On Tue, May 10, 2011 at 08:44:23AM +0200, Adrian Schröter wrote:
Am Tuesday, 10. May 2011, 00:04:10 schrieb Juergen Weigert:
On May 09, 11 23:00:01 +0200, Kay Sievers wrote: ...
People should not be allowed to throw their nitpicking issues that way at other people. They could use *their* own obs project to build the distro with *their* rules, and then fix the packages themselves.
Guess, that would be fine treatment for their personal issues, instead of annoying everybody else.
I take it, you are royally annoyed, and just want me in another universe.
I have heared earlier complaint about unfair rejects, but so far no good suggestion was made how to improve things. I can easily change all rejects into delegate to human -- this provokes delays instead. As of today, the manual review queue size is 170 submits.
Other options I could think of:
- implement the checks as rpmlint, so you get to see them earlier.
or into the source validator package, so it complains already at commit time. The earliest possible point.
I would say rpmlint is better and in earliest phase ;-). The source validator is for cases rpmlint cannot handle. Plus it is almost impossible to build package in Factory without rpmlint (I expect our autobuild checkin team reject all packages with BuildIgnore rpmlint). On the other hand it's trivial send package without check of source validators.
- silently change spec-files during check-in.
nope, this is what we want to get rid of. We could integrate it in the upcomming spec file formater of osc though. But I doubt that it would have all required data to run there, right ?
Yes, please do not modify spec file at all! Having strict checks with a description how to fix the problem is the best approach. Regards Michal Vyskocil