This test fails very often in CI. This should hopefully improve it.
With a downclocked CPU I've got 2 errors in 10 tries. After adding
```
await page.waitForLoadState();
```
I got no errors in 20 tries. But my testing methodology is unreliable so there's no guarantee that this helps:
* `Locator.click()` documentation says that it already waits by default for caused navigation to finish
* I added some `date` time measurement and `waitForLoadState` was clocking in about 5-15 ms, which shouldn't be causing the issue for `expect` that times out in 3000ms
...but e2e issues were `element(s) not found` on timeout, not wrong state, so my assumption is that somehow it is asserting the wrong snapshot of a page that isn't fully loaded
Additionally, I've converted _revertion_ logic to _make consistent before testing_, which is better at handling test retries in case that the _revertion_ part of the code was never ran in previous failure. In such case test retry is guaranteed to fail as well, at least when testing locally, not sure about CI.
Additionally, a completely unrelated comment fix in Playwright config.
Reviewed-on: https://codeberg.org/forgejo/forgejo/pulls/12871
Reviewed-by: Gusted <gusted@noreply.codeberg.org>
* remove two unused strings I identified while doing other things
* update two strings per request of @mahlzahn while avoiding a whole separate PR for this
* move 126 strings to JSON, some are remapped with a better structure
* previous migration: forgejo/forgejo!12280
Reviewed-on: https://codeberg.org/forgejo/forgejo/pulls/12718
Reviewed-by: Robert Wolff <mahlzahn@posteo.de>
Example for how that would look like on code.forgejo.org:
Login screen examples:
|||
|-|-|
Many instances would benefit from being able to communicate such information before a user signs up or logs in after a long break.
## Testing
Disabled unless admin explicitly decides to create a template override, nothing to test.
In theory CSS could break but IDK how to test against that given the custom template requirement.
Also need to know that template customizations are provided as-is and shift all accountability of maintaining them to admins.
## Documentation
https://codeberg.org/forgejo/docs/pulls/1945
Reviewed-on: https://codeberg.org/forgejo/forgejo/pulls/12597
Reviewed-by: Robert Wolff <mahlzahn@posteo.de>
Reviewed-by: Beowulf <beowulf@beocode.eu>
Reviewed-by: Gusted <gusted@noreply.codeberg.org>
Previous similar PR: https://codeberg.org/forgejo/forgejo/pulls/11879.
Moved strings from INI to JSON. Some directly, some with keys updated to be consistent. The latter was done carefully, making sure all usages are updated, and was tested locally.
There are more deletions than insertions because some languages also had some extra empty lines removed.
Reviewed-on: https://codeberg.org/forgejo/forgejo/pulls/12280
Reviewed-by: Robert Wolff <mahlzahn@posteo.de>
Followup to https://codeberg.org/forgejo/forgejo/pulls/6203
Currently it is logging an error wherever a template is rendered in language that doesn't have all plural strings covered. For example, Esperanto isn't well maintained.
Since more plural strings were migrated in v15 to new format, these errors became much more common. However, for all languages but the base one (English) they are completely harmless and just indicate an incomplete translation.
However, for base (English) they indicate a bug in either template or en-US.json, which should be still logged as an error.
The error is being logged by `LookupPluralByForm`, which is called by `TrPluralStringAllForms` and (`TrPluralString` through `LookupPluralByCount`). I originally intended to just pass log func directly to `LookupPluralByForm` from both, but since `TrPluralString` isn't calling `LookupPluralByForm` directly, it didn't look clean, so I went with passing a flag around instead and implemented logging logic in `LookupPluralByForm` itself.
I little concern is with that the so-called "default lang" is configurable, and if it is configured to something with less than 100% completion, it will cause fallback bugs, as well as a lot of logging of this as an error. But this is why changing "default lang" is a bad idea in the first place, and broken fallbacks should be greater concern than junk in the logs.
Reviewed-on: https://codeberg.org/forgejo/forgejo/pulls/12183
Reviewed-by: Beowulf <beowulf@beocode.eu>
Co-authored-by: 0ko <0ko@noreply.codeberg.org>
Co-committed-by: 0ko <0ko@noreply.codeberg.org>
* migrate 17 strings related to repository migrations from INI to JSON
* changed templates to get rid of unhelpful prefix `repo`
* migrate 4 strings related to counters
* also changed templates to get rid of `repo`, but it had to be done anyway to use `TrPluralString`
* Unhardcode the header on migraiton type selector page, which I haven't noticed in https://codeberg.org/forgejo/forgejo/pulls/6795 or in two other PRs I did on this template since
Reviewed-on: https://codeberg.org/forgejo/forgejo/pulls/11879
Reviewed-by: Gusted <gusted@noreply.codeberg.org>
Reviewed-by: Beowulf <beowulf@beocode.eu>
Our package registries types have many things in common, so they have many shared strings like
```json
"packages.installation": "Installation",
```
but not all and not consistently. For example, some packages duplicate these strings:
```json
"packages.*.install": "To install the package, run the following command:",
"packages.*.registry": "Setup this registry from the command line:",
"packages.*.repository": "Repository info",
"packages.*.dependencies": "Dependencies",
```
while other packages have them customized:
```json
"packages.cargo.install": "To install the package using Cargo, run the following command:",
"packages.composer.install": "To install the package using Composer, run the following command:",
"packages.conan.install": "To install the package using Conan, run the following command:",
"packages.conda.install": "To install the package using Conda, run the following command:",
```
While it's ok for them to have custom strings, I don't see need to duplicate the same generic strings. Even though Weblate provides tools to simplify maintenance of duplicated strings, our files still end up being bloated.
In this PR I deduplicated the strings listed above under a `common` key. There already was a generic shared string for Dependencies, that I just re-used.
Also fixed strings in Arch packages, which were saying Depends with meaning of Dependencies.
Reviewed-on: https://codeberg.org/forgejo/forgejo/pulls/11345
Reviewed-by: Gusted <gusted@noreply.codeberg.org>
Followup to #10702 where two new strings were hardcoded in the template.
## Changes
- added as translatable strings
- made the template use a variable to avoid making it a long unreadable
- an unrelated change where tests related to this area of the template are moved to a separate file
- they were created in Forgejo: c92b4b12c8, 192177fc88
## Testing
Testing translations of all individual strings is considered not needed.
Added test for tmpl logic which was missing and that I've touched.
Reviewed-on: https://codeberg.org/forgejo/forgejo/pulls/11333
Reviewed-by: Luis Adame <luisadame@noreply.codeberg.org>
Reviewed-by: Gusted <gusted@noreply.codeberg.org>
Replaces https://codeberg.org/forgejo/forgejo/pulls/11271
This rule was problematic and didn't quite make sense:
* the intention is to have a navbar that is 48px+1px border
* we apply `min-height:49px` to: `#navbar` with intention of making it 49px tall total, with border
* but we also apply it to `#navbar .navbar-left/right`
* elements `#navbar .navbar-left/right` become 49px tall
* their parent `#navbar` becomes 49px+1px border = 50px, overriding what was done in step 2
This height missmatch had an incompatibility with this definition for open menu inside of navbar on mobile, causing jitter when the menu is opened/closed, because it was expecting navbar body to be 48px, but due to the rule that was removed it was actually 49px.
```css
#navbar.navbar-menu-open .navbar-left .navbar-mobile-right {
min-height: 48px;
}
```
The fix is to only apply `min-height` to `#navbar`'s children. At least `.navbar-left` is always expected to be present unconditionally on all pages. This does make the navbar 1px shorter. If we still want a 50px navbar but without the bug let me know and I'll adjust.
### Preview by @panc
From https://codeberg.org/forgejo/forgejo/pulls/11271, also applies to this PR.
| Before | After |
|----|----|
| <video src="/attachments/515fdfc1-cb97-46af-88ac-9bb0e216a996" title="Screencast From 2026-02-13 19-56-32" controls></video> | <video src="/attachments/133d1499-4660-402d-8082-f407b3644e5c" title="Screencast From 2026-02-13 19-57-29" controls></video> |
Reported-by: panc <panc@noreply.codeberg.org>
Reviewed-on: https://codeberg.org/forgejo/forgejo/pulls/11274
Reviewed-by: Robert Wolff <mahlzahn@posteo.de>
Replaces #10789
Convert the layout from flex-list helpers to a CSS-native grid. This allows to having buttons in different rows aligned to each other while keeping the layout responsive, i.e. looking good on both desktop and mobile.
### Preview (desktop)
|Before|After|
|-|-|
|||
### Preview (mobile)
|Before|After|
|-|-|
|||
### Preview (Guest)
|Before|After|
|-|-|
|||
## Testing
Automated tests were added to make sure that actions in this list are still working, and for basic template logic. No tests were added for layout because layout being correct is an abstract concept that is difficult to explain to Playwright.
Reviewed-on: https://codeberg.org/forgejo/forgejo/pulls/11127
Reviewed-by: Antonin Delpeuch <wetneb@noreply.codeberg.org>
## Changes
I've made releases list more usable on narrow viewports while trying keep the layout unchanged for desktop viewports.
To support these changes large amount of Tailwind classes were converted to regular CSS to be applied conditionally via `@media`. While it was possible to just adjust the Tailwind classes to achieve the same behavior, there's a positive effect which is that the repeating HTML of releases generated by template's range is much less verbose and contains fewer long duplicated lines of Tailwind classes.
## Preview
### Desktop
Not much changed, but the dot between tag and release name is no more.
|Before|After|
|-|-|
|||
### Mobile
|Before|After|
|-|-|
|||
Reviewed-on: https://codeberg.org/forgejo/forgejo/pulls/11080
Reviewed-by: Gusted <gusted@noreply.codeberg.org>
Reviewed-by: Beowulf <beowulf@beocode.eu>
Closes#10534
The primary code change is that `TestMissingTranslationHandling` was converted to a unit test.
However translation unit tests were a mix of mostly unrelated old (INI) and new (JSON) code with unclear licensing. It was cleaned up with help of Git history.
Reviewed-on: https://codeberg.org/forgejo/forgejo/pulls/10539
Reviewed-by: Gusted <gusted@noreply.codeberg.org>
Followup to https://codeberg.org/forgejo/forgejo/pulls/9359. `templates/repo/issue/list.tmpl` had buttons changed which had `.disabled` on them conditionally.
* adjust devtest page to have such buttons
* implement `.disabled` for the newer buttons that does the same thing as the `.ui` buttons do: apply custom opacity, disable pointer events
* this is the most boring way of implementing this
Reported-by: @Gusted
Reviewed-on: https://codeberg.org/forgejo/forgejo/pulls/10410
Reviewed-by: Gusted <gusted@noreply.codeberg.org>
* in both `/admin` and `/admin/cron`: use new buttons, they are slightly more compact, which i think fits this place well as currently rows here feel too tall
* in `/admin/cron`: use `octicon-play` consistently with `/admin` instead of `octicon-triangle-right`
* in `/admin`: replace verbose template HTML with Range-based generator
* added integration test to verify page content
<!--start release-notes-assistant-->
## Release notes
<!--URL:https://codeberg.org/forgejo/forgejo-->
- User Interface features
- [PR](https://codeberg.org/forgejo/forgejo/pulls/10270): <!--number 10270 --><!--line 0 --><!--description ZmVhdCh1aSk6IGltcHJvdmUgYWRtaW4gZGFzaGJvYXJkIGNyb24gbGlzdA==-->feat(ui): improve admin dashboard cron list<!--description-->
<!--end release-notes-assistant-->
Reviewed-on: https://codeberg.org/forgejo/forgejo/pulls/10270
Reviewed-by: Michael Kriese <michael.kriese@gmx.de>
Reviewed-by: Otto <otto@codeberg.org>
Co-authored-by: 0ko <0ko@noreply.codeberg.org>
Co-committed-by: 0ko <0ko@noreply.codeberg.org>
* show/hide resolved: got rid of classes `ui labeled button` - ignoring them overriding each other, they were contributing three useful CSS properties in total:
```css
padding: 0;
background: none;
font-weight: 500;
```
and were also contributing an undesired margin:
* also fixed inconsistent font size between two views
* additionally got rid of verbose fomantic classes since there was already a rule that could accept their properties
* got rid of unused class `add-code-comment-right`
* added testing for the hide/show/resolve/unresolved buttons, but not related to CSS, just thought it might be useful
<!--start release-notes-assistant-->
## Release notes
<!--URL:https://codeberg.org/forgejo/forgejo-->
- User Interface changes without a feature or bug label
- [PR](https://codeberg.org/forgejo/forgejo/pulls/10268): <!--number 10268 --><!--line 0 --><!--description Y2hvcmUodWkpOiBjbGVhbnVwIHJldmlld3MgY3NzLCBpbXByb3ZlIGNvbnNpc3RlbmN5-->chore(ui): cleanup reviews css, improve consistency<!--description-->
<!--end release-notes-assistant-->
Reviewed-on: https://codeberg.org/forgejo/forgejo/pulls/10268
Reviewed-by: Gusted <gusted@noreply.codeberg.org>
Reviewed-by: Otto <otto@codeberg.org>
Co-authored-by: 0ko <0ko@noreply.codeberg.org>
Co-committed-by: 0ko <0ko@noreply.codeberg.org>
Addition to the user menu in navbar:
* display a link to `/devtest` if the instance run mode is dev instead of prod
* I think this is useful because:
* devtest pages are in general useful in development, but are not easy to discover by newcomers
* if the instance uses incorrect run mode, this entry should annoy it's admin into fixing the issue
Improvements to `/devtest`:
* fix error pages links
* add a few headers and explanation for what this page is for
Reviewed-on: https://codeberg.org/forgejo/forgejo/pulls/10263
Reviewed-by: Earl Warren <earl-warren@noreply.codeberg.org>
Reviewed-by: Gusted <gusted@noreply.codeberg.org>
Co-authored-by: 0ko <0ko@noreply.codeberg.org>
Co-committed-by: 0ko <0ko@noreply.codeberg.org>
Converted 8 buttons, fixed a few bugs with markdown editor, added some testing for template logic. The fixes are not auto-testable within reasonable effort and the changes like button conversion have nothing specific to be tested for.
### Convert form buttons to new buttons
|Before|After|
|-|-|
|||
### Justify "Delete release" button away from other buttons
|Before||
|-|-|
|After||
|Before||
|-|-|
|After||
### Fix markdown editor on mobile
|Before|After|
|-|-|
|||
### Buttons preview on mobile
|Before|After|Before|After|
|-|-|-|-|
|||||
Reviewed-on: https://codeberg.org/forgejo/forgejo/pulls/9940
Reviewed-by: Gusted <gusted@noreply.codeberg.org>
Replaced dropdowns in the navbar with JS-less ones from https://codeberg.org/forgejo/forgejo/pulls/7906.
Also made some changes to the dropdown component:
* fixed variable name
* painted backgrounds (hover, focus) are now consistently applied to the actual interactive items (`<a>`, `<button>`), not to `<li>`. This is consistent with how backgrounds are conditionally applied to pre-selected (`.active`) items and is better, as it allows to place additional things to `<li>`...
* ...`<hr>` can now be placed in some `<li>` instead of requiring splitting into multiple `<ul>`. This is simpler in code and I am guessing this should be better for a11y as screen readers can cast one continuous list instead of multiple ones. But have no hard proof that this is actually better. My main motivation was to avoid ugly mistake-prone tmpl logic where unconditional `<ul>` was getting closed and reopened inside of a condition.
I should note that on mobile all items, including these dropdowns, are hidden in another dropdown, and it stays JS-dependand for now. So this PR only makes this part of the UI JS-less for desktop.
Reviewed-on: https://codeberg.org/forgejo/forgejo/pulls/10025
Reviewed-by: Robert Wolff <mahlzahn@posteo.de>
Reviewed-by: Gusted <gusted@noreply.codeberg.org>
Co-authored-by: 0ko <0ko@noreply.codeberg.org>
Co-committed-by: 0ko <0ko@noreply.codeberg.org>
`create-page-form` is currently only referenced in CSS files.
I've tried researching this with `git log -S 'create-page-form'` and here are the results:
* 80701d45bb - first mention. Introduced `create-page-form` as a Less function/inclusion or something
* 8e0a69f86a - added another usage of it in Less
* 4465c58f4b - seemingly added usage of it as of an actual ID to `gogs.css` (later renamed to `index.css`), still no use in templates
* 045f14fbd0 - unrelated mention in minified bundle
* 3d3faa2624 - ditto + did one property change to it like if it actually changed anything
* d9c67a8c90 - removed built bundle
* 202803fc69 - turned the strange Less thing into a proper pack of `#create-page-form` like if it was used, for some reason, duplicated the new code twice
* ee9e83b230 - removed some of the unused selectors by retiring the `.captcha-field` class, likely not knowing that nothing under `#create-page-form` is used
Reviewed-on: https://codeberg.org/forgejo/forgejo/pulls/9933
Reviewed-by: Gusted <gusted@noreply.codeberg.org>
Co-authored-by: 0ko <0ko@noreply.codeberg.org>
Co-committed-by: 0ko <0ko@noreply.codeberg.org>
These helpers are harmless for the normal Forgejo UI but are causing the viewport to break if the page is taller than the screen, causing overflow scroll in both directions.
Example: 404 page on Codeberg breaks because Codeberg is making the content of all pages full screen height before footer.
Originally added in bb50ab2861 but were likely made unused by some change at a later point, maybe in 91b5aa0e5f. I was not able to find a viewport where toggling this class via devtools would make any visual difference.
Reviewed-on: https://codeberg.org/forgejo/forgejo/pulls/9753
Reviewed-by: Gusted <gusted@noreply.codeberg.org>
Fix regression of https://codeberg.org/forgejo/forgejo/pulls/5589 & https://codeberg.org/forgejo/forgejo/pulls/7006.
Both added invisible input fields with `required` attribute into the forms used for creating issues and adding comments via `combomarkdowneditor.tmpl`. These fields were not related to the form and were only used for the modal dialogs.
* When JS is on, modals are moved out of the form to the bottom of the page by JS that handles modals
* When JS is off, the forms are stuck with invisible inputs preventing form submission
* Another side effect is that request data contains these unwanted fields
This was fixed by adding attribute `disabled` to unwanted inputs. Then JS removes it on initialization. Thanks to @Beowulf for helping with choosing the approach!
When a field is `disabled`, the browser doesn't consider it in form validation and doesn't even include it in the request data.
Reviewed-on: https://codeberg.org/forgejo/forgejo/pulls/9614
Reviewed-by: Gusted <gusted@noreply.codeberg.org>
Reviewed-by: Beowulf <beowulf@beocode.eu>
