This has been instrumented with a warning for a number of years and is generally considered an anti-pattern.

The good news is that this isn't actually a pattern we've had to move away from, it was just never really a thing in recent history.

Despite that, we have seen in two cases in recent years that changes to code accidentally resulted in this indirectly happening and we missed it for a while because it isn't enforced.

• 2018: T191116: Wikimedia\Rdbms\Database::tableName: use of subqueries is not supported this way.
• 2022:

In T314189#8216809, @Ladsgroup wrote:

We are getting 1.3M/day warnings for bad use of table name in building queries in RC.

It's probably something […] turned on this week but this needs fixing.

This was caused by:

SpecialRecentChangesLinked: Pass query builder instead of SQL

https://gerrit.wikimedia.org/r/c/mediawiki/core/+/825078

And fixed in:

In T314189#8237478, @gerritbot wrote:

Change 831082 merged by jenkins-bot:

[mediawiki/core@master] rdbms: Allow SubQuery objects in SelectQueryBuilder as table

https://gerrit.wikimedia.org/r/831082

Proposal

Turn the warning into an exception. As suggested by @aaron:

In T191116#8269609, @aaron wrote:

Can this warning become an exception now (or the check removed completely)?