tags 567905 + fixed-upstream pending thanks On Tue, Feb 23, 2010 at 10:05:31AM +0100, Philipp Kern wrote: > On Mon, Feb 01, 2010 at 05:39:55PM +1300, martin f krafft wrote: > > If I build a arch:all package without --arch-all, then sbuild will > > skip the build, saying e.g.: > > > > backupninja_0.9.7-4~bpo50+1.dsc: i386 not in arch list or does not > > match any arch wildcards: all -- skipping > > > > It also returns with exit code 1. I would prefer if it exited with > > 0 since there wasn't really an error — the behaviour was > > well-defined and exactly what I wanted. > > For the autobuilders it's actually an error because it shouldn't have > been tried in the first place. Why does the exit code matter to you? > Would batch mode help? We could define another exit code there. > > In the general case it does not build anything so I think exit code > 1 is not entirely insane in user mode.
This one is certainly a corner case where either exit status could be appropriate depending upon how you look at it. - it's not built anything (fail?) - it's failed the arch check (fail?) - but we asked sbuild to "do nothing", so it actually succeeded at what we requested it to do--the package didn't fail to build, there just aren't any binaries to build for this arch. Due to the latter consideration, I'm inclined to go with Martin's suggestion. I've implemented this on git master, but only for "user mode"; in "buildd mode" it still results in a nonzero exit. Note: once buildd can parse the sbuild build status info as in my patch from last week, a zero exit would be perfectly OK for buildd to accept--it would get the "skipped" status from the sbuild output rather than "successful" or "failed". Regards, Roger -- .''`. Roger Leigh : :' : Debian GNU/Linux http://people.debian.org/~rleigh/ `. `' Printing on GNU/Linux? http://gutenprint.sourceforge.net/ `- GPG Public Key: 0x25BFB848 Please GPG sign your mail.
signature.asc
Description: Digital signature
