About missing reviews

[Thorsten Leemhuis]

On 01.01.2007 20:48, Kevin Fenzi wrote:
> On Mon, 01 Jan 2007 12:56:14 +0100
> fedora at leemhuis.info (Thorsten Leemhuis) wrote:
>> The rules we used today are also slightly different then they were
>> one, two or three years ago -- thus your argument holds true for all
>> our packages, not just those from fedora.us (albeit the rules that
>> were used might be bit more different). That why I wrote yesterday
>> somewhere else on this list that we sooner or later might need to
>> re-review each package in CVS. But I don't think we have the manpower
>> for that (or let's say: we IMHO have more important things to do
>> ATM), especially with the "let's review all the core stuff when it
>> gets merged into the extras framework for F7" on the horizon.
> Yeah, I think it would be a wonderfull world if we could re-review
> every package on every cycle or something, but thats not gonna
> happen. ;) 

Well, there is a reason behind the sentence "the oldest packages get 
visited first" ;-) -- e.g. I suspect we are not able to visit everything 
during one devel period.

>> I'd suggest this: for the devel period towards F8 build a "Re-review
>> SIG" (or let the QA sig handle it) that just goes trough most of/all
>> the packages in CVS; the oldest packages get visited first. 
> [...]
>> I'd even
>> say those SIG members should get allowed to fix everything directly
>> in CVS even if the package is owned by somebody else. That eventually
>> could speed up the handling of the effort a lot; we just need to make
>> sure the packager sees what was changed and gets noticed *why* it was
>> changed to educate them.
> I don't think thats a very good idea personally... If a re-review of a
> package shows problems, why not file them as bugs?  The maintainer can
> explain why they might not be real bugs, etc.. also as a bonus this
> could show a package that could be orphaned due to lack of maintainer
> response. 

For everything none obvious filing bugs should be mandatory. But is 
there is still a manual
> Requires: python-abi = %(python -c "import sys ; print sys.version[:3]")
somewhere (¹) why not fix it directly in cvs? Bugzilla  is IMHO to much 
overhead for such easy fixes. Hint: But the real solution for those 
obvious "bugs" is probably to write a check-script that files bugs 
automatically.

CU
thl

(¹) -- from a extras checkout devel:
$ grep python-abi devel/*/*.spec | wc -l
87