Re: Newer vegastrike : can we package that big ?
Hello !
[replying to this particular mail, but in effect to the whole thread]
Ivan Vucica wrote:
> Perhaps someone should bug upstream to consider producing a smaller
> version of the dataset. I'm sure they could strip some quality off of
> textures, and kick some content out to produce VegaStrike-lite. Or
> you, Vincent, could do that :)
>
> We could have:
> vegastrike-data
> vegastrike-data-lite (provides: vegastrike-data)
> vegastrike-data-full (provides: vegastrike-data)
>
> vegastrike-data-full would be separately hosted (as suggested above),
> perhaps by upstream following the
> if-they-can-host-1gb-they-can-host-2gb-too logic" ^^
That's a good idea, except for the part 'Or you, Vincent, could do
that' ;-)...
Anres Mejia wrote:
> The idea I'm referring to is that the orig source could contain only a "dummy"
> file and the upstream source is actually downloaded from somewhere into
> debian/tmp, unpacked, and built from using plain old debian/rules.
This is calling for a nightmare of FTBSes : if I remember correctly,
the grid that rebuilds every once in a while the archive is not
connected to the network - that will be a RC bug everytime. And people
should be able to rebuild a package from the source on a CD (without a
network connection).
I must admit that I currently don't have the courage to spawn a
discussion about sourceless packages, even if that could be a way to
alleviate the load of the mirrors.
I think I will finish the binary package in a first step, as it works
reasonably fine with old data, and upload it as is. Then, I'll probably
strip the music, which will lead to a source tarball of 330MB (which
will also be the Arch: all package generated from it). That is big, but
reasonable. This does not solve the installed-size problem, but that
probably can wait.
Cheers, and thanks for the input !
Vincent
--
Vincent Fourmond, Debian Developer
http://vince-debian.blogspot.com/
-- pretty boring signature, isn't it ?
Reply to: