Various fixes to allow the daily update-driver-submodules workflow to complete.

[casasnovas]

Looks like the workflow never really worked as expected due to differences in the github worker host versus local runs.

Thanks to @LuKP17 for reporting the issue, which he caught because the drivers README listing supported drivers and their version was stale.

A good run with this branch lead to this PR: https://github.com/xcp-ng/hypervisor-dev/pull/34/changes#diff-7ed8b37d946b3f15d59f978f9047536de899f2949a4041580d4374cee5396e84

It correctly updated all srpm that had changed since last manual runs, as well as their companion source repos. The README changes correctly reflect this.

[LuKP17]

In the resulting drivers README, the links of the alt drivers are missing the "-alt" suffix in the name of the repo, leading to a "Page not found".
Do you want to add this fix now or in a future PR?

[nathanael-h]

Not modified by the PR, but you might want to look for :

https://github.com/xcp-ng/hypervisor-dev/blob/quentin-fix-update-workflow/.github/workflows/update-driver-submodules.yml#L60

git submodule add git@github.com/xcpng-rpms/$(basename "$submodule").git "drivers/8.3/source/$(basename "submodule")"

• repo URL typo, should have a : after .com, like git@github.com:xcpng-rpms/$(basename "$submodule").git
• org name misses an - it should be xcp-ng-rpms

Missmatch write a file with - and test one with _ in the name :
https://github.com/xcp-ng/hypervisor-dev/blob/quentin-fix-update-workflow/.github/workflows/update-driver-submodules.yml#L108
printf '%s\n' "${missing_sources[@]}" > /tmp/missing_sources.txt

https://github.com/xcp-ng/hypervisor-dev/blob/quentin-fix-update-workflow/.github/workflows/update-driver-submodules.yml#L147

test -f /tmp/missing-sources.txt && {

[casasnovas]

Thanks for your good catches guys! I've (presumably) fixed them with the few extra commits on the branch.

Sadly I cannot do a full test-run now that I'm using the PAT secret because it is not passed down to workflow, but I've run it locally and can confirm it fixed the -alt URL. I'll retest after merging and address any newly found issue in a new PR if that's OK, unless you can already see some issues with my iterative fixes.

[nathanael-h]

Ok for me!

[LuKP17]

I will try to review next wednesday I would say, I'm busy with QA for a few days.
I'm not overly familiar with these languages but it looks like I can infer most of the changes made.

[casasnovas]

I will try to review next wednesday I would say, I'm busy with QA for a few days. I'm not overly familiar with these languages but it looks like I can infer most of the changes made.

I would say it's not urgent as @ydirson has some concerns over pushing source code as branches into the SRPMs repos. Even though this predates this PR, I am assuming it will likely change on the implementation side until we have a solution everyone is happy with.