Since Packit creates separate projects for each PR IIUC, it is not necessary
to - at least from Copr POV - require maintainers to trigger the build. We
can simply start them.
TBH, I am not 100% sure if I get this right.
Since Packit creates separate projects for each PR
Yes, by-default packit is the owner of that Copr project.
it is not necessary to - at least from Copr POV - require maintainers to trigger the build.
You mean GitHub project maintainer, don't you?
If the user or organisation is approved (and it is I think) and user has merge permission in that project, we trigger the build automatically. (I think the write access is the one we are speaking about, maintain is higher than write.)
If this does not work, there might be some bug during the process and we should fix it.
this would make things so much easier on our side
@praiskup if you are fine with such behaviour, as the COPR project lead, I'm pretty sure we can deliver :)
I think the write access is the one we are speaking about, maintain is higher than write
That's what I meant ... from Copr perspective, you don't need to require write
permissions as the users can not make any damage. The only thing that user
can affect is the build itself, or crack the concrete builder VM. The VMs are
reused for other builds, but only for buidls in the same project (and as each
PR means separate Copr project, it is just fine).
this would make things so much easier on our side
Packit would be actually even more convenient to use.
@praiskup Thanks for the explanation.
Can we get rid of the whitelist as well then? (I think you discussed this with @TomasTomecek some time ago.)
I'm not sure what do you mean by whitelist in this context.
OK, we have two-step verification:
write permission in that project).
- Allow namespace/user to build in Copr in general (=whitelist).
Ah, from Copr perspective _you_ (packit user) are the user who is allowed
to create projects and submit builds. So the problem like who/what is able
to build in Copr through GH pull request is solely in your hands.
What I mean..., yes - there _are_ some Copr rules which packit user needs to follow,
like legal POV and others, even packit docs say that. But if someone crosses the border through PR, I'm sure it
would be logged somewhere and packit moderators can then administratively
remove such affected build, etc.
Can we get rid of the whitelist as well then?
I'm reading this question once more, and now I'm sure it was not targeted to me :-) sorry for noise.
Can we get rid of the whitelist as well then?
I'm reading this question once more, and now I'm sure it was not targeted to me :-) sorry for noise.
No worries, thanks for the great elaboration.
Can we get rid of the whitelist as well then? (I think you discussed this with @TomasTomecek some time ago.)
I'd say we can't get rid of it for good. Though we could create a better solution which would be more straightforward (e.g. after installing packit, we'd redirect to our page where we would provide "rules" of the service which our users would need to "accept").
To clarify what we want to do:
Updating the Packit upstream work column so we can groom it and start the work.
I'd like to clarify the outcome of this card:
/packit build)?I'd like to clarify the outcome of this card:
- based on the title and description of the issue, we want to allow _anyone_ creating PR against a repository to trigger the build (change included in the PR)
+1
- do we also want to apply similar set of rules to comments (e.g.
/packit build)?
(change in PR allows only initial build after creating the PR, not comments by author of the PR without write permissions)
a. keep as is
b. allow author of the PR to trigger the build by commenting too (ignoring if author has or has not the write permissions to the repository)
I'd say that we should allow PR authors to do /packit build: on the other hand, if some random Joe Foo Bar comments on a PR where he's not the author, we should not change commit statuses and probably comment in that PR that the user doesn't have permissions to do that.
Most helpful comment
+1
I'd say that we should allow PR authors to do
/packit build: on the other hand, if some random Joe Foo Bar comments on a PR where he's not the author, we should not change commit statuses and probably comment in that PR that the user doesn't have permissions to do that.