[Date Prev][Date Next] [Thread Prev][Thread Next] [Date Index] [Thread Index]

Re: RFS: ballview : new package version

On Thursday 25 January 2007 15:55, Andreas Moll wrote:
> > Sorry if I missed something obvious, but why not calling it directly
> > from the sources instead of duplicating it in /debian ?
> All right, I will call the script script directly.
And if it is in the upstream source already, then there will be no issue with 
the diff's permissions when creating new files.

> > > I will reduce the set of supported architecures in the control file to:
> > > Architecture: i386 amd64 ia64 sparc
> > > I guess that a package does not have to support all arches, or am I
> > > wrong?
> >
> > The policy says that restricting architecture support should only be
> > used when the program is not portable. I think that is not the case of
> > BALLview. There is at least one more architecture on which it would be
> > definitely useful: powerpc. For slow architectures on which it does not
> > make sense to do some structural biology, it is still the responsability
> > of their lead developpers to decide that a package should not build. So
> > if it only requires to fix the configure script to support all Debian
> > arches, I would say: go for it!
> The problem is, that my application is based on roughly
> 500,000 lines of code,
> that we developed in the last years. I can not guarantee
> that it will work on
> the other platforms, especially because we are dependent
> on external libraries,

Since the libraries are built automatically through the Debian build daemon, 
you can to a certain degree expect a homogenous and consistent behaviour 
across all platforms. This is what renders Debian fairly unique.

> that may show unsuspected behaviour.
If a library shows unsuspected behaviour, then it is not your bug but one of 
that library. The inter-platform consistency of Debian may be one of the best 
ways to locate an issue.

> What is even worse: I have no computer available to test
> our software for example
> on a power pc architecture. What happens if my package
> fails to build on such a machine
> and I get a  critical "Fails to build"error?
Once you are a Debian developer, and I hope that you will apply to become one, 
you have access to machines allowing you to investigate that issue yourself. 
However, a bug on a platfrom that nobody uses will remain unnoticed.  You 
already made the acquaintance of Charles who is using the PowerPC and cares 
for your package. I myself looked into packaging your work quite a while ago 
(i386,x86_64) and am likely to have a look - just ask.  You may be interested 
to say hello to the Debian-Med community of Debian 
http://www.debian.org/devel/debian-med/ that features many list subscribers 
with a background in structural biology - be it from the wet or the dry lab.

A major reason to use Debian IMO is its community.

Best regards


Attachment: pgplfLTbByD27.pgp
Description: PGP signature

Reply to: