- Issue created by @catch
- last update
about 1 year ago 30,434 pass, 1 fail - @catch opened merge request.
- Status changed to Needs review
about 1 year ago 8:29am 27 October 2023 - 🇬🇧United Kingdom catch
Another one on the daily branch run:
https://git.drupalcode.org/project/drupal/-/jobs/233490
The problem here is that it's hitting packagist and github at all, it should not be. So... would just adding --lock allow the test to still pass?
- last update
about 1 year ago 30,434 pass, 1 fail - Status changed to Needs work
about 1 year ago 3:01pm 27 October 2023 - 🇺🇸United States smustgrave
Seems to have failure.
But should drupal/core-* be used like recommended in https://www.drupal.org/project/drupal/releases/10.1.5 →
- Status changed to Needs review
about 1 year ago 9:43pm 27 October 2023 - 🇬🇧United Kingdom catch
Think I might have it:
1. Build tests exclude /vendor when copying the code base.
2. This tests calls composer update, which ends up downloading real packages.
3. Sometimes that goes wrong.If we copy /vendor, nothing gets downloaded, so nothing breaks.
I debated whether we should change the default or only this test, at least if no tests fail due to changing the default, I think we should change that, since calling out to github during actual tests (not just as a prerequisite to running tests) is not ideal. If we really want tests to depend on packagist/composer that should be explicitly opt-in.
- 🇬🇧United Kingdom longwave UK
I don't think tests should ever depend on packagist; networks are just too unreliable and we will run into repeated random issues such as this one.
- 🇬🇧United Kingdom catch
Bumping to critical because afaict this is our most frequent random test failure on gitlab at the moment.
- 🇳🇱Netherlands spokje
Changes makes sense...
...now, I don't think in about one year from now I will look at that file iterator and think to myself "Of course we included the vendor-directory, otherwise we would trigger packagist".
TLDR: A comment why
getCodebaseFinder()
doesn't exclude vendor (after this change) will probably save somebody some time within a year. Otherwise this is a nice find and good to go. - Status changed to RTBC
about 1 year ago 2:07pm 28 October 2023 - 🇺🇸United States smustgrave
Was posted in #bugsmash and reviewing the MR change makes sense. Going to mark but also agree with @Spokje comment may be useful so someone doesn't have to use the git blame to find this ticket.
- last update
about 1 year ago Custom Commands Failed - First commit to issue fork.
- Status changed to Fixed
about 1 year ago 3:46pm 29 October 2023 Automatically closed - issue fixed for 2 weeks with no activity.