Problem/Motivation

Probably a fair assumption in real life, but not in the murky world of update path tests.

#3290810: Remove updates added prior to 9.4.0 (9.4.4 for ckeditor) and add 9.4.0 database dumps.

Steps to reproduce

Proposed resolution

Remaining tasks

User interface changes

API changes

Data model changes

Release notes snippet

CommentFileSizeAuthor
#8 3296108.patch713 bytescatch
#5 3296108.patch668 bytescatch
#2 3296108.patch586 bytescatch

Comments

catch created an issue. See original summary.

catch’s picture

Version: 10.1.x-dev » 9.4.x-dev
Status: Active » Needs review
StatusFileSize
new586 bytes
spokje’s picture

Status: Needs review » Reviewed & tested by the community

RTBC if TestBot agrees, and Yay! to less murky worlds in general and specifically for update path tests.

dww’s picture

  1. We already saw this patch pass tests at #3292008-39: [Ignore] In space (and/or this issue), no one can hear patches scream III, vs the fail in #38.
  2. The change is trivially small, and does exactly what it needs to.
  3. In Slack, we agreed this doesn't need its own dedicated tests.
  4. This doesn't seem like the kind of bug that needs a CR or a release note.

Perhaps premature, but I'm going to RTBC, anyway. I know no committer would actually commit this until this bot comes back green, and it can prove me wrong and NW if it has to... 😉

Thanks!

catch’s picture

StatusFileSize
new668 bytes

Same patch just without a line missing from the diff *cough*.

dww’s picture

Hah, whoops. ;) Thanks!

mondrake’s picture

Status: Reviewed & tested by the community » Needs review
+++ b/core/modules/mysql/mysql.install
@@ -17,6 +17,10 @@ function mysql_requirements($phase) {
+      if ($connection->driver() !== 'mysql') {

Maybe overkill, but I would suggest

      if (!($connection->driver() === 'mysql' && $connection->getProvider() === 'mysql')) {

so we ensure we only execute for the mysql driver provided by the mysql module? (other modules, even installed, may provide a driver named mysql)

catch’s picture

StatusFileSize
new713 bytes

#7 might possibly be overkill but it won't do any harm.

spokje’s picture

If we wanna go backport to 9.4.x, looks like we're gonna need a separate patch for it.

mondrake’s picture

Status: Needs review » Reviewed & tested by the community

Sqlite and PgSql do not have a hook_requirements() implementation so this looks fine to me, thanks

dww’s picture

Version: 9.4.x-dev » 9.5.x-dev
Related issues: +#2733675: Warning when mysql is not set to READ-COMMITTED

Just did a little git archeology. There is no mysql_requirements() in 9.4.x. That was introduced in #2733675: Warning when mysql is not set to READ-COMMITTED, which wasn't backported to 9.4.x. The earliest core with this bug is 9.5.x. Therefore, bumping the version, and ignoring the fact that #8 doesn't apply to 9.4.x...

Thanks,
-Derek

p.s. Also saving credit for all the reviews here.

  • xjm committed fe941ff on 10.1.x
    Issue #3296108 by catch, dww, Spokje, mondrake: mysql_requirements()...

  • xjm committed ecd9a8e on 10.0.x
    Issue #3296108 by catch, dww, Spokje, mondrake: mysql_requirements()...

  • xjm committed 9b6af36 on 9.5.x
    Issue #3296108 by catch, dww, Spokje, mondrake: mysql_requirements()...
xjm’s picture

I didn't understand this issue at first, but after reading the whole hook, this is clearly a bug that it displays its warnings on other drivers. With the information in #10, I agree that the scope is complete here.

Thanks @dww also for the archaeology on why it didn't work on 9.4.x.

Committed to 10.1.x and cherry-picked to 10.0.x and 9.5.x. Thanks!

xjm’s picture

Status: Reviewed & tested by the community » Fixed

Status: Fixed » Closed (fixed)

Automatically closed - issue fixed for 2 weeks with no activity.