On my machine `npm test` fails with a lot of ESLint errors for files in vendor folder. I guess mediawiki-vagrant creates the folder.
Change-Id: I5466ad8c490b39bef9a31f5442fecd2d11895b80
Depends-On: Ib7f0bd15dd0a9255e1e5140907e800478b658b92
Using MakeGlobalVariablesScript hook will add config variables to
the HTML meaning cached HTML will not be able to access it leaving
it as undefined. This will mess up any existing caching.
Additional changes:
Because I'm paranoid, make sure that value is never undefined.
Bug: T170648
Change-Id: Icb8e34104c57b1e69b7e035ef89d5b4c3371132f
Wikivoyage pages have more than 3 related articles as they make use
of the {{#related:}} magic word.
After speaking with Nirzar we should allow this project to show more than 3.
This change allows this while keeping the existing behaviour on other wikis and
will pave the way for removing a bunch of code from this extension.
Additional changes:
* Cleanup skinStyles definitions
* Limit cards to 30% maximum width and give margin top to account for
situations where the number of cards are multiple of 3
** In Minerva hardcode the max-width to pixels.
* Margins are switched from hardcoded 10px to percentage based. Yes
this changes the right margin slightly but is more maintable and visually
the same.
Bug: T164765
Change-Id: I41119de3228c2df799f740d4bd00082101c21b97
The following sniffs are failing and were disabled:
* MediaWiki.Commenting.FunctionComment.MissingParamComment
* MediaWiki.Commenting.FunctionComment.MissingParamTag
* MediaWiki.Commenting.FunctionComment.ParamNameNoMatch
* MediaWiki.FunctionComment.Missing.Public
Change-Id: I591fd0cd60147b9f5ddd3d1b9e74f79a0bb3f595
In order to avoid further confusion, clarify that we're bucketing users
for experiments.
Bug: T167236
Depends-On: I752cdf068ca25bffb38229380785e7da1a208049
Change-Id: I6508ab8fa3d436ae295eb047e170fdc5417d25fc
Currently, the mw.eventLog.Schema class samples per pageview. However,
we expect that if a user is bucketed for a session, then all
EventLogging events logged during that session are in the sample.
Unlike Popups in I4f653bba, the RelatedArticles instrumentation does
rely on another mw.eventLog.Schema behavior: distinguishing between
static and dynamic data. This and the use of mw.eventLog.Schema could be
removed in a follow-on change.
Bug: T167236
Change-Id: I9583cb98264efd7dc46b45bbce5323036a100536
* Move across all files
* Rename ext-card- prefix to ext-related-articles- prefix
** Since all code using these prefixes is JS
we do not have to worry about cached HTML
Bug: T137021
Change-Id: I784fd132c36329fa0dcc49fe2804460061940347
We're going to want to use a newer version of JSHint
anyway to support use of ES6 in our browser tests so now
seems a good time to do this.
Bug: T149202
Change-Id: I5526b020cfc12c0e065ad15ed711a0e3a7bff1bc
This change removes RelatedArticles from beta features.
It repurposes the BlacklistSkins config to be a whitelist that describes
when related pages should be shown in the footer.
To avoid enabling related pages on desktop, this depends on a config
change that makes use of the new configuration option.
Simplify browser tests to reflect the new status quo.
Bug: T146436
Bug: T160076
Depends-On: I522e0182d1c3d9261bd0561c3ec0c789b6431c7a
Change-Id: I84da1fb33a901a6365375d00d192af35422ff0b0
Since we're always depending on the event logging module, we can
make it a dependency in extension.json, rather than loading it
lazily.
Bug: T157375
Change-Id: Ia57f390586e123c6010a7daf23a3d851daf079ce
Log the `feature-enabled` event for users who are eligible to see the
ReadMore feature. Otherwise log the 'feature-disabled' event.
Suporting changes:
* Register the latest version of the RelatedArticles schema.
Bug: T157375
Change-Id: I1557de88c7602aa066833bd55b83e6ee51d27178
This allows us to decouple the event logging code from the
`ext.relatedArticles.readMore' module. As a result we'll be able to
log events without loading the above mentioned module.
Bug: T157375
Change-Id: I55cf4f40cafc88d6baeb5cc1b41fc2d6bcd2fbb9
Additional change:
* Correct how we expose RelatedArticlesEnabledSamplingRate so that
it includes the wg prefix and is consistent with other config values
Bug: T157372
Change-Id: I274de542cf461de60f903ddbc0353a4116016007
On the beta cluster, the browser test should pass.
Add a new test to reflect the two different states.
Bug: T157165
Change-Id: I2a777436dfa3a14d58855449c39cc144b7149a62
This previously passed with a false positive as it replicated
the scenario `ReadMore is not present in minerva stable` by running
on the mobile domain.
This change makes it run on the desktop domain and pass (it now
fails given it runs on Minerva skin)
Bug: T157165
Change-Id: Idb2002b0f2d378cc0b6df73ad1381bfee32448c9
