[packagers] qa, workflow, here we go again

Dag Wieers dag at wieers.com
Tue Oct 26 16:05:55 CEST 2010


On Tue, 26 Oct 2010, David Hrbáč wrote:

> Dne 19.10.2010 13:51, Dag Wieers napsal(a):
>> On Tue, 19 Oct 2010, David Hrbáč wrote:
>> Not possible until we have a central buildsystem. I am not planning to
>> commit something before I know for sure it builds for all distributions
>> (at least those that are possible).
>>
>> And I don't have the time, nor the hardware to separate the environments.
>
> Is it really so hard to test it within mock environment as we do so?!?

Yes, if that means waiting for the additional builds to complete and 
having to change my buildsystem. The easiest solution is a central 
buildsystem so that I have the same (and no more) rights than any other 
packager.


> Right now I see another clamav update within rpmforge channel and
> there's no commit to clamav. And you know, there's a lot of people
> offering HW/Virtual to you, includig myself...

Let's be honest here, I am not interested to spend my (personal) time at 
the moment to fix the fact that I forget to commit the changes to SPEC 
files from time to time after the builds work.

The reason why I don't commit right away, because I don't work serialized 
_and_ my tree contains unfinished work. So simply committing all changes 
does not work either. Selecting which changes I want to commit is also 
very slow due to subversion (and/or my system).

If we have a central buildsystem, I could work exactly the same way as I 
do now, except that for any official build I am required to commit my 
packages. So the requirement to commit is the incentive, rather than an 
extra task I need to remember to do at some time after the build has 
worked fine (and no additional rebuild is required).

Doing additional mock builds does not help in reducing my time, instead I 
would sacrifice time and require more time in the future when building. 
Hence I loose.

-- 
--   dag wieers,  dag at wieers.com,  http://dag.wieers.com/   --
[Any errors in spelling, tact or fact are transmission errors]


More information about the packagers mailing list