Patches for trivial bugs sitting in bugzilla -> trivial patch policy?
manisandro at gmail.com
Thu Jun 26 23:27:23 UTC 2014
On 27.06.2014 00:55, Kevin Fenzi wrote:
> On Fri, 27 Jun 2014 00:39:52 +0200
> Sandro Mani <manisandro at gmail.com> wrote:
>> So the thing that should be avoided IMO is not defining well enough
>> how the procedure should work, to avoid getting swamped with patches
>> which require additional work to apply. The requirement to fill out
>> post a "New Package Request" style form to the bug should allow most
>> of the work to be scripted, and ideally the proven packager would
>> just need to look at the patch, and if happy, fire the script.
>> So maybe such a policy could look something like this . Contrarily
>> to what I wrote initially, I think it might actually make sense to
>> allow also non-packagers to file such requests, and it would provide
>> another way of showing off experience which will eventually lead to
>> one getting sponsored.
> A few comments:
> - Not sure 'trivial' really covers the things you list in examples.
> FTBFS could be more than trivial depending on the patch. Perhaps the
> entire thing could be 'simple patches' or 'easy patches'
Right. So the list Yaakov Selkowitz posted elsewhere in this thread
gives a good set of possible FTBFS causes. I'm not sure all FTBFS should
be covered by such a policy, but if it is only a matter of fixing bad
BRs, format security issues or makefile issues, these should fall under
the accepted category. If the FTBFS fix involves porting for API/ABI
breaks, then IMO that does not fall under this category. But yeah,
simple might be more appropriate terminology.
> - 'flags' are a pain to get bugzilla folks to add and manage. How about
> a blocker bug instead? Then interested maintainers could simply cc to
> the blocker bug to notice when new things are added. You just add the
> patch bug to the blocks of the tracker.
I guess this would work just as well!
> - Should this be rawhide only? That would avoid 'trivial' patches that
> cause a problem from affecting users that aren't as able to debug
What about rawhide for all and stable releases for packagers only?
Draft updated accordingly.
More information about the devel