On Apr 5, 2011, at 4:13 PM, Per Øyvind Karlsen wrote: >> > Hm, okay, so better obviously needs to be done. >
Have a go at fixing if you wish. The core problem is this: rpmbuild can take either *.spec or *.src.rpm as inputs. noobs complain loudly (because n00biness is embarassing) when rpmbuild doesn't detect and exit gracefully when args are misused. There's also the general problem of random input rpmbuild -ba /dev/random with segfaults (though I doubt there's any serious security issue, rpmbuild is usually run on machines protected in other ways, and I'll start fuzzing rpmbuild for buffer overrun's and such when gcc and make and binutils and other commonly used tools on build systems start being routinely subjected to fuzz-busters ...). > For what currently is though, is it supposed to be broken or...? > I truly care only about not being bothered with silly mindless RFE's about error messages and exit codes. Its not like rpmbuild can be changed in any meaningful way for zillions of reasons called "build recipes" and "package monkeys" who expect rpmbuild to continue working -- good/bad/ugly -- exactly as its always done. See the %_initrddir thread for a perfectly common bug report (not hard to "fix", just multiple definitions to handle multiple dialects that unccordinated vendors have chosen is, well, there's no win there either ...) What made me squawk is that you cross-dressed (your obvious fix) as if it were an enhancement to support utf8. That isn't the issue with the stoopid n00by "sanity check" needs to prevent confusing *.spec and *.src.rpm inputs as args to the same executable. > -- > Regards, > Per Øyvind > ______________________________________________________________________ > RPM Package Manager http://rpm5.org > Developer Communication List rpm-devel@rpm5.org ______________________________________________________________________ RPM Package Manager http://rpm5.org Developer Communication List rpm-devel@rpm5.org