We've been ignoring what the Bullet gem reports for at least 6 years
(maybe more), but we were still updating the gem and maintaining the
code in `config/environments/` (which caused conflicts every time we run
`rails app:update` to upgrade to a new Rails version). Maintaining it
isn't a huge effort, but it's infinitely bigger than the benefits we get
from it, which are zero.
Adding `includes` everywhere we query for records would be a huge
maintenance effort and would make the code less readable, so I don't
think it's worth it. We might do it occasionally if we detect a
performance bottleneck.
We could also use a gem to automatically avoid the N+1 queries problem,
like Goldiloader [1], ArLazyPreload [2] or JitPreload [3]. Benchmarks
show that the performance improvements obtained by using these gems is
about less than 10% (it depends a lot on the page being loaded, though),
which IMHO doesn't justify adding yet another gem that patches
ActiveRecord and that could be incompatible with other gems doing so.
There are a couple of open pull requests (at the time of writing,
they've been open for about two years) in the Rails repository [4][5] to
automatically avoid N+1 queries as well. For now, we'll hope something
similar is integrated in Rails itself in the future.
[1] https://github.com/salsify/goldiloader
[2] https://github.com/DmitryTsepelev/ar_lazy_preload
[3] https://github.com/clio/jit_preloader/
[4] Pull request 45231 in https://github.com/rails/rails/
[5] Pull request 45413 in https://github.com/rails/rails/
The config.file_watcher option still exists but it's no longer included
in the default environtment file. Since we don't use it, we're removing
it.
The config.assets.assets.debug option is no longer true by default [1],
so it isn't included anymore.
The config.active_support.deprecation option is now omitted on
production in favor of config.active_support.report_deprecations, which
is false by default. I think it's OK to keep it this way, since we check
deprecations in the development and test environments but never on
production environments.
As mentioned in the Rails upgrade guide, sprockets-rails is no longer a
rails dependency and we need to explicitly include it in our Gemfile.
The behavior of queries trying to find an invalid enum value has changed
[2], so we're updating the tests accordingly.
The `favicon_link_tag` method has removed the deprecated `shortcut`
link type [3], so we're updating the tests accordingly.
The method `raw_filter` in ActiveSupport callbacks has been renamed to
`filter` [4], so we're updating the code accordingly.
[1] https://github.com/rails/rails/commit/adec7e7ba87e3
[2] https://github.com/rails/rails/commit/b68f0954
[3] Pull request 43850 in https://github.com/rails/rails
[4] Pull request 41598 in https://github.com/rails/rails
The `database:` argument from the `connected_to` method was deprecated
in Rails 6.1 [1] and removed in Rails 7.0 [2]. The ros-apartement gem
has already introduced fixes for this issue [3][4], but there has been
no release including these fixes.
So we're applying the fix in our code.
Note that, since Apartment already overwrites the `connected_to` method
by creating a `connected_to_with_tenant` method, we're calling the
`connected_to_without_tenant` method inside the patch, which is
equivalent to ActiveRecord's original `connected_to`.
[1] Pull request 37874 in https://github.com/rails/rails
[2] Pull request 40530 in https://github.com/rails/rails/pull
[3] Pull request 194 in https://github.com/rails-on-services/apartment
[4] Pull request 243 in https://github.com/rails-on-services/apartment
Byebug hasn't been maintained for years, and it isn't fully compatible
with Zeitwerk [1]. On the other hand, Ruby includes the debug gem since
version 3.1.0. We tried to start using at after commit e74eff217, but
couldn't do so because our CI was hanging forever in a test related to
machine learning, with the message:
> DEBUGGER: Attaching after process X fork to child process Y
(Note this message appeared with debug 1.6.3 but not with the version
we're currently using.)
So we're changing the debug gem fork mode in the test so it doesn't hang
anymore when running our CI. We tried to change the test so it wouldn't
call `Process.fork`, but this required changing the code, and since
there are no tests checking machine learning behavior with real scripts,
we aren't sure whether these script would keep working after changing
the code.
[1] Issue 564 in https://github.com/deivid-rodriguez/byebug
The initialjs-rails gem hasn't been maintained for years, and it
currently requires `railties < 7.0`, meaning we can't upgrade to Rails 7
while we depend on it.
Since the code in the gem is simple, and we were already rewriting its
most complex part (generating a background color), we can implement the
same code, only we're using Ruby instead of JavaScript. This way, the
avatars will be shown on browsers without JavaScript as well. Since
we're adding a component test that checks SVG images are displayed even
without JavaScript, we no longer need the test that checked images were
displayed after AJAX requests.
Now the tests show the user experience better; people don't care about
the internal name used to select the initial (which is what we were
checking); they care about the initial actually displayed.
Note initialjs generated an <img> tag using a `src="data:image/svg+xml;`
attribute. We're generating an <svg> tag instead, because it's easier.
For this reason, we need to change the code slightly, giving the <svg>
tag the `img` role and using `aria-label` so its contents won't be read
aloud by screen readers. We could give it a `presentation` role instead
and forget about `aria-label`, but then screen readers would read the
text anyway (or, at least, some of them would).
SassC/Libsass has been deprecated for years and has been replaced by
Dart Sass. However, the dartsass-rails gem, maintained by the Rails
team, doesn't support sprockets integration and doesn't allow glob
imports (using `@import something/**/*` or similar). In particular,
dartsass-rails needs to start a separate browser that makes it less
straightforward to change a file and reload the browser.
So we're using sassc-embedded, which provides Dart Sass integration with
sprockets. While there's no guarantee this gem will be maintained a few
years from now, we know for sure that SassC/Libsass won't be maintained
at all, so using sassc-embedded is an improvement over our current
situation.
On my machine, this change reduces compilation times by about 35%.
Note we still depend on the `sassc-rails` gem, for two reasons.
First, we're still importing CSS/Sass content from a couple of gems
(mainly, social-share-button and font-awesome) and we don't know how to
import this content without the `sassc-rails` gem.
And, second, it provides support for glob imports. Without it, we'd have
to manually add every single (S)CSS file we import to the
`application.scss` file instead of being able to write things like
`@import admin/**/*";`.
Note we're removing the `sass` gem from `Gemfile.lock`. We should have
done it as part of e210682ac, but when we developed that branch, it
didn't contain the changes where we removed another gem depending on the
`sass` gem (which we removed in commit 2fa713c64), so Bundler didn't
delete it. However, now that we're changing the Gemfile, Bundler is
finally removing the no-longer-needed `sass` gem and its dependencies.
SCSS Lint was based on Ruby Sass, which has been deprecated since 2018
and doesn't support some of the latest features in Dart Sass.
Since we're going to migrate to Dart Sass, we're also migrating to
Stylelint. In order to provide all the funcionality SCSS Lint had, we
also need to install a few Stylelint plugins. We also need to install
the `postcss-scss` package so Stylelint can read SCSS files.
We're still getting the linting errors we used to get in
`legislation_process.scss` because of the `max-nesting-depth` and
`selector-max-compound-selectors` rules, as well as the warnings we used
to get in `layout.scss` because the `ssb-whatsapp_app` HTML class
contains undescores in their name.
We're also getting a couple of new linting errors, which could be false
positives:
* The `scss/no-unused-private-members` rule reports errors in
`_consul_settings.scss` because of the `$-zf-*` variables, but I'm not
too worried about this one because these lines won't be necessary after
updating Foundation.
* The `scss/selector-no-redundant-nesting-selector` reports an error in
`datepicker_overrides.scss`, but removing the unnecessary nesting
selector would make it harder to understand the code (assuming it'd
work in the first place).
We've changed some files due to few differences between SCSS Lint rules
and their Stylelint equivalents:
* The `@stylistic/string-quotes` rule detects a case in `admin.scss`
that StringQuotes didn't detect.
* The `function-url-quotes` rule detects a case in `mixins/icons.scss`
that UrlQuotes didn't detect.
* The `@stylistic/declaration-bang-space-before` rule detects a case in
`sdg/goals/show.scss` that BangFormat didn't detect.
There are also a couple of rules that don't behave exactly like they
used to:
* The equivalents of SpaceBetweenParens and SpaceAfterComma don't cover
parenthesis or commas in mixin parameters; we haven't found rules that
detect these cases.
* DisableLinterReason probably has an equivalent that behaves
differently but, since we never disable linters inline, we aren't
adding its equivalent rule.
Note we're removing the SpaceAfterVariableColon rule because its
equivalent, `scss/dollar-variable-colon-space-after`, reports cases
where we add spaces to indent several variable assignments (which we do
a lot in the `_consul_settings.scss` file). We might add this rule again
if we stop aligning consecutive assignments.
We're also removing the QualifyingElement rule because its equivalent,
`selector-no-qualifying-type: true`, behaves differently. For example,
in this code:
```
a.qualifying {
color: inherit;
}
p {
&.qualifying {
color: inherit;
}
}
```
With the QualifyingElement from SCSS Lint, the first rule is incorrect
but the second one is correct. With the selector-no-qualifying-type rule
from Stylelint, on the other hand, both rules are incorrect.
Personally, I never liked the QualifyingElement rule, and we were
working around it anyway, so we aren't applying
selector-no-qualifying-type.
For reference, here's a full list of the SCSS Lint rules we had enabled
and their Stylelint equivalents.
BangFormat
"@stylistic/declaration-bang-space-after": "never"
"@stylistic/declaration-bang-space-before": "always"
BorderZero
declaration-property-value-disallowed-list:
border:
- none
ColorKeyword
color-named: "never"
DebugStatement
at-rule-disallowed-list:
- debug
DeclarationOrder
order/order:
- dollar-variables
- custom-properties
- type: at-rule
name: extend
- type: at-rule
name: include
hasBlock: false
- declarations
- type: at-rule
name: include
hasBlock: true
- rules
ElsePlacement # Apparently replaced by the combination of:
scss/at-else-closing-brace-space-after: always-intermediate
scss/at-else-empty-line-before: never
scss/at-if-closing-brace-space-after: always-intermediate
scss/at-if-closing-brace-newline-after: always-last-in-chain
EmptyLineBetweenBlocks:
ignore_single_line_blocks: true
rule-empty-line-before:
- "always-multi-line"
- ignore:
- after-comment
- first-nested
EmptyRule:
block-no-empty: true
FinalNewline
"@stylistic/no-missing-end-of-source-newline": true
HexLength
color-hex-length: "short"
HexNotation
@stylistic/color-hex-case: "lower"
HexValidation
color-no-invalid-hex: true
IdSelector
selector-max-id: 0
ImportPath
scss/load-no-partial-leading-underscore: true
scss/at-import-partial-extension: never
Indentation
"@stylistic/indentation": 2
LeadingZero
@stylistic/number-leading-zero: "always"
NameFormat
scss/at-function-pattern: "^(-?[a-z][a-z0-9]*)(-[a-z0-9]+)*$"
scss/at-mixin-pattern: "^(-?[a-z][a-z0-9]*)(-[a-z0-9]+)*$"
scss/dollar-variable-pattern: "^(-?[a-z][a-z0-9]*)(-[a-z0-9]+)*$"
scss/percent-placeholder-pattern: "^(-?[a-z][a-z0-9]*)(-[a-z0-9]+)*$"
NestingDepth
max-nesting-depth: 4
PlaceholderInExtend
scss/at-extend-no-missing-placeholder: true
PrivateNamingConvention
scss/no-unused-private-members: true
PropertySpelling
property-no-unknown: true
PseudoElement
selector-pseudo-element-colon-notation: "double"
selector-pseudo-element-no-unknown: true
SelectorDepth
selector-max-compound-selectors: 5
SelectorFormat # Not always followed; ssb-whatsapp_app
custom-property-pattern: "^([a-z][a-z0-9]*)(-[a-z0-9]+)*$"
selector-class-pattern: "^([a-z][a-z0-9]*)(-[a-z0-9]+)*$"
Shorthand
shorthand-property-no-redundant-values: true
SingleLinePerProperty
"@stylistic/declaration-block-semicolon-newline-after": always-multi-line
SingleLinePerSelector
@stylistic/selector-list-comma-newline-after": always
SpaceAfterComma
"@stylistic/value-list-comma-space-after": always
SpaceAfterPropertyColon
"@stylistic/declaration-colon-space-after": always-single-line
SpaceAfterPropertyName
"@stylistic/declaration-colon-space-before": never
SpaceAfterVariableName
scss/dollar-variable-colon-space-before: never
SpaceAroundOperator
scss/operator-no-unspaced: true
SpaceBeforeBrace
@stylistic/block-opening-brace-space-before: always
SpaceBetweenParens
"@stylistic/function-parentheses-space-inside": never
"@stylistic/selector-attribute-brackets-space-inside": never
"@stylistic/selector-pseudo-class-parentheses-space-inside": never
"@stylistic/media-feature-parentheses-space-inside": never
StringQuotes
"@stylistic/string-quotes": double
TrailingSemicolon
"@stylistic/declaration-block-trailing-semicolon": always
TrailingWhitespace # Note it was enabled by the gem and not by us
"@stylistic/no-eol-whitespace": true
TrailingZero
"@stylistic/number-no-trailing-zeros": true
UnnecessaryMantissa
"@stylistic/number-no-trailing-zeros": true
UnnecessaryParentReference
scss/selector-no-redundant-nesting-selector: true
UrlQuotes
function-url-quotes: always
VendorPrefixes
value-no-vendor-prefix: true
selector-no-vendor-prefix: true
property-no-vendor-prefix: true
at-rule-no-vendor-prefix: true
media-feature-name-no-vendor-prefix: true
ZeroUnit:
length-zero-no-unit: true
Although the gem is called jquery-fileupload-rails, the node package is
called blueimp-file-upload.
Note we're using the same version as provided by the gem.
The jquery-fileupload gem provided a `basic.js` file (which we were
requiring), which had the following content:
```
//= require jquery-fileupload/vendor/jquery.ui.widget
//= require jquery-fileupload/jquery.iframe-transport
//= require jquery-fileupload/jquery.fileupload
``
This file isn't available in the Node.js package, so we're adapting its
contents in our application.js file. Since we're already requiring
jQuery UI widget, we're omitting that line.
We're upgrading to this version in order because it's the first version
that works with our current version of execjs when running the
`autoprefixer:info` task. Since it's been such a long time since the
last time we upgraded autoprefixer-rails (version 8.2.0, which is the
version we were using, was released on March 2018), and we're getting a
warning when compiling the assets with the latest version, we're going
to do it gradually; version 10.2.5.1 was released on May 2021, so it
also seems like a reasonable middlepoint.
Note we're moving the `browserlist` file to `.browserlistrc` because
that's the expected locations in new versions of autoprefixer-rails.
Also note that we're changing this file so we support browsers released
in the last 7 years, so we still support more or less the same browsers
we supported before this change (about 98% of the browsers out there).
We might reduce it to 5 years in the future.
Bumps [autoprefixer-rails](https://github.com/ai/autoprefixer-rails) from 8.2.0 to 10.2.5.1
- [Changelog](https://github.com/ai/autoprefixer-rails/blob/master/CHANGELOG.md)
- [Commits](https://github.com/ai/autoprefixer-rails/compare/8.2.0...10.2.5.1
Since Foundation hasn't released a new gem for years, we haven't been
able to upgrade Foundation to its most recent version.
Thanks to this change, we'll be able to do so.
Note we're using motion-ui version 2.0.3 because version 2.0.5 (the
latest at the moment) requires Dart Sass.
