Silence the deprecation warning instead to avoid regressing T372193.
I filed T372592 about the bigger problem we have here.
This reverts commit fe8526caf2.
Bug: T372499
Change-Id: I47884f99e9c18fffd22ffdc27f5c61515b60c8f8
Record if a talk page is not empty, rather than empty.
This helps prevent Parsoid from mislabelling pages as empty when they
contain extension tags it needs to call out to the legacy parser to
process. Metadata from the legacy parse is merged into the overall
metadata for the page.
For example, a templatestyles extension tag parsed in the context of a
talk page would set the empty state which would then fail to cleared if
the overall parse of the page wasn't empty.
Bartosz (@matmarex) proposed this idea in T371125#10053043
Setting and using these flags all appear to be post-retrieval from
cache.
Alternative to I1deb679ef0e19dc3a36e377c183dd1b4ab0003e9
Bug: T371125
Change-Id: I5c1877f7f9eb73f88a33e001ca3c2f3d06bb90e4
While investigating primary database queries as part of T370304,
we found that EventDispatcher currently issues a primary DB query
to fetch the previous revision of the newly created revision on
every edit, but discards the result for non-talk pages. So, as an
optimization, have it fetch the previous revision only after the page
has been established to be a talk page.
Bug: T370304
Change-Id: I301f5c3c002c6953ef0b3919a33667a809342b84
The previous approach no longer works due to changes in MediaWiki.
We must now pass the data between the hook which has access to request
state needed to determine if we should show this, and the hook that is
capable of showing it in the right place.
Bug: T362496
Change-Id: I290422c4249d288badd7c7d25e5e7e435dfd1800
It's unlikely we'll do another feature A/B test any time soon.
Bug: T322492
Bug: T341491
Depends-On: Ia3712e2930fcd971bce44f568430602ce3949f23
Change-Id: I1ef4191f9466b7420a2fead571615ed6d49f873e
I've been reading some discussions in Special:DiscussionToolsDebug
and saw a good number of comments where the first paragraph has
a bullet point (`*`), and the following paragraphs are indented (`:`).
Change-Id: I1afadc49565e843c97286a3744941ffb4062ddec
urldecode() should be used for decoding URL query parameters,
rawurldecode() should be used for decoding URL paths.
Bug: T367977
Change-Id: I7a7b14da85fb89f612c701d2746803d830017842
WMF config change should be deployed first: Ica7dba4f17eb3e57a1a2415cc970c54a67eac696
The default value for this preference was already '1', and this code
was a no-op unless it was overridden to '0' like it was in WMF config.
Bug: T357221
Change-Id: I6b09bbea9bfb7a8f8ed17114dc59e657c5b66e91
Changes to the use statements done automatically via script
Addition of missing use statement done manually
Change-Id: I492ddae3c3e9f81d551ea8b348b7148893f4f590
Currently, DT is making more than 30 exact queries back to back in every
request. It is clear the caching is completely broken.
It is because when the properties don't exist (majority of the cases),
it returns an empty array making the caching noop.
Tested in mwdebug1002 and it fixes the issue.
Bug: T364693
Change-Id: I182ae121999df1a04cfb7399bc49891587a37074
It needs to only look at the end of the added comments, and ignore
whatever is going on at the beginning, since the only thing it really
cares about is a functioning "Reply" button at the end.
Bug: T363285
Change-Id: Ia337be754deda741617d1343972f3e0a21c41b05
The PHP code should never see a `<span class="mw-headline">` node
since MediaWiki change If04d72f427ec3c3730e757cbb3ade8840c09f7d3,
but we have to support it for our old integration tests (T363031).
Improve code comments to explain this and move the handling to
one place, so that it can be deleted more easily in the future.
Follow-up to 08f61b2609.
Change-Id: I5ab9d3373a6911c1456c30d844b66576b278a1b5
