Re: [Rpm-maint] [rpm-software-management/rpm] Fall back to /tmp if TMPDIR is bogus (#1567)
So... I think it's better to just honestly fail if the setup is borken, rather than try build fallbacks into silly shell scripts. -- You are receiving this because you are subscribed to this thread. Reply to this email directly or view it on GitHub: https://github.com/rpm-software-management/rpm/pull/1567#issuecomment-815583729___ Rpm-maint mailing list Rpm-maint@lists.rpm.org http://lists.rpm.org/mailman/listinfo/rpm-maint
Re: [Rpm-maint] [rpm-software-management/rpm] Fall back to /tmp if TMPDIR is bogus (#1567)
Closed #1567. -- You are receiving this because you are subscribed to this thread. Reply to this email directly or view it on GitHub: https://github.com/rpm-software-management/rpm/pull/1567#event-4567318242___ Rpm-maint mailing list Rpm-maint@lists.rpm.org http://lists.rpm.org/mailman/listinfo/rpm-maint
Re: [Rpm-maint] [rpm-software-management/rpm] Fall back to /tmp if TMPDIR is bogus (#1567)
Yup, this sounds like papering over some issue in either local setup or something else, and I don't think we should do that. -- You are receiving this because you are subscribed to this thread. Reply to this email directly or view it on GitHub: https://github.com/rpm-software-management/rpm/pull/1567#issuecomment-800990192___ Rpm-maint mailing list Rpm-maint@lists.rpm.org http://lists.rpm.org/mailman/listinfo/rpm-maint
Re: [Rpm-maint] [rpm-software-management/rpm] Fall back to /tmp if TMPDIR is bogus (#1567)
Containers aside, I'm still now sure why the `TMPDIR` variable would be set to anything other than a directory? -- You are receiving this because you are subscribed to this thread. Reply to this email directly or view it on GitHub: https://github.com/rpm-software-management/rpm/pull/1567#issuecomment-800989288___ Rpm-maint mailing list Rpm-maint@lists.rpm.org http://lists.rpm.org/mailman/listinfo/rpm-maint
Re: [Rpm-maint] [rpm-software-management/rpm] Fall back to /tmp if TMPDIR is bogus (#1567)
> @DemiMarie - interesting. I suppose you have some kind of script to set that > all up - care to share? I do! Two of them in fact. One uses podman and another uses bwrap. I also have the Containerfile I use. I don’t mind sharing them, but be warned that they have not been cleaned up and probably have bugs. They will almost certainly require tweaking for use on anyone else’s system. > Thing is, that sounds almost like a ticket out of this fakechroot trap we're > in and would love to replace it with something more robust, but haven't found > the time to get it done. Yes and no. Having to bring an entire distro along just to run the tests seems like mega-bloat to me. That said, if the tests are that tightly coupled to the host environment, it might be the best option -- You are receiving this because you are subscribed to this thread. Reply to this email directly or view it on GitHub: https://github.com/rpm-software-management/rpm/pull/1567#issuecomment-800497202___ Rpm-maint mailing list Rpm-maint@lists.rpm.org http://lists.rpm.org/mailman/listinfo/rpm-maint
Re: [Rpm-maint] [rpm-software-management/rpm] Fall back to /tmp if TMPDIR is bogus (#1567)
Ack. In fact depending on the host is far from ideal either, because then we need to worry about all manner of strange details in which hosts may vary, this case may well be one of those (I've seen my share of fakechroot weirdling behavior due to external symlinks etc). So having the test-suite run in a known version of a known distro, in a container, would be far better in many ways. Fedora is as good as any for that image, but we'd probably want an image that does *not* include rpm it, so we could put our development version as the system rpm there. But this is getting a little outside the scope here :sweat_smile: -- You are receiving this because you are subscribed to this thread. Reply to this email directly or view it on GitHub: https://github.com/rpm-software-management/rpm/pull/1567#issuecomment-800250510___ Rpm-maint mailing list Rpm-maint@lists.rpm.org http://lists.rpm.org/mailman/listinfo/rpm-maint
Re: [Rpm-maint] [rpm-software-management/rpm] Fall back to /tmp if TMPDIR is bogus (#1567)
Correct; it's a Fedora-based image, similar to our [ci/Dockerfile](https://github.com/rpm-software-management/rpm/blob/master/ci/Dockerfile). You're right - Re-reading her comment, it does seem she's using an empty image, after all. Something like toolbox, in fact. I wonder, though - what difference would that make (i.e. bind-mounting everything versus just the sources), with regards to how the testsuite behaves? -- You are receiving this because you are subscribed to this thread. Reply to this email directly or view it on GitHub: https://github.com/rpm-software-management/rpm/pull/1567#issuecomment-800168774___ Rpm-maint mailing list Rpm-maint@lists.rpm.org http://lists.rpm.org/mailman/listinfo/rpm-maint
Re: [Rpm-maint] [rpm-software-management/rpm] Fall back to /tmp if TMPDIR is bogus (#1567)
@dmnks - but you have an actual OS image in Podman, right? AIUI @DemiMarie has an essentially empty image where even the OS is populated with bind-mounts from the host, which is basically how we use fakechroot. -- You are receiving this because you are subscribed to this thread. Reply to this email directly or view it on GitHub: https://github.com/rpm-software-management/rpm/pull/1567#issuecomment-800149253___ Rpm-maint mailing list Rpm-maint@lists.rpm.org http://lists.rpm.org/mailman/listinfo/rpm-maint
Re: [Rpm-maint] [rpm-software-management/rpm] Fall back to /tmp if TMPDIR is bogus (#1567)
FWIW, the way I use Podman here is pretty much the same - I just bind-mount my git checkout into the container and do the building and testing (including running the testsuite) from there. I do have a fancy [script](https://github.com/dmnks/dotfiles/blob/master/utils/bin/codebox) for making the management of such throwaway development containers a breeze, but that's completely optional :) -- You are receiving this because you are subscribed to this thread. Reply to this email directly or view it on GitHub: https://github.com/rpm-software-management/rpm/pull/1567#issuecomment-800133584___ Rpm-maint mailing list Rpm-maint@lists.rpm.org http://lists.rpm.org/mailman/listinfo/rpm-maint
Re: [Rpm-maint] [rpm-software-management/rpm] Fall back to /tmp if TMPDIR is bogus (#1567)
@DemiMarie - interesting. I suppose you have some kind of script to set that all up - care to share? Thing is, that sounds almost like a ticket out of this fakechroot trap we're in and would love to replace it with something more robust, but haven't found the time to get it done. -- You are receiving this because you are subscribed to this thread. Reply to this email directly or view it on GitHub: https://github.com/rpm-software-management/rpm/pull/1567#issuecomment-800022408___ Rpm-maint mailing list Rpm-maint@lists.rpm.org http://lists.rpm.org/mailman/listinfo/rpm-maint
Re: [Rpm-maint] [rpm-software-management/rpm] Fall back to /tmp if TMPDIR is bogus (#1567)
> I routinely do a `make check` in Podman containers and never encountered this > issue. Is there some specific set-up (of the testsuite) that you're referring > to? The container image itself is practically empty ― everything is bind-mounted from the host. To be fair, it is possible that the container image that I am using is buggy, or that I am misusing podman. Additionally, I have my source directory bind-mounted at `/source` and my build directory bind-mounted at `/build`. -- You are receiving this because you are subscribed to this thread. Reply to this email directly or view it on GitHub: https://github.com/rpm-software-management/rpm/pull/1567#issuecomment-799703639___ Rpm-maint mailing list Rpm-maint@lists.rpm.org http://lists.rpm.org/mailman/listinfo/rpm-maint
Re: [Rpm-maint] [rpm-software-management/rpm] Fall back to /tmp if TMPDIR is bogus (#1567)
I routinely do a `make check` in Podman containers and never encountered this issue. Is there some specific set-up (of the testsuite) that you're referring to? -- You are receiving this because you are subscribed to this thread. Reply to this email directly or view it on GitHub: https://github.com/rpm-software-management/rpm/pull/1567#issuecomment-799609306___ Rpm-maint mailing list Rpm-maint@lists.rpm.org http://lists.rpm.org/mailman/listinfo/rpm-maint
Re: [Rpm-maint] [rpm-software-management/rpm] Fall back to /tmp if TMPDIR is bogus (#1567)
@DemiMarie pushed 1 commit. a49f073f15436d0bdea4125dd6189d6ee7e6cac5 Fall back to /tmp if TMPDIR is bogus -- You are receiving this because you are subscribed to this thread. View it on GitHub: https://github.com/rpm-software-management/rpm/pull/1567/files/e05be91780a5c2011a5b8bd9933d944c36d93c0f..a49f073f15436d0bdea4125dd6189d6ee7e6cac5 ___ Rpm-maint mailing list Rpm-maint@lists.rpm.org http://lists.rpm.org/mailman/listinfo/rpm-maint
Re: [Rpm-maint] [rpm-software-management/rpm] Fall back to /tmp if TMPDIR is bogus (#1567)
> I'd be more interested in understanding what makes $TMPDIR bogus in this case. I suspect that it is a fakechroot bug, but I have not yet been able to find a workaround other than this. -- You are receiving this because you are subscribed to this thread. Reply to this email directly or view it on GitHub: https://github.com/rpm-software-management/rpm/pull/1567#issuecomment-795965786___ Rpm-maint mailing list Rpm-maint@lists.rpm.org http://lists.rpm.org/mailman/listinfo/rpm-maint
Re: [Rpm-maint] [rpm-software-management/rpm] Fall back to /tmp if TMPDIR is bogus (#1567)
I'd be more interested in understanding what makes $TMPDIR bogus in this case. -- You are receiving this because you are subscribed to this thread. Reply to this email directly or view it on GitHub: https://github.com/rpm-software-management/rpm/pull/1567#issuecomment-795494743___ Rpm-maint mailing list Rpm-maint@lists.rpm.org http://lists.rpm.org/mailman/listinfo/rpm-maint
[Rpm-maint] [rpm-software-management/rpm] Fall back to /tmp if TMPDIR is bogus (#1567)
This allows the testsuite to run in a Podman container. You can view, comment on, or merge this pull request online at: https://github.com/rpm-software-management/rpm/pull/1567 -- Commit Summary -- * Fall back to /tmp if TMPDIR is bogus -- File Changes -- M scripts/check-files (2) -- Patch Links -- https://github.com/rpm-software-management/rpm/pull/1567.patch https://github.com/rpm-software-management/rpm/pull/1567.diff -- You are receiving this because you are subscribed to this thread. Reply to this email directly or view it on GitHub: https://github.com/rpm-software-management/rpm/pull/1567 ___ Rpm-maint mailing list Rpm-maint@lists.rpm.org http://lists.rpm.org/mailman/listinfo/rpm-maint