[repo-coord] Packaging Process - PLEASE READ

[Rudolf Kastl]

Am Fr, den 21.05.2004 schrieb Rudolf Kastl um 4:47:
> Hello.
> 
> I have thought a bit about it and i think there should be a general
> process when building new packages.I have not a complete process chain
> (can do that professionally in tools like aris if you are really
> interested). I will go and put that on my page if its complete.
> I am working to extend this document so i would be happy about your
> ideas. I think we have to get to a generalize some things without beeing
> too limiting. with coordination of so many packagers like currently i
> think its not possible to have clean interfaces between the repos
> without a few general rules. 
> 
> 
> 
> A. The Source
> 1. we need to specify checks for the source code. Means checking if the
> source is alright or no in terms of malicously changed.
> Taking already tested source of other distros makes sense but its still
> adviseable to verify again.
> (rudolf kastl/newrpms)
> 
> 2. before you even touch the source check if its already packaged. dont
> override already existing packages... that happened way to often
> already.
> (rudolf kastl/newrpms)
> 
> 
> B. The Spec
> 1. perl and sed should be avoided, clean patches are preferred, unless
> really necassery. 
> (rudolf kastl/newrpms)
> 
> 2. buildmacros should all be defined on top of the spec. everything else
> is considered broken. there can be exclusions when its not possible to
> act otherwise in a special case. 
> (rudolf kastl/newrpms)
> 
> 3. other major distros should be checked for patches that might give
> crucial fixes. not all send everything upstream.
> (rudolf kastl/newrpms)
> 
> 
> 
> 
> C. The Buildprocess
> 1. building should always happen in a non root environment (thats an
> understatement). Even if its chrooted and works fine, theres no way one
> could verify if they also build as non root.
> (rudolf kastl/newrpms)
> 
> 2. other major distros should be checked for patches that might give
> crucial fixes. not all send everything upstream.
> (rudolf kastl/newrpms)
> 
> 3. it has to be verified that the rpm installs clean and runs fine
> before its released. That includes basic functionality checking.
> (rudolf kastl/newrpms)
> 

replying to my own post real quick since i think theres a general
misunderstanding here...

what i wanna do here is to collect valuable ideas for the optimal
process...
for some things that might look scary to you there are quite simple
solutions...

regards,
rudolf kastl
> 
> 
> 
> 
> _______________________________________________
> repo-coord mailing list
> repo-coord at atrpms.net
> http://lists.atrpms.net/mailman/listinfo/repo-coord