Thanks for the change! I just realized that my patch has another problem: when checking assignments, it should also check that the receiver is the same in both branches, and this would avoid the case in LoadBalancer. I also find excessive the case in GlobalFunctions, and that would be solved by implementing a check on the complexity of the 'if' condition, as proposed in gerrit comments. Ideas on how to perform such check are welcome.
Il giorno lun 11 feb 2019 alle ore 19:42 Bartosz Dziewoński < matma.rex@gmail.com> ha scritto:
On 2019-02-11 18:42, Daimona wrote:
Hi, All patches in the codesniffer repo have a sample run against mwcore set
up
in CI. As can be seen in [0], the current version is triggered 13 times
by
MW core. No idea about extensions, though. Daimona
[0]:
https://integration.wikimedia.org/ci/job/mw-tools-codesniffer-mwcore-testrun...
Thanks for that link. I looked at them and submitted a change (please do not merge it) to demonstrate what changes this would require: https://gerrit.wikimedia.org/r/c/mediawiki/core/+/489759
In my opinion most of these changes are clear improvement or harmless, except for the pattern in LBFactorySimple.php/LoadBalancer.php, which is a little tricky and probably clearer in the original version.
-- Bartosz Dziewoński
Wikitech-l mailing list Wikitech-l@lists.wikimedia.org https://lists.wikimedia.org/mailman/listinfo/wikitech-l