[aur-general] TU application for Thomas Dziedzic

[Xyne]

> Hello,
> 
> I am the former maintainer of the sage-mathematics package in AUR. At
> the time I asked for inclusion in community it was not packaged in the
> install file, but using a "regular" PKGBUILD. This method stopped
> working later. The main reason for not including it was the mentioned
> duplication of packages. I believe this issue would be difficult to
> solve. The packages sage uses are mostly altered in a way to make them
> integrate into the "distribution".  I do not know if there is a complete
> list of changes they made, even a complete list of packages they
> changed. 
> 
> I was sometimes thinkig about a arch-sage-distribution, that consists of
> the arch core packages, sage as base system, and that takes care of not
> installing packages that are included in sage from the arch repos.  I am
> not sure if this is technically doable or advisable. One would have two
> package managers, pacman and spkg.
> 
> Regards Stefan

Thanks for the clarification.

I don't think the duplication itself should be an issue. I agree that
it is somewhat a waste of diskspace, but overall I would not expect it
to be significant. Bandwidth might be an issue, but if the Sage package
fully provides many of its internal packages then users who intend to
use Sage would not need to download those other packages. This might
require some initial work on the PKGBUILD to make sure that the
provided packages behave as expected and can be used to satisfy other
dependencies, but that should not require continuous effort.

Ideally the package would simply use external deps and optdeps but I
understand that the alteration of packages makes this difficult. I
would prefer the upstream devs to work together to provide the
necessary interfaces by default so the Sage devs wouldn't have to hack
everything together. Hopefully this will change in the future.

Given what we have now, the question is whether the extra diskspace is
worth the total amount of compilation time that users currently incur
for building sage from source. Given the package's popularity, I'm
inclined to say it is.

If Thomas' application is accepted, he can even occasionally nag
upstream to clean up their build scripts and modularize the package.*


Regards,
Xyne


* New Arch Linux position: UN - upstream nagger :P