Learn more about these different git repos.
Other Git URLs
See also: #817
The current situation with volume id substitutions is very unsatisfactory
Not fixed on a particular approach yet. This is open to discussion
Metadata Update from @mikem: - Issue set to the milestone: 1.17
Agreed that it's rather icky.
I really would like to see the mechanisms and rules for generating both image file names and volume IDs be much more 'procedural' than they currently are, and live in exactly one place. That could be in Pungi, or it could be in Koji, but only if all Pungi tasks that produce images ran through Koji (currently AIUI they do not, specifically traditional installer images are not built in Koji).
The easiest way I can see of improving this for now is for the Koji tasks which actually use this code in Koji - which AFAICS is only the livecd-creator and livemedia-creator tasks - to allow Pungi to pass them desired volume IDs (and while we're at it, filenames), which they currently do not. That would allow the actual work of generating the names to be done only on the Pungi side.
There are ways we could improve how Pungi actually does that work, also, but that'd be out of scope for Koji of course.
Metadata Update from @tkopecek: - Issue tagged with: discussion
I'm wondering why the volume-ids weren't simply a task option to begin with. Perhaps folks worried that users would declare them inconsistently? @ausil do you recall?
(because certainly, just passing them in is easier all around)
https://bugzilla.redhat.com/show_bug.cgi?id=1145264 I think is what led to the current behaviour. It boiled down to we needed a short less terrible path and put a patch in koji. making it be passed in is probably the best option, with the old behaviour of truncation being the default. @lsedlar what would it take to make the changes to pungi to pass in the volumeid?
https://lists.fedoraproject.org/pipermail/infrastructure/2014-November/015138.html Where I said koji was the wrong place.
This should be fairly easy to change in Pungi. The tricky part might be doing it in a way that supports older Koji that does not accept the argument.
PR #1227
Commit 4f45059 fixes this issue
Login to comment on this ticket.