[DX] Exclude .tests.info files from requiring a "type =" line.

[klonos]

These are not modules or themes or layouts. The "right" way to do it would be to perhaps require a type = test line, but I say we make this optional and instead rely on the fact that these live within /test subdirectories.

This came up in backdrop-contrib/honeypot#6 and elsewhere where modules would not be installed because we check and validate .info files in every subdir inside a project. Lets ease DX and improve UX.

Steps to reproduce:
1)
2)
3)

Backdrop version 1.2.2

---

PR: backdrop/backdrop#2491

[quicksketch]

Rather than ignoring test directories, I'd recommend just stopping at the first .info file found, as suggested in backdrop-contrib/honeypot#6 (comment). It will accomplish the same thing in hopefully a slightly more robust approach.

[quicksketch]

Oh, that other issue is in Honeypot's queue, not core. Reopening to fix this here.

[pifbits]

I'm still having the issue honeypot. Manually install via cmd/file manager works tho. it's also happening with mimemail, civicrm, and a few others. I'm just testing config modules for now. I'm not even at the point of dropping tables and re-importing a db-copy from the d7 production site yet.

[jenlampton]

The files that are causing problems are the ones named *.tests.info so can't we just stop the checker from checking those specific files? I'm going to attempt a PR, I'll see how hard that is.

[jenlampton]

I'm unable to reproduce this problem on backdrop 1.7.0 or above. Can someone confirm that it's still an issue?

[herbdool]

I can't reproduce either using honeypot.

[klonos]

This has now become an issue with #3105 which we are aiming to get in for the upcoming 1.11

[herbdool]

I added a PR to exclude .tests.info in the Updater class backdrop/backdrop#2491. The function was already smart enough to first try to fetch the .info that matches the directory name, and, if that didn't work, it would take the first .info file in the list. That seemed to exclude tests.info in my testing, but this PR will make sure.

I haven't been able to recreate the issue so perhaps someone else can test.

[quicksketch]

That PR looks good to me @herbdool. I made a minor change to include the period in the substr() check, per @opi's comment that we shouldn't match on a string like footests.info.

[quicksketch]

I've merged backdrop/backdrop#2491 into 1.x and 1.12.x. Thanks for the fix @herbdool and review from @opi!

[klonos]

Follow-up: #3821