current school
day, I left alone.
[[#top|RETURN]]
=== 2024 Aug 20: SDS Gavintech: Attendance Code / Reason Picklists Slow ===
This task changes the ''SDS Gavintech'' app shared component that is used
by all or most screens with an Attendance Code / Reason picklist so that it
internally uses the database more efficiently and thus performs better.
For a typical scenario of displaying 20 attendance codes, it now performs
this using exactly 1 database call rather than 21 database calls.
This benefit is seen on every screen using this picklist, not just 1.
While the impact of this change alone on a single user action is likely too
small for a user to notice, it still contributes to efficiency, and in
aggregate, performance improvements users would notice.
That change was entirely in the single PHP source file
<code>overrides/classes/class_dropdowns.php</code>, in its single routine
<code>attendanceCode()</code>.
This task also changes the internals of the 1 screen Staff Menu ->
Attendance -> Input Absences such that it now uses the same, now improved,
app component for its Reason picklist, rather than duplicating it.
That change was entirely in the single PHP source file
<code>pages/attendance/input_absence.php</code>.
[[#top|RETURN]]
== Internal Design Changes ==
=== 2024 Apr 19: SDS Laravel: Consolidate Eloquent model classes to /app/Models ===
An idiomatic PHP Laravel project has its main files grouped into the
specific hierarchy of folders documented here:
https://laravel.com/docs/11.x/structure
Of particular relevance, the folder <code>/app/Models</code> is where all
of the Laravel Eloquent model classes go.
Prior to the performance of this task, ''SDS Laravel'' conformed to the
idiomatic folder layout in all ways except with respect to the Eloquent
model classes, which it had instead spread across about a dozen
<code>/app/FooModel</code> folders.
This task moved/renamed all of the Eloquent model classes, such that for
each class <code>/app/FooModel/Bar.php</code>, it was moved/renamed to
<code>/app/Models/Foo/Bar.php</code>. While that described nearly all of
those classes, a handful started with other names, but they were
appropriately gathered under <code>/app/Models</code> too.
This task also included a normalization of how first-party classes were
referred to in other code, such that any which weren't referred to by their
fully-qualified names before were referred to in that way after.
[[#top|RETURN]]
=== 2024 May 8: SDS Laravel: Sever Coupling of PHP Class Names to Database Data ===
Laravel Eloquent has a particular more-advanced feature for defining
relationships between model classes that it refers to as a
''polymorphic relationship'', such that a relationship is not just between
2 specific model classes, but rather between 1 model class and 1 set of all
model classes that fulfill a particular role.
https://laravel.com/docs/8.x/eloquent-relationships#polymorphic-relationships
Prior to the performance of this task, ''SDS Laravel'' utilized such an
Eloquent relationship, between <code>Group</code> on one side, and the set
comprising <code>Page</code> and <code>SubPage</code> on the other side.
And it did so in a design-flawed way that led to tight coupling between the
exact fully-qualified PHP class names of the latter 2 model classes, and
the database data, such that those fully-qualified class names
(<code>App\SubModel\Page</code> and <code>App\SubModel\SubPage</code>) were
stored in the database (specifically in the <code>group_pages</code>
table's <code>class_type</code> column) and the fact that these matched was
relied on by the application for correct behavior.
This tight coupling manifested as broken app behavior when attempting to
move/rename the <code>Page</code> and <code>SubPage</code> classes (to
<code>App\Models\Sub\Page</code> and <code>App\Models\Sub\SubPage</code>),
because then their names no longer matched what was in the database.
The most common kind of breakage was that, for almost every app screen that
was gated with a user group privilege check for the screen, the app thought
no user had any privilege to use it; for example, this error message
displayed when trying to app general search screen:
    No access to Search.index
    You do not appear to have permission to access https://api-latest.smus.ca/Search as [User's Name]. If you believe you should have access
    to open a help ticket request permission, or you can go back / home and try again.
And had users been able to get past that, certain other screens or
functionality that rely on displaying or editing user group privileges for
pages and subpages would have had problems, such as failing to see existing
privileges made before the rename (they might have worked to re-add the
privileges, then using the new hard-coded class names in the database).
Without correcting the design flaw, renaming these model classes would have
required a corresponding database migration to replace the old
hard-coded-in-data PHP class names with their new ones, which carried more
complexity and risk.
The more specific design flaw was the use of Eloquent's
<code>morphToMany()</code> relationship builder without specifying an
explicit constant string to represent each morphed class in the database,
resulting in Eloquent defaulting to using its fully-qualified PHP class
name instead.
Both <code>Page</code> and <code>SubPage</code> had a copy of this method
(this version is post-rename) which was affected:
    public function groups(){
        return $this->morphToMany(\App\Models\Sub\Group::class, 'class', 'group_pages');
    }
This task made the simplest possible fix to this design flaw by explicitly
configuring Eloquent to use an explicit constant string per each morphed
class, by adding the following in the file
<code>app/Providers/AppServiceProvider.php</code>:
    Relation::enforceMorphMap([
        'App\SubModel\Page' => 'App\Models\Sub\Page',
        'App\SubModel\SubPage' => 'App\Models\Sub\SubPage',
        // TODO: Migrate database to use non-PHP-class-looking slugs
        // like following instead so they don't look tightly coupled.
        // 'page' => 'App\Models\Sub\Page',
        // 'sub_page' => 'App\Models\Sub\SubPage',
    ]);
This was the simplest possible fix, and ostensibly the most elegant, by
working with the system, using an Eloquent feature expressly designed for
this very scenario, rather than say trying to avoid using the
<code>morph</code> features (although doing so may be better longer term),
and also by not requiring any database changes, so it works with the
database as-is.
https://laravel.com/docs/8.x/eloquent-relationships#custom-polymorphic-types
Following the changes of this task, there was still room for an even more
elegant solution to follow that alters the database to use a
non-php-class-looking constant in the database data to represent each
class, so the app isn't forever using the old class name there, but that
was considered higher complexity and risk and suited to be done as a
separate task another day.
Note that a consequence of using <code>Relation::enforceMorphMap()</code>
is that it makes Eloquent treat the entire app more strictly, such that if
we try to use <code>morphToMany()</code> or similar Eloquent methods with
any other classes besides <code>Page</code> and <code>SubPage</code>, it
will require us to add an entry to the above list for it, rather than
silently defaulting to the old behavior for other unspecified classes. I
consider this strictness to be a good thing.
[[#top|RETURN]]
=== 2024 Aug 21: SDS Gavintech: Sever Coupling of Main Menu Builder to Write/Edit Reports Screen ===
This task refactored ''SDS Gavintech'' to sever the coupling between the
main menu builder, defined in the single PHP source file
<code>overrides/modules/sidebar_definition/smus_sidebar_definition.php</code>,
and the screen known alternately as
My Courses -> $course_name -> Reports & Marks -> Write reports -> $student_name
and
Staff Menu -> Advisor stuff -> Edit Reports -> $student_name -> $course_name,
defined in the single PHP source file
<code>pages/reports/reports_entry_spell.php</code>.
That coupling took the form of the main menu builder fetching a list of
students from the database on behalf of the screen and storing that list of
students in the PHP session variable
<code>$_SESSION["reports_entry_spell"]</code>, which was subsequently read
by the screen. The use of this session variable was one-way; the menu
builder only set it and the screen only read it. The menu builder set it
unconditionally from scratch, and so could have just used a PHP "global"
instead, which would have saved the app from the overhead of serializing
and deserializing its value to disk, which may have consisted of a thousand
"Student" PHP objects. The menu builder set it anew every time the user
loaded/reloaded any app screen, and not just when viewing the specific
screen it was used on, meaning almost all the time it consumed resources to
do that work for nothing.
The changes of this task consisted mainly of placing a stripped down clone
of the main menu builder source file into the top of the screen source
file, just the parts that built the student list. The changes otherwise
consisted of entirely eliminating the use of the PHP session variable.
The menu builder source file had no changes except for removing the 2 lines
that set the session variable; the screen source file used a local variable
to hold the student list instead of reading it from the session variable;
the stripped down clone it gained totalled 70 lines of code, which was
8% of the menu builder source file's lines.
Following the changes of this task, the actual performance of the app was
identical, using exactly the same count of database calls per each screen
as prior to the changes of this task, and also the same page generation
time. This was tested both on the app home screen, which used the menu
builder logic but not the screen logic, and on the screen, which used both.
Despite the duplicate database call main logic, this would be explained by
these database calls being mediated by both an app SQL query cache and an
app PHP object cache, which recognized that the results of the first
instance of the calls by the menu builder could be reused by the screen.
This decoupling resulted in the app being easier to maintain, as one can
now change either the menu builder or the screen arbitrarily without being
concerned that changes to one would affect the other, as conceptually the
two things should be completely separate.
[[#top|RETURN]]
== Changes to Third-Party Dependencies ==
=== 2024 May 3: SDS Laravel: Stop Requiring dbal, revisionable-upgrade, laravel-cors ===
Prior to the performance of this task, the ''SDS Laravel'' Composer config
file <code>composer.json</code> explicitly required these 3 third-party
PHP library dependencies, which were not actually used by the app:
* doctrine/dbal (^3.5)
* fico7489/laravel-revisionable-upgrade (*)
* fruitcake/laravel-cors (^2.0)
This task updated <code>composer.json</code> to remove those unused
explicit dependencies, which also had the effect of removing 6 other
Composer-resolved dependencies.
[[#top|RETURN]]
=== 2024 May 6: SDS Laravel: Reflect rename of fzaninotto to fakerphp ===
This task updated ''SDS Laravel'' <code>composer.json</code> to reflect that
the third-party PHP library dependency <code>fzaninotto/faker</code> was
renamed to <code>fakerphp/faker</code> (in fact, that rename occurred prior
to the release of Laravel 8), and also require its latest version.
[[#top|RETURN]]
=== 2024 May 6: SDS Laravel: Upgrade PHPUnit from 9.x to 10.x ===
This task updated ''SDS Laravel'' <code>composer.json</code> to require the
latest PHP-8.1-compatible version of the PHP dev dependency PHPUnit from
<code>9.6.19</code> to <code>10.5.20</code>.
[[#top|RETURN]]
=== 2024 Jul 15: SDS Laravel: Upgrade Laravel Mix from 5.x to 6.x and Upgrade Webpack ===
This task updated ''SDS Laravel'' so that its JavaScript build pipeline is
implemented by Laravel Mix 6 rather than 5, and likewise a newer major
version of Webpack.
This task changed these dependencies in <code>package.json</code>:
* cross-env (^7.0.3 removed as laravel-mix has its upgrade built-in)
* postcss (^8.4.39 added)
* laravel-mix (^5.0.9 to ^6.0.49)
* vue-loader (^15.4.2 to ^15.11.1)
This newer pipeline should be usable in the same way as the prior one, with
<code>npm run prod</code> and <code>npm run dev</code> etc, but Laravel Mix
6 makes it considerably more streamlined and ostensibly considerably less
fragile than the Laravel Mix 5 version was. Much less underlying Webpack
API details are exposed in the "scripts" commands, and also cross-env is no
longer directly used and is dropped as a dependency.
The following substitution was performed in <code>package.json</code>:
    "scripts": {
        "dev": "npm run development",
        "development": "cross-env NODE_ENV=development node_modules/webpack/bin/webpack.js --progress --hide-modules --config=node_modules/laravel-mix/setup/webpack.config.js",
        "watch": "npm run development -- --watch",
        "watch-poll": "npm run watch -- --watch-poll",
        "hot": "cross-env NODE_ENV=development node_modules/webpack-dev-server/bin/webpack-dev-server.js --inline --hot --config=node_modules/laravel-mix/setup/webpack.config.js",
        "prod": "npm run production",
        "production": "cross-env NODE_ENV=production node_modules/webpack/bin/webpack.js --no-progress --hide-modules --config=node_modules/laravel-mix/setup/webpack.config.js"
    },
    "scripts": {
        "dev": "npm run development",
        "development": "mix",
        "watch": "mix watch",
        "watch-poll": "mix watch -- --watch-options-poll=1000",
        "hot": "mix watch --hot",
        "prod": "npm run production",
        "production": "mix --production"
    },
Two other files each had a single line change or addition for
compatibility with the Laravel Mix version change.
The following substitution was performed in <code>webpack.mix.js</code>:
    mix.js('resources/js/app.js', 'public/js')
        .sass('resources/sass/app.scss', 'public/css');
    mix.js('resources/js/app.js', 'public/js')
        .vue()
        .sass('resources/sass/app.scss', 'public/css');
That is Laravel Mix 6 no longer assumes one is always using Vue, so now one
opts into processing its templates explicitly with the <code>.vue()</code>.
The following substitution was performed in <code>resources/js/app.js</code>:
    window.Vue = require('vue');
    window.Vue = require('vue').default;
Note that Laravel Mix 6 is required for Vue 3 compatibility but it also
works with Vue 2; the latter is still in use as of this task's changes.
[[#top|RETURN]]
=== 2024 Jul 15: SDS Laravel: Replace laravel/helpers With laravel/framework ===
This task updated ''SDS Laravel'' <code>composer.json</code> to purge the
PHP library dependency <code>laravel/helpers</code>, and also updated any
''SDS Laravel'' PHP source files that used it to instead use
<code>laravel/framework</code> built-ins.
These 4 PHP source files were updated:
* app/Http/Controllers/Auth/RegisterController.php
* resources/views/admin/subjects/edit_rotating_courses/show.blade.php
* resources/views/student/student_mark_collections/edit.blade.php
* resources/views/student/student_mark_collections/show.blade.php
For each of the above 4 files, there were either of these 3 subsitutions:
    str_contains(...)
    str_random(...)
    str_replace_array(...)
    Str::contains(...)
    Str::random(...)
    Str::replaceArray(...)
Note that <code>laravel/helpers</code> was just a single PHP source file
providing trivially thin wrapper functions over
<code>laravel/framework</code> 6+ built-ins that emulated ones from Laravel
5.8, and we were already mostly using the newer versions, so this task just
replaced the last 7 instances where we weren't.
[[#top|RETURN]]
=== 2024 Jul 17: SDS Laravel: Upgrade Font Awesome from 4.x plus 5.x plus CDN to 6.x ===
This task updated ''SDS Laravel'' to require the latest version of the
JavaScript library dependency Font Awesome, to 6.x from a mixture of 3
different versions 4.x plus 5.x plus CDN.
This task changed these dependencies in <code>package.json</code>:
* @fortawesome/fontawesome-free (^5.15.4 to ^6.6)
* font-awesome (^4.7 removed as @fortawesome/fontawesome-free is its upgrade)
This task deleted references to this CDN dependency in 3 app Blade template
PHP source files:
* <nowiki>https://use.fontawesome.com/releases/v5.7.1</nowiki>
This task also made a small tweaks to a few other files for compatibility,
replacing font/symbol/style name <code>fa</code> with <code>far</code>.
[[#top|RETURN]]
=== 2024 Jul 17: SDS Laravel: Stop Requiring tempusdominus, bootstrap-select, jquery-ui, lodash ===
Prior to the performance of this task, the ''SDS Laravel'' config
file <code>package.json</code> explicitly required these 5 third-party
JavaScript library dependencies, which were not actually used by the app:
* bootstrap-select (^1.13.18 removed as not in use)
* jquery-ui (^1.13.3 removed as not in use)
* lodash (^4.17.21 removed as only used indirectly by laravel-mix)
* tempusdominus-bootstrap-4 (^5.39.2 removed as not in use)
* tempusdominus-core (^5.19.3 removed as not in use)
This task updated <code>package.json</code> to remove those unused
explicit dependencies.
This task also updated 5 other app files for compatibility, typically by
deleting explicit "require" or similar statements to these libraries since
they weren't actually used otherwise.
However, these 2 files were special:
* resources/views/admin/subjects/edit_subject/show.blade.php
* resources/views/application/studentinfo/index.blade.php
Those 2 files actually DID use removed dependencies, and so technically
they were broken by this task. But the 2 screens in question were not
currently in use and so temporarily breaking them was acceptable. One
screen was used historically but not any more. The other screen was still
under construction. This task added comments to those 2 files explaining
that they will need to be modified to use alternate and modern dependencies
if they are going to be completed or used again.
[[#top|RETURN]]
=== 2024 Jul 18: SDS Laravel: Upgrade axios ===
This task updated <code>package.json</code> to upgrade the JavaScript library
dependency <code>axios</code> from <code>0.27.2</code> to <code>1.7.2</code>.
[[#top|RETURN]]
=== 2024 Jul 25: SDS Laravel: Upgrade Vue.js from 2.x to 3.x and Upgrade Co-Dependent JS deps ===
This task updated <code>package.json</code> to require the latest versions
of <code>vue</code> as well as of all of its co-dependent JavaScript
library dependencies.
To be specific, it made these dependency changes:
* @formkit/pro (^0.119.14 added)
* @formkit/themes (^1.6.5 added)
* @formkit/vue (^1.6.5 added)
* css-loader (^7.1.2 added)
* vue (^2.7.16 to ^3.4.34)
* vue-loader (^15.11.1 to ^17.4.2)
* vue-bootstrap-datetimepicker (^5.0.1 removed as @formkit/pro "datepicker" is used in its' place)
* vue-template-compiler (^2.7.16 removed as vue 3 has its upgrade built-in)
Several changes for compatibility were made to the app source files.
Partially rewrote source file <code>resources/js/app.js</code> to account
for Vue 3 not having a single global state like Vue 2 but multiple possible
object-scoped states. Many of the same changes were previously made in
''SDS Gavintech'' which served as a guide.
Rewrote source file <code>resources/js/components/inputs/DateTime.vue</code>
to use <code>@formkit/pro</code> rather than
<code>vue-bootstrap-datetimepicker</code>. The new version is a clone from
''SDS Gavintech'' which had the corresponding changes made earlier.
Updated source file <code>resources/js/components/admin/finance/Fees.vue</code>
in several ways (some previously done in ''SDS Gavintech'', some brand new).
As Vue 3 removed the filters feature, their use was replaced with ordinary
function calls; <code>x|format_year</code> was replaced with <code>format_year(x)</code>.
As Vue 3 reversed the precedence of <code>v-if</code> plus <code>v-for</code>
in the same HTML/template tag, added a child <code>template</code> tag of
the <code>tr</code> that used both, to force the expected precedence.
Updated source file <code>resources/js/components/application/AddressInfo.vue</code>
to fix a syntax/logic error that Vue 2 overlooked but Vue 3 doesn't.
The logic on the System Settings -> Pages screen to fold/unfold the page
hierarchy was disabled as it was incompatible with Vue 3 and not essential
so the tree is just permanently unfolded. This was done by small changes in
the source file <code>resources/views/system/page/index.blade.php</code>.
The System Settings -> Groups screen was altered to be read-only because
the logic for adding/editing/deleting database records was incompatible
with Vue 3, a fix was elusive, and this was a good enough 90% solution for
the short term given the infrequent use of the screen; a follow-up task can
be performed later to enable editing on this screen again when needed.
This was done by small changes in the source file
<code>resources/views/system/group/index.blade.php</code>.
Formatting configuration for date pickers was updated on a dozen or two
screens replacing any <code>LL</code> and <code>LLL</code> shorthands with
their longhands, namely <code>D MMM YYYY</code> and
<code>D MMM YYYY HH:mm</code> respectively, since the new date pickers
don't support the shorthands, only the longhands.
This task also resulted in unintended changes in the app's visual
appearance, such as fonts being a bit smaller or some elements being in
slightly different locations. The cause of this proved elusive, and since
it isn't actually a functional break, this didn't hold back the Vue 3
upgrade. A follow-up task can be performed later if changes to the visuals
are desired.
[[#top|RETURN]]
=== 2024 Aug 2: SDS Laravel: Upgrade directorytree/ldaprecord-laravel from 2.x to 3.x ===
This task updated ''SDS Laravel'' <code>composer.json</code> to require the
latest version the PHP library dependency
<code>directorytree/ldaprecord-laravel</code>, going from <code>2.7.3</code>
to <code>3.3.3</code>.
This task also made 3 distinct sets of app source code changes to be
compatible with that upgrade.
See also for context:
* https://ldaprecord.com/docs/laravel/v3/upgrading/
* https://ldaprecord.com/docs/laravel/v3/release-notes/
* https://github.com/DirectoryTree/LdapRecord-Laravel/releases
The first set of changes...
This task updated the 1 PHP config source file <code>config/ldap.php</code>
to account for <code>logging</code> now being an array. There was this 1
substitution:
    'logging' => env('LDAP_LOGGING', true),
    'logging' => [
        'enabled' => env('LDAP_LOGGING', true),
    ],
The second set of changes...
For context, <code>directorytree/ldaprecord-laravel</code> had a PHP trait
named <code>LdapRecord\Laravel\Auth\MultiDomainAuthentication</code> which
was deprecated before version <code>2.7.3</code> and removed in version
<code>3.0.0</code>. The 1 SDS PHP source file
<code>app/Http/Controllers/Auth/LoginController.php</code> used that trait.
This task updated <code>LoginController</code> to clone into itself the
used portions of <code>MultiDomainAuthentication</code>, thus removing the
external dependency of the former on the latter.
As the removed trait was itself a sub-trait of
<code>LdapRecord\Laravel\Auth\CreatesUserProvider</code>,
<code>LoginController</code> now composed that directly and not indirectly.
Also <code>LoginController</code> gained the new protected function
<code>getLdapGuard</code> cloned from the removed trait.
Note that the function <code>getLdapGuardFromRequest</code> was ''not''
cloned from the removed trait since <code>LoginController</code> already
had its own version that overrode it.
The third set of changes...
For context, each of these 5 SDS classes composed one of the 2 classes
<code>LdapRecord\Models\Model</code> or <code>LdapRecord\Models\Scope</code>:
* app/Ldap/ExternalUser.php
* app/Ldap/Scopes/OnlyStaffUsers.php
* app/Ldap/Scopes/OnlyStudents.php
* app/Ldap/SmusStudents.php
* app/Ldap/SmusUser.php
For each of those 5, it was updated such that for any property or method it
contained which overrode or implemented a same-named one from
<code>Model</code> or <code>Scope</code>, that property or method had
explicit type annotations added to match the originals. This was required
for the SDS code to satisfy a PHP or Laravel stricture so the code runs.
[[#top|RETURN]]
=== 2024 Aug 2: SDS Laravel: Upgrade intervention/image from 2.x to 3.x and Fix Import Photos Screen ===
This task made 3 distinct groups of changes to ''SDS Laravel'' all related
to image handling and the Admin Menu -> Migration -> Import Photos screen.
==== Changes For intervention/image Upgrade ====
This task also made 2 distinct sets of changes to be compatible with the
<code>intervention/image</code> upgrade.
For context, as part of the major update of <code>intervention/image</code>
from version 2.x to 3.x, it was also split up into multiple libraries, with
the core remaining under the old name and the optional Laravel-specific
add-ons being in the new <code>intervention/image-laravel</code> library.
We only need to directly require the latter, which brings in the former.
See also for context:
* https://image.intervention.io/v3/introduction/upgrade
* https://github.com/Intervention/image-laravel/blob/main/README.md
* https://image.intervention.io/v3/modifying/resizing
The first set of changes...
This task added the 1 PHP config source file <code>config/image.php</code>
that is standard per the above-linked README.
The second set of changes...
This task replaced the use of the no longer existing PHP class
<code>Intervention\Image\ImageServiceProvider</code> with the existing
<code>Intervention\Image\Laravel\Facades\Image</code>.
In addition, the class is now used directly by its full name in the 2 PHP
source files where the image handling functionality is actually used, which
are <code>app/Models/File.php</code> and
<code>app/Http/Controllers/Admin/Migration/ImportPhotosController.php</code>.
This task has purged any explicit registration in <code>config/app.php</code>
of any class of this library in <code>providers</code> or
<code>aliases</code>; the <code>Image</code> alias is now purged.
The PHP source code that directly used the Intervention Image libraries was
updated for compatibility with several method renames or substitutions.
Changes were made to reflect the rename of the library static method
<code>make</code> to <code>read</code>.
Changes were made to reflect the changes with the set of image manipulation
methods for image resizing. The <code>resize</code> method of version 2 had
taken a third argument by which one specified the resize should keep the
original aspect ratio. The <code>resize</code> method of version 3 doesn't
take that argument, and instead the version 3 method <code>scale</code> is
explicitly for resizing that maintains the aspect ratio.
[[#top|RETURN]]
==== Bug/Security Fixes In Photos Import ====
This task also updated the 2 files
<code>app/Http/Controllers/Admin/Migration/ImportPhotosController.php</code>
plus
<code>resources/views/admin/migration/import_photos/index.blade.php</code>
to fix bugs tangential to the image file path handling.
One fix was in the <code>index</code> function so that the directory path
reported to the user was the same one that <code>store</code> actually
used, from <code>Storage::disk('photos')</code>, rather than being either
of a couple of hard-coded alternatives toggled on an environment variable
that isn't actually used.
A second fix was to stop passing a <code>path</code> hidden form field
value from <code>index</code> to <code>store</code>, whose value was set
from the incorrect path in <code>index</code>, and which <code>store</code>
validated as present and subsequently didn't use. Besides being dead code
from the disuse, if this had been used it would have been a major security
vulnerability, as it would have let web clients directly control actions
against server local file system paths.
A third fix was in the <code>store</code> such that logic to skip over
files that were not of type <code>image/jpeg</code> was faulty and never
skipped any kind of file, and now it skips those it is supposed to.
The specific logic error was one of operator precedence, where it was of
the form <code>!x == y</code> where it meant to be <code>!(x == y)</code>;
the fixed version made it be <code>x !== y</code>.
[[#top|RETURN]]
==== Changes For league/flysystem Indirect Upgrade ====
This task also updated
<code>app/Http/Controllers/Admin/Migration/ImportPhotosController.php</code>
to be compatible with the <code>league/flysystem</code> upgrade from
version 1 to version 3 that was indirect by way of the Laravel 9 upgrade.
See also for context:
* https://flysystem.thephpleague.com/docs/upgrade-from-1.x/
* https://laravel.com/docs/9.x/upgrade#flysystem-3
There were 3 substitutions for compatibility in the <code>store</code> function.
There was 1 like this:
    Storage::disk('photos')->getMimeType($file)
    Storage::disk('photos')->mimeType($file)
And there were 2 like this:
    Storage::disk('photos')->getDriver()->getAdapter()->applyPathPrefix($file);
    Storage::disk('photos')->path($file);
This task also updated <code>config/filesystems.php</code> to add the
element <code>'throw' => true</code> to each <code>disks</code> element.
This is to help preserve Laravel 8 behavior that we might have been relying
on when a write operation fails. For example, write operations such as
<code>write</code> by default no longer throw an exception when a write
operation fails and instead returns <code>false</code>; until we can audit
our code for how we currently check for such failures, this config change
should make it less likely our code was broken by the dependency update.
Note that there are still other kinds of default behavior changes that we
should audit our code for handling of. For example, write operations will
overwrite existing files by default, so if we don't want that then we
should explicitly be checking for their existence first. Also, reading a
nonexisting file now returns null rather than an exception.
[[#top|RETURN]]
=== 2024 Aug 2: SDS Laravel: Upgrade laravel/framework from 8.x to 10.x and Upgrade Co-Dependent PHP deps ===
This task made 5 distinct groups of changes to ''SDS Laravel'' all related
to upgrading it from Laravel 8 to Laravel 10.
==== Summary of PHP Dependency Upgrades or Removals ====
This task updated <code>composer.json</code> to require the latest
PHP-8.1-compatible versions of <code>laravel/framework</code> as well as of
all of its co-dependent PHP library dependencies. In some cases, upgrading
co-dependents meant removing them entirely in favor of Laravel built-ins.
To be specific, it made these dependency changes:
* barryvdh/laravel-debugbar (^3.7 to ^3.13.5)
* directorytree/ldaprecord-laravel (^3.3.3 unchanged)
* etern8ty/beanstream (dev-master unchanged and is custom fork)
* fakerphp/faker (^1.23.1 unchanged)
* fideloper/proxy (^4.4.2 removed as laravel/framework has its upgrade built-in)
* goldspecdigital/laravel-eloquent-uuid (^8.0.1 removed as laravel/framework has its upgrade built-in)
* guzzlehttp/guzzle (^7.9.2 unchanged)
* intervention/image-laravel (^1.3 unchanged)
* juliomotol/laravel-auth-timeout (^3.1.1 to ^4.1)
* lab404/laravel-impersonate (^1.7.5 unchanged)
* laravel/framework (^8.83.27 to ^10.48.18)
* laravel/tinker (^2.9 unchanged)
* laravel/ui (^3.4.6 to 4.5.2; but it recommends using Laravel Breeze or Laravel Jetstream instead)
* mockery/mockery (^1.6.12 unchanged)
* nunomaduro/collision (^5.11 to ^7.10)
* phpunit/phpunit (^10.5.29 unchanged)
* spatie/laravel-ignition (^1.7 to ^2.8)
* staudenmeir/eloquent-has-many-deep (^1.14.4 to ^1.19.4)
[[#top|RETURN]]
==== Specify Minimum PHP Dependency Versions Used Are Stable ====
This task updated <code>composer.json</code> to specify an app default
<code>minimum-stability</code> of <code>stable</code> rather than
<code>dev</code>.
This is per the Laravel 10 upgrading recommendation:
* https://laravel.com/docs/10.x/upgrade#updating-minimum-stability
While the setting could instead have been removed as <code>stable</code> is
its default, being explicit seemed better here.
Note that this didn't cause any changes to what PHP library versions are
actually installed by Composer versus the old setting, but is a good
constraint going forward.
[[#top|RETURN]]
==== Changes For laravel/framework Upgrade's Removal of "dates" Model Attribute ====
This task also updated 51 PHP source files to be compatible with a breaking
change made by Laravel itself with version 10.
Laravel supported a "dates" model attribute through version 9, and then
Laravel 10 removed it. The function of this was to enumerate database/model
fields that were supposed to be automatically converted to Carbon DateTime
objects; so under Laravel 8, any "dates" declarations would be respected,
while under Laravel 10 they would be ignored.
See also:
* https://laravel.com/docs/10.x/upgrade#model-dates-property
Compare:
* https://laravel.com/api/9.x/Illuminate/Database/Eloquent/Concerns/HasAttributes.html
* https://laravel.com/api/10.x/Illuminate/Database/Eloquent/Concerns/HasAttributes.html
As a result, simply upgrading ''SDS Laravel'' from Laravel 8 to 10 resulted
in many parts of the app breaking in various ways including when simply
visiting the post-login home screen, as PHP died with errors like
<code>Call to a member function format() on int</code>.
To fix this, any instances of <code>protected $dates = ['x',...]</code>
in model classes were replaced with
<code>protected $casts = ['x'=>'datetime',...]</code>
which was the more modern way to get the same functionality, which exists
in both Laravel 8 and 10. For the few model classes that already had other
<code>$casts</code> declarations, the replacements were merged with those.
While the "dates" change could have been its own task that was merged to
trunk prior to and separately from the current Laravel 10 upgrade task, it
was combined with the latter to streamline testing, as both had potential
impacts over a large fraction of the app.
[[#top|RETURN]]
==== Changes For laravel/framework Upgrade's Miscellaneous Changes ====
This task also updated <code>app/Providers/AuthServiceProvider.php</code>
to remove the call <code>$this->registerPolicies();</code> from
<code>boot()</code>because Laravel 10 now calls it automatically, as
recommended here:
* https://laravel.com/docs/10.x/upgrade#register-policies
[[#top|RETURN]]
==== Changes For goldspecdigital/laravel-eloquent-uuid Removal ====
This task also updated these 5 PHP source files to be compatible with the
replacement of <code>goldspecdigital/laravel-eloquent-uuid</code> with a
Laravel built-in:
* app/Models/Application/Application.php
* app/Models/User.php
* app/Models/User/Student.php
* app/Models/User/Teacher.php
* app/Models/User/UserContract.php
These further 3 files also referenced the trait but commented out, so not
current users but possible past or future users:
* app/Models/Application/AppUser.php
* app/Models/User/Address.php
* app/Models/User/Guardian.php
For each of the above 8 files, there were these 2 line subsitutions:
    use GoldSpecDigital\LaravelEloquentUUID\Database\Eloquent\Uuid;
    use Uuid;
    use Illuminate\Database\Eloquent\Concerns\HasUuids;
    use HasUuids;
Here is a description of the above built-in feature in Laravel 9.3+:
https://laravel.com/docs/11.x/eloquent#uuid-and-ulid-keys
The purpose of that reimplemented functionality was to empower use of
generated UUIDs for primary key fields of some database tables instead of
the serially generated integers that ''SDS Laravel'' more typically uses;
Laravel Eloquent only gained built-in support for UUIDs with version 9.3.
This task also deleted the single PHP file
<code>app/Models/Traits/Uuids.php</code> as it appeared to be unused.
[[#top|RETURN]]
==== Changes For fideloper/proxy Removal ====
This task also updated <code>app/Http/Middleware/TrustProxies.php</code> to
be compatible with the replacement of <code>fideloper/proxy</code> with a
Laravel built-in. The changes were in 2 spots.
First was this substitution:
    use Fideloper\Proxy\TrustProxies as Middleware;
    use Illuminate\Http\Middleware\TrustProxies as Middleware;
Second was this substitution:
    protected $headers = Request::HEADER_X_FORWARDED_ALL;
    protected $headers =
        Request::HEADER_X_FORWARDED_FOR |
        Request::HEADER_X_FORWARDED_HOST |
        Request::HEADER_X_FORWARDED_PORT |
        Request::HEADER_X_FORWARDED_PROTO |
        Request::HEADER_X_FORWARDED_AWS_ELB;
See also https://laravel.com/docs/9.x/upgrade under "Trusted Proxies".
[[#top|RETURN]]
==== Changes For juliomotol/laravel-auth-timeout Upgrade ====
This task also updated
<code>app/Http/Middleware/AuthTimeoutMiddleware.php</code> to be
compatible with the <code>juliomotol/laravel-auth-timeout</code> upgrade.
There was this 1 substitution:
    use JulioMotol\AuthTimeout\Middleware\AuthTimeoutMiddleware as BaseMiddleware;
    use JulioMotol\AuthTimeout\Middlewares\CheckAuthTimeout as BaseMiddleware;
Note that <code>juliomotol/laravel-auth-timeout</code> must be upgraded
simultaneously with Laravel since the former's versions 3.1.1 and 4.1
respectively require Laravel 8 and 10 respectively.
See https://github.com/juliomotol/laravel-auth-timeout/blob/master/CHANGELOG.md
for more change details and upgrade notes on that.
[[#top|RETURN]]
== Documentation Changes ==
=== 2024 Jul 16: SDS Laravel: Make README etc Say This App is SDS Rather Than Laravel ===
This task updated ''SDS Laravel'' <code>composer.json</code> to update
primary documentation attributes so they describe the actual SDS
application this is rather than the Laravel template used to build it.
There were these substitutions:
    "name": "laravel/laravel",
    "description": "The Laravel Framework.",
    "keywords": ["framework","laravel"],
    "name": "smus/sds",
    "description": "School Data System (SDS)",
    "keywords": [],
This task also completely rewrote <code>README.md</code> so it describes
this SDS application rather than the Laravel framework.


[[#top|RETURN]]
[[#top|RETURN]]

Latest revision as of 18:46, 20 August 2024


This document consists of multiple parts; for a directory to all of the parts, see SDSM:Index.

Description

This part of the SDS Modernization (SDSM) document enumerates a not necessarily exhaustive list of historical changes or improvements that were made to SDS, made by Darren Duncan if by whom is not otherwise specified.

It exists to provide visibility into historical progress of the SDS Modernization (SDSM) project about which this document otherwise mainly just describes the current state of SDS.

This list often but not always corresponds to the list of Git pull requests at https://git.smus.ca which most readers of this document are not privileged to see directly.

RETURN

Fixes to Broken Behavior

2024 Apr 18: SDS Laravel: User Preferences Screen Fails to Display

Context: User darren.duncan attempted to visit their SDS Laravel User Preferences screen, example urls https://api.smus.ca/User/Preferences (production) and https://api-latest.smus.ca/User/Preferences (preview).

Observed Behavior: In production, a non-graceful generic 500 internal server error message was displayed rather than the expected User Preferences screen. In preview, the revealed underlying symptom was that the code line in PreferencesController of $addresses = $user->guardian->addresses; was dying with Attempt to read property addresses on null. The problem occurred for user darren.duncan but not for other users.

Expected Behavior: The User Preferences screen should always display, regardless of whether the user in question is a guardian or has guardians. If some functionality of the screen is not applicable for users that aren't or don't have guardians, the screen should either hide that functionality or explain why it doesn't apply or both.

Applied Solution: The dying code line was replaced with code that only tried to fetch and return an address list when it was valid to do so, and it returned an empty list otherwise. As a result, the User Preferences screen never failed to display under this scenario.

Caveat: While the User Preferences screen no longer fails to display, as of the completion of this fix, it still remained true that the functionality to actually display the fetched address list in question was never implemented, and so for every user, the screen carried the placeholder message Shows address info instead of each individual address.

RETURN

2024 May 21: SDS Laravel: Timeout Feature Broken In Sub-Directory Hosted Apps

SDS Laravel has a security feature such that if a user is inactive for a period of time, 15 minutes typically, they will be automatically logged out of the app. The app displays a countdown timer on the top right hand corner of the screen that ticks for every second of activity. Javascript code running in the client web browser will normally reset the countdown to the full amount whenever a user interface (UI) event occurs that it considers continued user activity, such as mouse movement over the screen.

This feature is implemented partly by the web client invoking the /session endpoint on the server.

That invocation fails on any SDS Laravel app instance that is hosted at a web address which is a sub-directory of the base url (meaning it is based at an address like https://foo.com/bar rather than at an address like just https://foo.com). This is because the web client is trying to unconditionally treat every app instance as if it is hosted directly at the base url with respect to its attempts to invoke the /session endpoint, and so it is invoking the wrong web address for instances that are at sub-directory urls.

As a result, for broken app instances, while the first UI activity indicating continued user activity will reset the countdown timer displayed on screen to the user, all subsequent UI activity will have no effect on the timer, and it will not reset, and the server will not be aware of this activity. Only a full page load like clicking a link to a new page will register as activity and reset the timer.

This breakage affects all SDS Laravel instances at https://sdsdev.smus.ca which are in sub-directories.

The primary change of this task fixes the problem by making the web client respect the actual location of the app when invoking /session. Mainly it is a 1-line change in the Laravel Blade template file resources/views/layouts/main.blade.php to use url: '{{route('session')}}' rather than url: '/session'.

An additional change of this task is to fix a problem where the app's background image doesn't display for the same underlying reason. The problem is that the web client is trying to load the image public/images/body.png from the wrong location. The fix updates 1 line in the CSS file public/css/smus_custom.css to background: url('../images/body.png') from background: url('/images/body.png'); the newly-relative url is relative to the location of the CSS file itself.

There is still additional broken behavior related to static asset loading like the background image example, affecting custom fonts for example, but these references are in the generated file public/css/app.css as parts of third-party dependencies, and so these were left alone.

RETURN

2024 Jun 12: SDS Laravel: Form Meister Screen Fails to Display

This task made 2 distinct changes.

The first change...

The SDS Laravel screen Admin Menu -> Online Forms -> Form Meister fails to display when visited. In production it displays a generic 500 internal server error message. In preview, the revealed underlying symptom was an attempt to invoke the non-existent PHP class App\Models\System\Form.

This task fixed the offending PHP class reference to instead refer to the existing PHP class App\Models\Abstract\Form. The result was that the Form Meister screen then displayed when visited.

(Note that the SDS Gavintech version of this screen was disabled in favor of the new version, so it was not functional for anyone.)

The second change...

For an idiomatic PHP Laravel project, each of its web controller classes has an unqualified PHP class name in the form FooController.

While nearly all of the 161 SDS Laravel controller classes followed that naming convention, these 3 were lone exceptions:

  • Admin/OnlineForms/EditFormCampaigns.php
  • Admin/OnlineForms/EditFormMeister.php
  • Admin/OnlineForms/MergeFormCampaign.php

This task renamed each of those 3 to add Controller to their unqualified names.

RETURN

2024 Jul 4: SDS Laravel: Functional Screens Not Distinguished in Main Menu Bar

The main navigation menus of SDS Laravel declare a hierarchical list of named screens, but a design decision was made such that, while the app is still incomplete and under development, the menus list many screens that do not yet exist in SDS Laravel as placeholders along with many which do.

There was a serious usability problem with how this was first implemented, such that each menu item did not distinguish in any way whether it was for a screen that existed or whether it was a placeholder; both kinds looked identical visually and both kinds were active hyperlinks that went somewhere, and a user could only tell which kind it was by choosing it and seeing where they went as a result. In the case of placeholders, every one un-helpfully went to the home screen.

This was a problem because users exploring SDS Laravel, whether to use it normally or to test it, would waste a lot of time searching for needles of functional screens in haystacks, with much trial and error, as about half of the total menu item count was placeholders.

This task made 2 related changes for usability without losing anything.

The first change is that now for each menu item which is a placeholder, its name is prefixed by "(N/A)" so it clearly stands out visually.

The second change is that now for each menu item which is a placeholder, its hyperlink is no longer functional, and a user isn't taken anywhere when clicking on it; they stay on their current screen.

Note that common established practice in desktop GUI apps is to use alternate fonts or colors for visual distinction, such as non-active menu items having grey text versus black text for active ones, but this task eschewed that solution given its greater complexity. A future enhancement can do this, however.

Note that an alternate solution could have been to just not display placeholders anymore, so all menu items went to screens, but it was deemed more useful to retain the placeholders.

Note that for the purposes of this menu logic, the determination of whether a screen was implemented or a placeholder was just whether or not a Laravel route was defined for it, in the source file /routes/web.php.

Implementing this change was a single line change in the source file /app/Models/Sub/Page.php.

RETURN

2024 Jul 23: SDS Laravel: Form Campaign Add Target Students Uncheck All Button Fails

The SDS Laravel screen Admin Menu -> Online Forms -> Form Campaign -> Consolidated Health Form -> Manage has broken behavior.

The "Add New Targets" screen section presents a list of students with a checkbox for each student. When visiting the screen, the checkboxes all default to a checked state. Often this is what the user wants but sometimes they want to just pick a few students instead, and to assist with this is an "Uncheck All" button.

Clicking the "Uncheck All" button as the user's first action on this screen succeeds. But if any boxes are manually checked afterwards, clicking "Uncheck All" again does nothing, and those boxes are still checked.

Furthermore, if SDS Laravel is updated to use Vue 3 from its current Vue 2, the same functionality manifests additional broken behavior such that the "Uncheck All" button has no effect even on initially visiting the page, so it doesn't even work a single time to uncheck all the boxes.

This task fixes the broken behaviour by just making a few small updates in the 1 PHP source file resources/views/admin/online_forms/form_campaigns/show.blade.php.

Primarily there was this 1 substitution in the "Uncheck All" button definition:

   onclick="$('#student_list input:checkbox').attr('checked', false);"
   onclick="Array.from(document.getElementsByClassName('checkbox_student_list_add_target')).forEach(function(checkbox) { checkbox.checked = false; });"

In support of that, there was also this 1 addition to each student checkbox:

   class="checkbox_student_list_add_target"

Those 2 line changes are all that was needed to fix the broken behavior.

The underlying problem appears to be that jQuery's mechanism to enumerate the set of checkbox DOM nodes fails to find them when any environmental change occurs such as when users click the checkboxes, so the JavaScript code to uncheck those boxes acts like the checkboxes don't exist. Note that built-ins like querySelectorAll() have the same problem. Whereas, getElementsByClassName() is resilient to such environmental changes and can still find the checkboxes after user actions.

This task also makes a user interface change to enhance users' experience, which is to add a complementing "Check All" button.

RETURN

2024 Aug 15: SDS Gavintech: Input Absences Screen Campus Filter Not Sticky and Other Bugs

The SDS Gavintech screen Staff Menu -> Attendance -> Input Absences has several bugs or otherwise sub-optimal behaviors, and this task makes 7+ distinct changes to respectively fix or improve those.

All of the changes are isolated to the single PHP source file pages/attendance/input_absence.php and so should not impact any part of SDS besides this one screen.

The first/primary change...

The filter toggle "Show all campuses"/"Restrict to campus only" had a bug such that the setting of this toggle was not sticky; as soon as the user took any action which reloaded the screen, such as adding/editing/deleting an input absence record, or such as flipping the filter toggle "Day students only"/"All students", the campus toggle would reset to its default of showing the current campus only. Besides this going against user expectations and making more work for them, it is also inconsistent with the behavior of the day students toggle which generally did persist through any corresponding user actions.

This task normalized the behavior of the screen so that both of these filter toggles behave the same and persist through any actions the user takes on the screen; they would only reset to defaults if the user leaves the screen entirely and returns.

In particular, this task's change makes things better because if one uses the screen to add an absence for a student in another campus, then they can actually see the record they just added on the screen after they submitted the form, while otherwise it would look more like their action had no effect, besides the transient status message of "Success: Added entries".

The second change...

This task also made a behavior change such that the current setting of the campus filter toggle is now reflected in the title of the screen with a "(campus only)" or "(all campuses)" suffix, so its more clear to users what the setting is. This also makes the screen more consistent with the other filters, such that the day students filter had a corresponding title suffix already, "(day students)" and "(all students)" respectively, and the date filter which shows the date its filtered to.

The third change...

This task also made a behavior change such that the value of the "Set date" form/button now defaults to the same date the screen is currently displaying, rather than the real-world current date, which is effectively like the screen not persisting their choice to an extent.

I feel that behavior is more useful and in line with user expectations or less surprising to them. It also means that this form is more thoroughly a localized version of the application global "Date" setting. That is, changing the date globally also affects this screen, but changing it on this screen only affects this screen, until the user leaves this screen and it resets to match the global setting.

The fourth change...

This task also made a behavior change such that the link to "Print list of absent day students" has been renamed to "Print list of absent students", and it now displays unconditionally instead of only when the screen is filtered to just day students.

This new behavior was deemed more useful such as if we have an emergency on campus and we have to evacuate, a full list would indicate day students with absences entered by Ally/Bonita and boarding student absences entered by the Health Centre.

However, there was no behavior change with respect to what actually happened when that link was clicked on; because of a prior bug, the link always led to a list of all students anyway, rather than filtering to show just day students.

The fifth change...

This task also fixed a bug in the "Existing absences" -> "Entered by" column such that the HTML for the link to email the person was malformed and the url did not go anywhere / got a 404, but with the fix it correctly goes to the email screen.

The sixth change...

This task also removed the link "Email Advisors & Students", which was never implemented and did nothing but reload the page when clicked on.

The seventh+ changes...

This task also further normalizes how urls are generated that point back to the current screen, reducing duplication of repeated details, and also normalizing that arguments for filters only appear in urls if they did before or are not defaults.

This task also normalizes some other implementation code, such as specifying that all the "Dropdown" calls return their results rather than printing them.

RETURN

Fixes to Slow Performance

2024 May 14: SDS Laravel: Main Menu Bar Heavy Database Use Very Slow Page Loads

SDS Laravel currently relies on the "pages" database table to canonically define its app main navigation menus and it currently has 201 records where each defines a menu item.

The main menus are hierarchical, and the "order" field of each "pages" record serves double duty to define the parent/child aka super/sub relationships of menu items as well as their display order. The "order" of a record is used as a unique key for the record for this purpose, as well as a foreign key from child to parent.

Each "order" is a positive integer N such that:

  • The result of log-100(N) is the item's tree level.
  • The result of div-by-100(N) is the item's parent item's "order" in the tree.
  • N itself is the item's display order relative to its siblings.

For example, a root level menu item has an "order" in 1..99, and for the root item with an "order" of 3, all of its children would have "order" in 301..399, and the child of 303 would have "order" in 30301..30399 etc.

(Very notably, the actual primary key field of "pages" is NOT used for describing menu parent/child relationships in conjunction with a second field, as is the more typical practice in other databases.)

SDS Laravel reads the entire "pages" database table and constructs a hierarchical menu in 2 different circumstances. The primary and most important circumstance is for nearly every single screen, those where the user is logged in, and it is a normal screen, and not an API, a menu for navigation appears at the top of each screen. The secondary circumstance is on the "System Settings"->"Pages" screen that is used to edit this database table, where the hierarchy is rendered as a left column.

Prior to the changes of this task, the hierarchical menu assembly was performed in a very inefficient manner such that the application made a separate database query for every single tree node to fetch its children, meaning that it was making 138 queries for nearly every screen displayed to the user just for making the navigation menu alone. And on the "Pages" screen it was making 427 additional queries (total 565) just to display the menu hierarchy on that one screen. For context, the sum total of all other database queries for all other purposes on a single screen was 20-30 or so.

Compounding the above query-per-node inefficiency was another kind of inefficiency where, for each menu node, the same query for its children was often done up to 3 times (or up to 5 times), where only the last query actually enumerated the list of child nodes, and prior times simply tested if there were any children, and if there were children, the query was run again to actually use them.

Following the changes of this task, the hierarchical menu assembly only involved 2 SQL queries as overhead for nearly every screen, and 2 more on the "Pages" screen, because it now selects all the "pages" records at once rather than a handful or zero at a time, and uses the same fetched records when building every menu node. The records are grouped when fetched in a PHP array of arrays, the root array keyed by parent "order" values, and the array value for each being an array of its child "pages" records.

This change had a massive impact on the user-visible screen loading speed of SDS Laravel, particularly when a developer runs the application and its database on opposite ends of a VPN tunnel where the added latency of each database call magnified the overhead time of each database query.

For example, before the changes, running the app on a local machine at home, where it talked to a database on a school server, took about 23..25 seconds (155 queries) on the fastest screen with a menu to display on that screen, and after the changes it took about 3.4-3.7 seconds (19 queries). And on the "Pages" screen it took about 1.5 minutes (581 queries) to display the screen before the changes, and about 4.1-4.2 seconds (20 queries) after the changes.

For a typical non-developer user scenario which didn't have that kind of added VPN latency, the time to load a faster screen before the changes was around 0.46-0.49 seconds, and after the changes it was about 0.12-0.13 seconds. For the "Pages" screen it was about 1.8-1.9 seconds before and about 0.1-0.3 seconds after.

This task kept its changes about as minimal as possible while making this change. There was no significant change to how screens or menus were defined, and anything else related to the application design, aside from the minimal needed to consolidate the database queries. While there is a lot of room to make other improvements to how menus are implemented, other such changes are out of scope for this task.

Here are specific numbers for before/after, school/home-VPN, on 4 screens.

SDS Laravel/API Screen: /

  • Blade Views Rendered: 1
  • Database Queries Run Before: 0
  • Database Queries Run After: 0
  • Eloquent Models Used Before: 0
  • Eloquent Models Used After: 0
  • HTTP Page Load Time at School Before: about same
  • HTTP Page Load Time at School After: 58.8ms,59.9ms,55.3ms,59.3ms,63.1ms
  • HTTP Page Load Time at Home/VPN Before: about same
  • HTTP Page Load Time at Home/VPN After: about same

SDS Laravel/API Screen: /home

  • Blade Views Rendered: 3
  • Database Queries Run Before: 192
  • Database Queries Run After: 56 (136 fewer)
  • Eloquent Models Used Before: 455
  • Eloquent Models Used After: 475
  • HTTP Page Load Time at School Before: 672ms,671ms,734ms,725ms,617ms
  • HTTP Page Load Time at School After: 362ms,416ms,364ms,271ms,256ms
  • HTTP Page Load Time at Home/VPN Before: 30.22s,30.78s,29.54s,29.90,29.83
  • HTTP Page Load Time at Home/VPN After: 9.30s,9.07s,9.21s,9.24s,9.35s

SDS Laravel/API Screen: /Event/Calendar/33297

  • Blade Views Rendered: 4
  • Database Queries Run Before: 155
  • Database Queries Run After: 19 (136 fewer)
  • Eloquent Models Used Before: 200
  • Eloquent Models Used After: 220
  • HTTP Page Load Time at School Before: 462ms,487ms,589ms,493ms,491ms
  • HTTP Page Load Time at School After: 126ms,131ms,125ms,121ms,125ms
  • HTTP Page Load Time at Home/VPN Before: 24.06s,25.20s,24.06s,24.67s,23.72s
  • HTTP Page Load Time at Home/VPN After: 3.38s,3.41s,3.39s,3.72s,4.73s

SDS Laravel/API Screen: /System/Page

  • Blade Views Rendered: 34
  • Database Queries Run Before: 581
  • Database Queries Run After: 20 (561=136+425 fewer)
  • Eloquent Models Used Before: 779
  • Eloquent Models Used After: 436
  • HTTP Page Load Time at School Before: 1.86s,1.85s,1.79s,1.88s,1.80s
  • HTTP Page Load Time at School After: 170ms,207ms,173ms,316ms,208ms
  • HTTP Page Load Time at Home/VPN Before: 1.5min,1.5min,1.5min
  • HTTP Page Load Time at Home/VPN After: 4.16s,4.17s,4.09s,4.12s,4.12s

RETURN

2024 Aug 20: SDS Gavintech: Input Absences Screen Heavy Database Use Very Slow

The SDS Gavintech app has widespread structural inefficiencies in the way that it uses the database, and this manifests as the primary cause of very slow app performance for users. The app may make 1000s or 10000s of distinct database calls every time a user visits any screen or performs any action on the screen that involves reloading the page, when it should only be making on the order of 1-20 database calls total each such time. As a result, each user action that loads/reloads a screen may carry a delay of 1-2 seconds at the bare minimum and often 10-20 seconds or several minutes, which cumulatively is a very poor user experience.

To be specific, that was the experience of the production SDS instance in early June of 2024. A pending migration as of this writing to newer production hardware would yield performance improvements of 2-5X alone, which helps significantly, but is still very noticeably slow. In contrast, changes to the app source code to correct the structural inefficiencies would instead yield further performance improvements in the 15X+ range, such that user actions then actually may seem instant. Meanwhile, any specific metrics given below for this task are all using the same newer hardware, to be apples to apples comparisons, and also each version is refreshed several times and the shortest generation time is used.

This task makes several distinct changes specific/isolated to the 1 screen Staff Menu -> Attendance -> Input Absences such that the logic specific to that screen is optimized in how it uses the database.

As a result of this task's changes, in order to display the screen, a total of exactly 8-9 database calls are made the screen-specific elements (4 trivial ones plus 4-5 main ones), rather than about 9594 calls.

Further, as a result of this task's changes, in order to bulk-add new absence records, a total in typical cases of exactly 4 database calls are made for the database updating logic (2 trivial ones plus 2 main ones), or a typical extreme maximum of about 24 database calls if adding around 1100 students, rather than a minimum of 6 calls and a maximum of about 4000-5000 respectively.

As a result of this task's changes, all user actions on the Input Absences screen that involve a screen load/reload now take about 0.1 seconds, which is essentially immediately, including displaying and bulk adding, as far as the work specific to the screen itself is concerned. Unlike the old logic, the new logic should scale and have essentially this same speed no matter how many students there are, rather than slowing down with more students.

This should be very noticeable to users, as every single individual action should feel a lot faster, which is even more significant given a typical user workflow of performing dozens to a hundred actions here each day.

As a result of this task's changes, the amount of server memory used also goes down by orders of magnitude, from about 168MB to about 4MB usually (2MB in some cases), a 42X savings, as far as the work specific to the screen itself is concerned. If widely applied, such changes can also mean SDS can scale to handle orders of magnitude more concurrent users on the same hardware without users seeing further slowdowns due to competition for resources with other concurrent users.

All of the changes are isolated to the single PHP source file pages/attendance/input_absence.php and so should not impact any part of SDS besides this one screen.

Prior to the changes of this task...

This is the overhead for every screen minus that specific to the main menus; it comprises 16 generally distinct simple database calls:

   Page generation time: 0.011
   Total SQL Queries: 16
   Total SQL Cache Hits: 15
   Total Local Cache Hits: 2
   Total Memcache Hits:
   Memory Usage: 2048K
   Realpath Usage: 87K

This is the overhead for every screen inclusive of the main menus, when using SDS with the effective date of 2024 May 6 and as a test/staff user "Darren Duncan" that is not a teacher/parent/student; the main menus alone add 1567 database calls:

   Page generation time: 0.231
   Total SQL Queries: 1583
   Total SQL Cache Hits: 7078
   Total Local Cache Hits: 10
   Total Memcache Hits:
   Memory Usage: 32768K
   Realpath Usage: 125K

This is the overhead for every screen inclusive of the main menus, when using SDS with the effective date of 2024 May 6 and as a teacher user "Carol Adamson", who also has a "My Courses" main menu but lacks some other staff menu items; the main menus alone add 2049 database calls:

   Page generation time: 0.27
   Total SQL Queries: 2065
   Total SQL Cache Hits: 9188
   Total Local Cache Hits: 65
   Total Memcache Hits:
   Memory Usage: 57344K
   Realpath Usage: 87K

This is Input Absences screen on effective date 2024 May 6 by user Darren Duncan including all the above overhead (user Carol Adamson doesn't have privileges to see this screen); the initial filter settings of current/Senior campus only plus all day plus boarding students (about 600 students) there are 10131 database calls total:

   Page generation time: 1.471
   Total SQL Queries: 10131
   Total SQL Cache Hits: 31468
   Total Local Cache Hits: 5351
   Total Memcache Hits:
   Memory Usage: 180224K
   Realpath Usage: 125K

This is the same Input Absences screen initial load but with the main menu generation disabled, so it is more clear what this screen alone is doing; there are 9594 database calls that are specific to this screen; many of those calls are the same as ones in the main menu generator, and those same calls are cached / only made once when both the main menu generator and this screen's logic are enabled:

   Page generation time: 1.406
   Total SQL Queries: 9610
   Total SQL Cache Hits: 23359
   Total Local Cache Hits: 5343
   Total Memcache Hits:
   Memory Usage: 167936K
   Realpath Usage: 126K

Note that all further numbers below likewise have the main menu generation disabled, to just highlight what Input Absences alone is doing.

This is the same thing but with the campus filter set to all campuses and showing both day and boarding students (about 1000 students); there are 3514 database calls specific to this screen; there are actually 2/3 fewer database calls despite showing more students:

   Page generation time: 0.499
   Total SQL Queries: 3530
   Total SQL Cache Hits: 17071
   Total Local Cache Hits: 289
   Total Memcache Hits:
   Memory Usage: 65536K
   Realpath Usage: 88K

This is the same thing but with the filter set to day students only, and campus filter still main only; there are 9541 database calls specific to this screen; unlike the campus filter, the day student filter only has a small effect:

   Page generation time: 1.462
   Total SQL Queries: 9557
   Total SQL Cache Hits: 20482
   Total Local Cache Hits: 5280
   Total Memcache Hits:
   Memory Usage: 165888K
   Realpath Usage: 126K

This is the same thing but with both filters toggled such that it shows day students only plus all campuses; there are 2942 database calls specific to this screen:

   Page generation time: 0.469
   Total SQL Queries: 2958
   Total SQL Cache Hits: 17827
   Total Local Cache Hits: 217
   Total Memcache Hits:
   Memory Usage: 55296K
   Realpath Usage: 89K

About 80-90% of the database calls specific to this screen are about fetching a list of currently enrolled students for the effective date, with or without campus/day filters, and using that list both to populate a picklist of students and also to filter a display of existing absense records, because it is making a handful of separate database calls for each of about 1000 students (both when all 1000 are used and when some aren't). It also makes a handful of separate database calls for each existing absense record displayed, but these tend to be on the order of dozens of records instead of 1000.

When the user performs an action like adding/editing/deleting an absence record on this screen, the actual database modification task is typically very fast, assuming just a single student updated, usually around 2-6 database calls for that, but then the page reloads carrying the large aforementioned overheads. As an exception to this, when the user performs a bulk addition of absence records, by selecting multiple students, each individual student involves 4-5 database calls, of which 2 are reads and 2-3 are updates, and so in the unusual case of adding all thousand at once, it can make around 4000-5000 database calls, and take around 15 seconds on the newer hardware to do so; and then the page reloads.

Following the changes of this task...

This is the same thing, on just displaying the Input Absences screen, for effective date 2024 May 6, for user Darren Duncan; there are 8-9 database calls specific to this screen; it is 9 when filtering to day students only, and 8 otherwise; it is unchanged by filtering on one or all campuses:

   Page generation time: 0.093
   Total SQL Queries: 24
   Total SQL Cache Hits: 25
   Total Local Cache Hits: 5
   Total Memcache Hits:
   Memory Usage: 4096K
   Realpath Usage: 54K
   Page generation time: 0.108
   Total SQL Queries: 25
   Total SQL Cache Hits: 26
   Total Local Cache Hits: 5
   Total Memcache Hits:ƒ
   Memory Usage: 2048K
   Realpath Usage: 92K

In addition, the logic for batch-adding adds almost no time to that in typical cases of 1 or few students, and otherwise a small fraction of 1 second for the extreme case of bulk-adding all thousand students at once.

The main general change made by this task is to clone several call chains worth of source code for several dozen routines across a handful of classes into input_absence.php and iteratively refactoring that whole code mass to its most concise specific-to-this-screen's-use-case version. No code shared with other screens was altered, allowing a guarantee that no other screens are affected by these changes, making it much lower risk, and a much simpler task to accomplish.

The most important effect of the main change is the consolidation of database calls. Typically how the logic worked before this task's changes was that in order to produce a list of something, first you had a query that just produced a set of record ids for the list we wanted, and then a separate query was made for each of those ids to get other fields we want for each record, sometimes in one call and sometimes in a handful, some for display purposes, and others for filtering purposes. Hence you may have around 3000-8000 calls to deal with a list of around 1000 students. This task changed things so that exactly 1 SQL query was done instead in place of that first id-fetching query that also gathered all the related details.

The 4-5 display-only main queries then were for these purposes:

  • One to derive the input time range for the effective date, eg 8:30am-3:20pm, from a period list.
  • One to display the picklist of students to add/edit absences for.
  • One to display the picklist of attendence codes.
  • One to display the list of existing absence records.
  • Iff filtering students by day, one to fetch a list of fee statuses to filter with.

(While in theory the last one could also have been inlined, I chose not to because it was relating fee statuses with "like" rather than equality.)

There was also these 3 add-only SQL statements per each batch of 100 students:

  • One to test that the user-supplied student ids match student records.
  • One to insert up to 100 absence records.
  • An indirectly called one to insert a log entry for the prior insert.

In the interest of security and scalability I compromised my adding logic between the pre-changes version of a separate database call per single student and using a single database call for the entire list of 1000+ students, into batches of 100, which is essentially the same performance as all-in-one but shouldn't be as likely to fail in pathological cases.

A related general change is that the main screen logic no longer turns database records into cached PHP objects that it then calls methods on, and instead just works with the records as regular PHP arrays which typically don't correspond to single database tables but joins of several. The logic goes from interacting heavily with the app's record object cache to generally not interacting with it at all, either to add or remove items. In particular, the logic no longer has up to a handful of repeatedly converting back and forth between record ids and record objects per record. Between this, and the greatly reduced use of the app's SQL query cache due to fewer database calls, likely is the main reason why so much less memory is used by PHP in the new version.

Note that while some database queries or other logic could possibly be streamlined or simplified further, I tried to maintain the exact effective logic as possible from before, in case there might be subtle behavior changes from such further streamlining. But an actual future rewrite of this screen such as under SDS Laravel could likely revisit this.

Generally speaking, the screen logic I overhauled was that where many similar database queries could be consolidated, while logic that only had individual distinct queries, such as related to getting the current school day, I left alone.

RETURN

2024 Aug 20: SDS Gavintech: Attendance Code / Reason Picklists Slow

This task changes the SDS Gavintech app shared component that is used by all or most screens with an Attendance Code / Reason picklist so that it internally uses the database more efficiently and thus performs better. For a typical scenario of displaying 20 attendance codes, it now performs this using exactly 1 database call rather than 21 database calls. This benefit is seen on every screen using this picklist, not just 1.

While the impact of this change alone on a single user action is likely too small for a user to notice, it still contributes to efficiency, and in aggregate, performance improvements users would notice.

That change was entirely in the single PHP source file overrides/classes/class_dropdowns.php, in its single routine attendanceCode().

This task also changes the internals of the 1 screen Staff Menu -> Attendance -> Input Absences such that it now uses the same, now improved, app component for its Reason picklist, rather than duplicating it.

That change was entirely in the single PHP source file pages/attendance/input_absence.php.

RETURN

Internal Design Changes

2024 Apr 19: SDS Laravel: Consolidate Eloquent model classes to /app/Models

An idiomatic PHP Laravel project has its main files grouped into the specific hierarchy of folders documented here:

https://laravel.com/docs/11.x/structure

Of particular relevance, the folder /app/Models is where all of the Laravel Eloquent model classes go.

Prior to the performance of this task, SDS Laravel conformed to the idiomatic folder layout in all ways except with respect to the Eloquent model classes, which it had instead spread across about a dozen /app/FooModel folders.

This task moved/renamed all of the Eloquent model classes, such that for each class /app/FooModel/Bar.php, it was moved/renamed to /app/Models/Foo/Bar.php. While that described nearly all of those classes, a handful started with other names, but they were appropriately gathered under /app/Models too.

This task also included a normalization of how first-party classes were referred to in other code, such that any which weren't referred to by their fully-qualified names before were referred to in that way after.

RETURN

2024 May 8: SDS Laravel: Sever Coupling of PHP Class Names to Database Data

Laravel Eloquent has a particular more-advanced feature for defining relationships between model classes that it refers to as a polymorphic relationship, such that a relationship is not just between 2 specific model classes, but rather between 1 model class and 1 set of all model classes that fulfill a particular role.

https://laravel.com/docs/8.x/eloquent-relationships#polymorphic-relationships

Prior to the performance of this task, SDS Laravel utilized such an Eloquent relationship, between Group on one side, and the set comprising Page and SubPage on the other side.

And it did so in a design-flawed way that led to tight coupling between the exact fully-qualified PHP class names of the latter 2 model classes, and the database data, such that those fully-qualified class names (App\SubModel\Page and App\SubModel\SubPage) were stored in the database (specifically in the group_pages table's class_type column) and the fact that these matched was relied on by the application for correct behavior.

This tight coupling manifested as broken app behavior when attempting to move/rename the Page and SubPage classes (to App\Models\Sub\Page and App\Models\Sub\SubPage), because then their names no longer matched what was in the database.

The most common kind of breakage was that, for almost every app screen that was gated with a user group privilege check for the screen, the app thought no user had any privilege to use it; for example, this error message displayed when trying to app general search screen:

   No access to Search.index
   You do not appear to have permission to access https://api-latest.smus.ca/Search as [User's Name]. If you believe you should have access
   to open a help ticket request permission, or you can go back / home and try again.

And had users been able to get past that, certain other screens or functionality that rely on displaying or editing user group privileges for pages and subpages would have had problems, such as failing to see existing privileges made before the rename (they might have worked to re-add the privileges, then using the new hard-coded class names in the database).

Without correcting the design flaw, renaming these model classes would have required a corresponding database migration to replace the old hard-coded-in-data PHP class names with their new ones, which carried more complexity and risk.

The more specific design flaw was the use of Eloquent's morphToMany() relationship builder without specifying an explicit constant string to represent each morphed class in the database, resulting in Eloquent defaulting to using its fully-qualified PHP class name instead.

Both Page and SubPage had a copy of this method (this version is post-rename) which was affected:

   public function groups(){
       return $this->morphToMany(\App\Models\Sub\Group::class, 'class', 'group_pages');
   }

This task made the simplest possible fix to this design flaw by explicitly configuring Eloquent to use an explicit constant string per each morphed class, by adding the following in the file app/Providers/AppServiceProvider.php:

   Relation::enforceMorphMap([
       'App\SubModel\Page' => 'App\Models\Sub\Page',
       'App\SubModel\SubPage' => 'App\Models\Sub\SubPage',
       // TODO: Migrate database to use non-PHP-class-looking slugs
       // like following instead so they don't look tightly coupled.
       // 'page' => 'App\Models\Sub\Page',
       // 'sub_page' => 'App\Models\Sub\SubPage',
   ]);

This was the simplest possible fix, and ostensibly the most elegant, by working with the system, using an Eloquent feature expressly designed for this very scenario, rather than say trying to avoid using the morph features (although doing so may be better longer term), and also by not requiring any database changes, so it works with the database as-is.

https://laravel.com/docs/8.x/eloquent-relationships#custom-polymorphic-types

Following the changes of this task, there was still room for an even more elegant solution to follow that alters the database to use a non-php-class-looking constant in the database data to represent each class, so the app isn't forever using the old class name there, but that was considered higher complexity and risk and suited to be done as a separate task another day.

Note that a consequence of using Relation::enforceMorphMap() is that it makes Eloquent treat the entire app more strictly, such that if we try to use morphToMany() or similar Eloquent methods with any other classes besides Page and SubPage, it will require us to add an entry to the above list for it, rather than silently defaulting to the old behavior for other unspecified classes. I consider this strictness to be a good thing.

RETURN

2024 Aug 21: SDS Gavintech: Sever Coupling of Main Menu Builder to Write/Edit Reports Screen

This task refactored SDS Gavintech to sever the coupling between the main menu builder, defined in the single PHP source file overrides/modules/sidebar_definition/smus_sidebar_definition.php, and the screen known alternately as My Courses -> $course_name -> Reports & Marks -> Write reports -> $student_name and Staff Menu -> Advisor stuff -> Edit Reports -> $student_name -> $course_name, defined in the single PHP source file pages/reports/reports_entry_spell.php.

That coupling took the form of the main menu builder fetching a list of students from the database on behalf of the screen and storing that list of students in the PHP session variable $_SESSION["reports_entry_spell"], which was subsequently read by the screen. The use of this session variable was one-way; the menu builder only set it and the screen only read it. The menu builder set it unconditionally from scratch, and so could have just used a PHP "global" instead, which would have saved the app from the overhead of serializing and deserializing its value to disk, which may have consisted of a thousand "Student" PHP objects. The menu builder set it anew every time the user loaded/reloaded any app screen, and not just when viewing the specific screen it was used on, meaning almost all the time it consumed resources to do that work for nothing.

The changes of this task consisted mainly of placing a stripped down clone of the main menu builder source file into the top of the screen source file, just the parts that built the student list. The changes otherwise consisted of entirely eliminating the use of the PHP session variable.

The menu builder source file had no changes except for removing the 2 lines that set the session variable; the screen source file used a local variable to hold the student list instead of reading it from the session variable; the stripped down clone it gained totalled 70 lines of code, which was 8% of the menu builder source file's lines.

Following the changes of this task, the actual performance of the app was identical, using exactly the same count of database calls per each screen as prior to the changes of this task, and also the same page generation time. This was tested both on the app home screen, which used the menu builder logic but not the screen logic, and on the screen, which used both. Despite the duplicate database call main logic, this would be explained by these database calls being mediated by both an app SQL query cache and an app PHP object cache, which recognized that the results of the first instance of the calls by the menu builder could be reused by the screen.

This decoupling resulted in the app being easier to maintain, as one can now change either the menu builder or the screen arbitrarily without being concerned that changes to one would affect the other, as conceptually the two things should be completely separate.

RETURN

Changes to Third-Party Dependencies

2024 May 3: SDS Laravel: Stop Requiring dbal, revisionable-upgrade, laravel-cors

Prior to the performance of this task, the SDS Laravel Composer config file composer.json explicitly required these 3 third-party PHP library dependencies, which were not actually used by the app:

  • doctrine/dbal (^3.5)
  • fico7489/laravel-revisionable-upgrade (*)
  • fruitcake/laravel-cors (^2.0)

This task updated composer.json to remove those unused explicit dependencies, which also had the effect of removing 6 other Composer-resolved dependencies.

RETURN

2024 May 6: SDS Laravel: Reflect rename of fzaninotto to fakerphp

This task updated SDS Laravel composer.json to reflect that the third-party PHP library dependency fzaninotto/faker was renamed to fakerphp/faker (in fact, that rename occurred prior to the release of Laravel 8), and also require its latest version.

RETURN

2024 May 6: SDS Laravel: Upgrade PHPUnit from 9.x to 10.x

This task updated SDS Laravel composer.json to require the latest PHP-8.1-compatible version of the PHP dev dependency PHPUnit from 9.6.19 to 10.5.20.

RETURN

2024 Jul 15: SDS Laravel: Upgrade Laravel Mix from 5.x to 6.x and Upgrade Webpack

This task updated SDS Laravel so that its JavaScript build pipeline is implemented by Laravel Mix 6 rather than 5, and likewise a newer major version of Webpack.

This task changed these dependencies in package.json:

  • cross-env (^7.0.3 removed as laravel-mix has its upgrade built-in)
  • postcss (^8.4.39 added)
  • laravel-mix (^5.0.9 to ^6.0.49)
  • vue-loader (^15.4.2 to ^15.11.1)

This newer pipeline should be usable in the same way as the prior one, with npm run prod and npm run dev etc, but Laravel Mix 6 makes it considerably more streamlined and ostensibly considerably less fragile than the Laravel Mix 5 version was. Much less underlying Webpack API details are exposed in the "scripts" commands, and also cross-env is no longer directly used and is dropped as a dependency.

The following substitution was performed in package.json:

   "scripts": {
       "dev": "npm run development",
       "development": "cross-env NODE_ENV=development node_modules/webpack/bin/webpack.js --progress --hide-modules --config=node_modules/laravel-mix/setup/webpack.config.js",
       "watch": "npm run development -- --watch",
       "watch-poll": "npm run watch -- --watch-poll",
       "hot": "cross-env NODE_ENV=development node_modules/webpack-dev-server/bin/webpack-dev-server.js --inline --hot --config=node_modules/laravel-mix/setup/webpack.config.js",
       "prod": "npm run production",
       "production": "cross-env NODE_ENV=production node_modules/webpack/bin/webpack.js --no-progress --hide-modules --config=node_modules/laravel-mix/setup/webpack.config.js"
   },
   "scripts": {
       "dev": "npm run development",
       "development": "mix",
       "watch": "mix watch",
       "watch-poll": "mix watch -- --watch-options-poll=1000",
       "hot": "mix watch --hot",
       "prod": "npm run production",
       "production": "mix --production"
   },

Two other files each had a single line change or addition for compatibility with the Laravel Mix version change.

The following substitution was performed in webpack.mix.js:

   mix.js('resources/js/app.js', 'public/js')
       .sass('resources/sass/app.scss', 'public/css');
   mix.js('resources/js/app.js', 'public/js')
       .vue()
       .sass('resources/sass/app.scss', 'public/css');

That is Laravel Mix 6 no longer assumes one is always using Vue, so now one opts into processing its templates explicitly with the .vue().

The following substitution was performed in resources/js/app.js:

   window.Vue = require('vue');
   window.Vue = require('vue').default;

Note that Laravel Mix 6 is required for Vue 3 compatibility but it also works with Vue 2; the latter is still in use as of this task's changes.

RETURN

2024 Jul 15: SDS Laravel: Replace laravel/helpers With laravel/framework

This task updated SDS Laravel composer.json to purge the PHP library dependency laravel/helpers, and also updated any SDS Laravel PHP source files that used it to instead use laravel/framework built-ins.

These 4 PHP source files were updated:

  • app/Http/Controllers/Auth/RegisterController.php
  • resources/views/admin/subjects/edit_rotating_courses/show.blade.php
  • resources/views/student/student_mark_collections/edit.blade.php
  • resources/views/student/student_mark_collections/show.blade.php

For each of the above 4 files, there were either of these 3 subsitutions:

   str_contains(...)
   str_random(...)
   str_replace_array(...)
   Str::contains(...)
   Str::random(...)
   Str::replaceArray(...)

Note that laravel/helpers was just a single PHP source file providing trivially thin wrapper functions over laravel/framework 6+ built-ins that emulated ones from Laravel 5.8, and we were already mostly using the newer versions, so this task just replaced the last 7 instances where we weren't.

RETURN

2024 Jul 17: SDS Laravel: Upgrade Font Awesome from 4.x plus 5.x plus CDN to 6.x

This task updated SDS Laravel to require the latest version of the JavaScript library dependency Font Awesome, to 6.x from a mixture of 3 different versions 4.x plus 5.x plus CDN.

This task changed these dependencies in package.json:

  • @fortawesome/fontawesome-free (^5.15.4 to ^6.6)
  • font-awesome (^4.7 removed as @fortawesome/fontawesome-free is its upgrade)

This task deleted references to this CDN dependency in 3 app Blade template PHP source files:

  • https://use.fontawesome.com/releases/v5.7.1

This task also made a small tweaks to a few other files for compatibility, replacing font/symbol/style name fa with far.

RETURN

2024 Jul 17: SDS Laravel: Stop Requiring tempusdominus, bootstrap-select, jquery-ui, lodash

Prior to the performance of this task, the SDS Laravel config file package.json explicitly required these 5 third-party JavaScript library dependencies, which were not actually used by the app:

  • bootstrap-select (^1.13.18 removed as not in use)
  • jquery-ui (^1.13.3 removed as not in use)
  • lodash (^4.17.21 removed as only used indirectly by laravel-mix)
  • tempusdominus-bootstrap-4 (^5.39.2 removed as not in use)
  • tempusdominus-core (^5.19.3 removed as not in use)

This task updated package.json to remove those unused explicit dependencies.

This task also updated 5 other app files for compatibility, typically by deleting explicit "require" or similar statements to these libraries since they weren't actually used otherwise.

However, these 2 files were special:

  • resources/views/admin/subjects/edit_subject/show.blade.php
  • resources/views/application/studentinfo/index.blade.php

Those 2 files actually DID use removed dependencies, and so technically they were broken by this task. But the 2 screens in question were not currently in use and so temporarily breaking them was acceptable. One screen was used historically but not any more. The other screen was still under construction. This task added comments to those 2 files explaining that they will need to be modified to use alternate and modern dependencies if they are going to be completed or used again.

RETURN

2024 Jul 18: SDS Laravel: Upgrade axios

This task updated package.json to upgrade the JavaScript library dependency axios from 0.27.2 to 1.7.2.

RETURN

2024 Jul 25: SDS Laravel: Upgrade Vue.js from 2.x to 3.x and Upgrade Co-Dependent JS deps

This task updated package.json to require the latest versions of vue as well as of all of its co-dependent JavaScript library dependencies.

To be specific, it made these dependency changes:

  • @formkit/pro (^0.119.14 added)
  • @formkit/themes (^1.6.5 added)
  • @formkit/vue (^1.6.5 added)
  • css-loader (^7.1.2 added)
  • vue (^2.7.16 to ^3.4.34)
  • vue-loader (^15.11.1 to ^17.4.2)
  • vue-bootstrap-datetimepicker (^5.0.1 removed as @formkit/pro "datepicker" is used in its' place)
  • vue-template-compiler (^2.7.16 removed as vue 3 has its upgrade built-in)

Several changes for compatibility were made to the app source files.

Partially rewrote source file resources/js/app.js to account for Vue 3 not having a single global state like Vue 2 but multiple possible object-scoped states. Many of the same changes were previously made in SDS Gavintech which served as a guide.

Rewrote source file resources/js/components/inputs/DateTime.vue to use @formkit/pro rather than vue-bootstrap-datetimepicker. The new version is a clone from SDS Gavintech which had the corresponding changes made earlier.

Updated source file resources/js/components/admin/finance/Fees.vue in several ways (some previously done in SDS Gavintech, some brand new). As Vue 3 removed the filters feature, their use was replaced with ordinary function calls; x|format_year was replaced with format_year(x). As Vue 3 reversed the precedence of v-if plus v-for in the same HTML/template tag, added a child template tag of the tr that used both, to force the expected precedence.

Updated source file resources/js/components/application/AddressInfo.vue to fix a syntax/logic error that Vue 2 overlooked but Vue 3 doesn't.

The logic on the System Settings -> Pages screen to fold/unfold the page hierarchy was disabled as it was incompatible with Vue 3 and not essential so the tree is just permanently unfolded. This was done by small changes in the source file resources/views/system/page/index.blade.php.

The System Settings -> Groups screen was altered to be read-only because the logic for adding/editing/deleting database records was incompatible with Vue 3, a fix was elusive, and this was a good enough 90% solution for the short term given the infrequent use of the screen; a follow-up task can be performed later to enable editing on this screen again when needed. This was done by small changes in the source file resources/views/system/group/index.blade.php.

Formatting configuration for date pickers was updated on a dozen or two screens replacing any LL and LLL shorthands with their longhands, namely D MMM YYYY and D MMM YYYY HH:mm respectively, since the new date pickers don't support the shorthands, only the longhands.

This task also resulted in unintended changes in the app's visual appearance, such as fonts being a bit smaller or some elements being in slightly different locations. The cause of this proved elusive, and since it isn't actually a functional break, this didn't hold back the Vue 3 upgrade. A follow-up task can be performed later if changes to the visuals are desired.

RETURN

2024 Aug 2: SDS Laravel: Upgrade directorytree/ldaprecord-laravel from 2.x to 3.x

This task updated SDS Laravel composer.json to require the latest version the PHP library dependency directorytree/ldaprecord-laravel, going from 2.7.3 to 3.3.3.

This task also made 3 distinct sets of app source code changes to be compatible with that upgrade.

See also for context:

The first set of changes...

This task updated the 1 PHP config source file config/ldap.php to account for logging now being an array. There was this 1 substitution:

   'logging' => env('LDAP_LOGGING', true),
   'logging' => [
       'enabled' => env('LDAP_LOGGING', true),
   ],

The second set of changes...

For context, directorytree/ldaprecord-laravel had a PHP trait named LdapRecord\Laravel\Auth\MultiDomainAuthentication which was deprecated before version 2.7.3 and removed in version 3.0.0. The 1 SDS PHP source file app/Http/Controllers/Auth/LoginController.php used that trait.

This task updated LoginController to clone into itself the used portions of MultiDomainAuthentication, thus removing the external dependency of the former on the latter.

As the removed trait was itself a sub-trait of LdapRecord\Laravel\Auth\CreatesUserProvider, LoginController now composed that directly and not indirectly.

Also LoginController gained the new protected function getLdapGuard cloned from the removed trait.

Note that the function getLdapGuardFromRequest was not cloned from the removed trait since LoginController already had its own version that overrode it.

The third set of changes...

For context, each of these 5 SDS classes composed one of the 2 classes LdapRecord\Models\Model or LdapRecord\Models\Scope:

  • app/Ldap/ExternalUser.php
  • app/Ldap/Scopes/OnlyStaffUsers.php
  • app/Ldap/Scopes/OnlyStudents.php
  • app/Ldap/SmusStudents.php
  • app/Ldap/SmusUser.php

For each of those 5, it was updated such that for any property or method it contained which overrode or implemented a same-named one from Model or Scope, that property or method had explicit type annotations added to match the originals. This was required for the SDS code to satisfy a PHP or Laravel stricture so the code runs.

RETURN

2024 Aug 2: SDS Laravel: Upgrade intervention/image from 2.x to 3.x and Fix Import Photos Screen

This task made 3 distinct groups of changes to SDS Laravel all related to image handling and the Admin Menu -> Migration -> Import Photos screen.

Changes For intervention/image Upgrade

This task also made 2 distinct sets of changes to be compatible with the intervention/image upgrade.

For context, as part of the major update of intervention/image from version 2.x to 3.x, it was also split up into multiple libraries, with the core remaining under the old name and the optional Laravel-specific add-ons being in the new intervention/image-laravel library. We only need to directly require the latter, which brings in the former.

See also for context:

The first set of changes...

This task added the 1 PHP config source file config/image.php that is standard per the above-linked README.

The second set of changes...

This task replaced the use of the no longer existing PHP class Intervention\Image\ImageServiceProvider with the existing Intervention\Image\Laravel\Facades\Image.

In addition, the class is now used directly by its full name in the 2 PHP source files where the image handling functionality is actually used, which are app/Models/File.php and app/Http/Controllers/Admin/Migration/ImportPhotosController.php. This task has purged any explicit registration in config/app.php of any class of this library in providers or aliases; the Image alias is now purged.

The PHP source code that directly used the Intervention Image libraries was updated for compatibility with several method renames or substitutions.

Changes were made to reflect the rename of the library static method make to read.

Changes were made to reflect the changes with the set of image manipulation methods for image resizing. The resize method of version 2 had taken a third argument by which one specified the resize should keep the original aspect ratio. The resize method of version 3 doesn't take that argument, and instead the version 3 method scale is explicitly for resizing that maintains the aspect ratio.

RETURN

Bug/Security Fixes In Photos Import

This task also updated the 2 files app/Http/Controllers/Admin/Migration/ImportPhotosController.php plus resources/views/admin/migration/import_photos/index.blade.php to fix bugs tangential to the image file path handling.

One fix was in the index function so that the directory path reported to the user was the same one that store actually used, from Storage::disk('photos'), rather than being either of a couple of hard-coded alternatives toggled on an environment variable that isn't actually used.

A second fix was to stop passing a path hidden form field value from index to store, whose value was set from the incorrect path in index, and which store validated as present and subsequently didn't use. Besides being dead code from the disuse, if this had been used it would have been a major security vulnerability, as it would have let web clients directly control actions against server local file system paths.

A third fix was in the store such that logic to skip over files that were not of type image/jpeg was faulty and never skipped any kind of file, and now it skips those it is supposed to. The specific logic error was one of operator precedence, where it was of the form !x == y where it meant to be !(x == y); the fixed version made it be x !== y.

RETURN

Changes For league/flysystem Indirect Upgrade

This task also updated app/Http/Controllers/Admin/Migration/ImportPhotosController.php to be compatible with the league/flysystem upgrade from version 1 to version 3 that was indirect by way of the Laravel 9 upgrade.

See also for context:

There were 3 substitutions for compatibility in the store function.

There was 1 like this:

   Storage::disk('photos')->getMimeType($file)
   Storage::disk('photos')->mimeType($file)

And there were 2 like this:

   Storage::disk('photos')->getDriver()->getAdapter()->applyPathPrefix($file);
   Storage::disk('photos')->path($file);

This task also updated config/filesystems.php to add the element 'throw' => true to each disks element. This is to help preserve Laravel 8 behavior that we might have been relying on when a write operation fails. For example, write operations such as write by default no longer throw an exception when a write operation fails and instead returns false; until we can audit our code for how we currently check for such failures, this config change should make it less likely our code was broken by the dependency update.

Note that there are still other kinds of default behavior changes that we should audit our code for handling of. For example, write operations will overwrite existing files by default, so if we don't want that then we should explicitly be checking for their existence first. Also, reading a nonexisting file now returns null rather than an exception.

RETURN

2024 Aug 2: SDS Laravel: Upgrade laravel/framework from 8.x to 10.x and Upgrade Co-Dependent PHP deps

This task made 5 distinct groups of changes to SDS Laravel all related to upgrading it from Laravel 8 to Laravel 10.

Summary of PHP Dependency Upgrades or Removals

This task updated composer.json to require the latest PHP-8.1-compatible versions of laravel/framework as well as of all of its co-dependent PHP library dependencies. In some cases, upgrading co-dependents meant removing them entirely in favor of Laravel built-ins.

To be specific, it made these dependency changes:

  • barryvdh/laravel-debugbar (^3.7 to ^3.13.5)
  • directorytree/ldaprecord-laravel (^3.3.3 unchanged)
  • etern8ty/beanstream (dev-master unchanged and is custom fork)
  • fakerphp/faker (^1.23.1 unchanged)
  • fideloper/proxy (^4.4.2 removed as laravel/framework has its upgrade built-in)
  • goldspecdigital/laravel-eloquent-uuid (^8.0.1 removed as laravel/framework has its upgrade built-in)
  • guzzlehttp/guzzle (^7.9.2 unchanged)
  • intervention/image-laravel (^1.3 unchanged)
  • juliomotol/laravel-auth-timeout (^3.1.1 to ^4.1)
  • lab404/laravel-impersonate (^1.7.5 unchanged)
  • laravel/framework (^8.83.27 to ^10.48.18)
  • laravel/tinker (^2.9 unchanged)
  • laravel/ui (^3.4.6 to 4.5.2; but it recommends using Laravel Breeze or Laravel Jetstream instead)
  • mockery/mockery (^1.6.12 unchanged)
  • nunomaduro/collision (^5.11 to ^7.10)
  • phpunit/phpunit (^10.5.29 unchanged)
  • spatie/laravel-ignition (^1.7 to ^2.8)
  • staudenmeir/eloquent-has-many-deep (^1.14.4 to ^1.19.4)

RETURN

Specify Minimum PHP Dependency Versions Used Are Stable

This task updated composer.json to specify an app default minimum-stability of stable rather than dev.

This is per the Laravel 10 upgrading recommendation:

While the setting could instead have been removed as stable is its default, being explicit seemed better here.

Note that this didn't cause any changes to what PHP library versions are actually installed by Composer versus the old setting, but is a good constraint going forward.

RETURN

Changes For laravel/framework Upgrade's Removal of "dates" Model Attribute

This task also updated 51 PHP source files to be compatible with a breaking change made by Laravel itself with version 10.

Laravel supported a "dates" model attribute through version 9, and then Laravel 10 removed it. The function of this was to enumerate database/model fields that were supposed to be automatically converted to Carbon DateTime objects; so under Laravel 8, any "dates" declarations would be respected, while under Laravel 10 they would be ignored.

See also:

Compare:

As a result, simply upgrading SDS Laravel from Laravel 8 to 10 resulted in many parts of the app breaking in various ways including when simply visiting the post-login home screen, as PHP died with errors like Call to a member function format() on int.

To fix this, any instances of protected $dates = ['x',...] in model classes were replaced with protected $casts = ['x'=>'datetime',...] which was the more modern way to get the same functionality, which exists in both Laravel 8 and 10. For the few model classes that already had other $casts declarations, the replacements were merged with those.

While the "dates" change could have been its own task that was merged to trunk prior to and separately from the current Laravel 10 upgrade task, it was combined with the latter to streamline testing, as both had potential impacts over a large fraction of the app.

RETURN

Changes For laravel/framework Upgrade's Miscellaneous Changes

This task also updated app/Providers/AuthServiceProvider.php to remove the call $this->registerPolicies(); from boot()because Laravel 10 now calls it automatically, as recommended here:

RETURN

Changes For goldspecdigital/laravel-eloquent-uuid Removal

This task also updated these 5 PHP source files to be compatible with the replacement of goldspecdigital/laravel-eloquent-uuid with a Laravel built-in:

  • app/Models/Application/Application.php
  • app/Models/User.php
  • app/Models/User/Student.php
  • app/Models/User/Teacher.php
  • app/Models/User/UserContract.php

These further 3 files also referenced the trait but commented out, so not current users but possible past or future users:

  • app/Models/Application/AppUser.php
  • app/Models/User/Address.php
  • app/Models/User/Guardian.php

For each of the above 8 files, there were these 2 line subsitutions:

   use GoldSpecDigital\LaravelEloquentUUID\Database\Eloquent\Uuid;
   use Uuid;
   use Illuminate\Database\Eloquent\Concerns\HasUuids;
   use HasUuids;

Here is a description of the above built-in feature in Laravel 9.3+:

https://laravel.com/docs/11.x/eloquent#uuid-and-ulid-keys

The purpose of that reimplemented functionality was to empower use of generated UUIDs for primary key fields of some database tables instead of the serially generated integers that SDS Laravel more typically uses; Laravel Eloquent only gained built-in support for UUIDs with version 9.3.

This task also deleted the single PHP file app/Models/Traits/Uuids.php as it appeared to be unused.

RETURN

Changes For fideloper/proxy Removal

This task also updated app/Http/Middleware/TrustProxies.php to be compatible with the replacement of fideloper/proxy with a Laravel built-in. The changes were in 2 spots.

First was this substitution:

   use Fideloper\Proxy\TrustProxies as Middleware;
   use Illuminate\Http\Middleware\TrustProxies as Middleware;

Second was this substitution:

   protected $headers = Request::HEADER_X_FORWARDED_ALL;
   protected $headers =
       Request::HEADER_X_FORWARDED_FOR |
       Request::HEADER_X_FORWARDED_HOST |
       Request::HEADER_X_FORWARDED_PORT |
       Request::HEADER_X_FORWARDED_PROTO |
       Request::HEADER_X_FORWARDED_AWS_ELB;

See also https://laravel.com/docs/9.x/upgrade under "Trusted Proxies".

RETURN

Changes For juliomotol/laravel-auth-timeout Upgrade

This task also updated app/Http/Middleware/AuthTimeoutMiddleware.php to be compatible with the juliomotol/laravel-auth-timeout upgrade.

There was this 1 substitution:

   use JulioMotol\AuthTimeout\Middleware\AuthTimeoutMiddleware as BaseMiddleware;
   use JulioMotol\AuthTimeout\Middlewares\CheckAuthTimeout as BaseMiddleware;

Note that juliomotol/laravel-auth-timeout must be upgraded simultaneously with Laravel since the former's versions 3.1.1 and 4.1 respectively require Laravel 8 and 10 respectively.

See https://github.com/juliomotol/laravel-auth-timeout/blob/master/CHANGELOG.md for more change details and upgrade notes on that.

RETURN

Documentation Changes

2024 Jul 16: SDS Laravel: Make README etc Say This App is SDS Rather Than Laravel

This task updated SDS Laravel composer.json to update primary documentation attributes so they describe the actual SDS application this is rather than the Laravel template used to build it.

There were these substitutions:

   "name": "laravel/laravel",
   "description": "The Laravel Framework.",
   "keywords": ["framework","laravel"],
   "name": "smus/sds",
   "description": "School Data System (SDS)",
   "keywords": [],

This task also completely rewrote README.md so it describes this SDS application rather than the Laravel framework.

RETURN