On Friday 06 September 2013 10:43:52 Hanchett, Paul wrote:
> If OBS and GBS are building the same package, why wouldn't they get the
> same result (assuming no circular references)?
Well, GBS is just not OBS.
GBS is made to build from a git source tree and make that easy - extended to
build multiple dependent packages.
But a full distro with it's cycles and cross-references is more complex. OBS
was built for exactly that.
>
> FYI, I'm finding a lot of build errors that are hard to understand-- Things
> like:
>
> - Configure script gives errors
> - Various compile errors (conflicting definitions/undeclared; warnings,
> with "treat warnings as errors" turned on..; etc.)
> - use of deprecated api's g_mutex_new, g_mutex_free, g_cond_timed_wait
> - ld errors
> - Multiple missing 1.x patch files on 2.x build
> - Missing libraries
> - "Architecture is excluded" (i686 with -Ai586 on build line)
> - rpmlint exceeds "badness" score
> - rpm error "installed but unpackaged files found"
> - Makefile.am: required directory does not exist (".intl)
> - use of %(x) construct causes lots of commentary from rpm...
> - quite a number and variety of other warnings
We might hit here differences between what is in the repo/binaries and what you
build from the source tree through gbs. What does the src.rpm contain vs. the
tarball you get through gbs.
>
> All but the last two would have halted the package build. I expect that a
> developer touching the package would have tried to rebuild the package and
> would have noticed and repaired any package defects.
>
> It is as though changes are getting checked in and peer reviewed without
> the package being built to confirm that the package is created...
There is a peer review, and they're built within the OBS. Maybe someone
eplains to us how the gerrit review-process to package - to OBS - to repo
works right now in Tizen.
> So, how do we fix this? Then, how do we verify that what is fixed for gbs
> is also correct for obs? (In these cases, it really should be--but if it's
> building in obs then something is not reciprocal...)
Really good points and we need to make this process available for all
participants.
> For what it's worth, I'm more than willing to help--I just want to be sure
> that what I'm doing is really the Right Thing To Do. :-)
>
> Thanks for listening,
> Paul
>
BR,
Jan-Simon
_______________________________________________
IVI mailing list
[email protected]
https://lists.tizen.org/listinfo/ivi
