Re: Packaging of static libraries
-----BEGIN PGP SIGNED MESSAGE-----
Hash: SHA1
On Thu, Apr 14, 2016 at 02:57:00PM +0200, Vincent Danjean wrote:
> > If users have such specialized needs, I think it is not only reasonable that
> > they build their own versions of their libraries; I expect them to prefer that.
> > So we should make that as easy as possible. I can imagine that Gentoo also
> > seems attractive to them, and it may be a good (or even better) solution. But
> > as Debian, I think the best we can do to help them is to make it easy to build
> > our packages from source with custom build flags. I'm guessing that we're
> > probably talking less than 100 people in the world, maybe less than 10, that
> > need this. It makes no sense to put a package in the archive just for them.
>
> I do not understand where your numbers come from.
I understood that this was a very specialized thing that some people in
research institutions would want. So I guessed a few people at Cern, for
example. There aren't that many projects that are running for months. But
it's still a wild guess, and I'm happy to believe you when you have better
numbers. How many people do you expect need this?
> I also know lots of HPC clusters installed with Debian. If Debian
> choose to favor safety wrt to performance (instead of trying to find
> a good compromise as currently), it will probably loose some users.
It's always a compromise, and you were talking about people who don't want
that. Those people will need to build their own libraries.
> And no, admins (and most users) do not prefer to recompile software
> instead of using the installed one (some users do not have enough
> computer science skills to do it and some admins are already
> overloaded and will not want to manage a derivative distribution)
It's all about priorities. If having optimized libraries is very important,
they will train someone to do that. If it is not worth the training, then
that's fine, but then it obviously isn't so important.
> > And changing the default compiler settings to fit their needs makes even less
> > sense.
>
> However, it already occurred : we compile by default with -O2, not
> with the compiler default (no -O options). Until now, the Debian
> project seems to agree that this is a good tradeoff between
> optimization and "code correction".
That's not what I meant. You seem to suggest that we should compile for
maximum performance, at the cost of security, because some people want that.
Or maybe that we offer both options, I'm not sure what you propose. My point
is that we should not do that; we decide on the compromise, and if it's too
much towards security and away from performance for some people, they need to
get a different distribution or compile things themselves. I think we should
make it easy for them to do that.
But I understand that you think the prospect of a small group of people leaving
us over this is unacceptable, and we should try to avoid it even at high cost.
I disagree. It's nicer if they want to keep using Debian, and as the universal
OS we obviously hope that our system works for everyone. But it's not the end
of the world if a small group of users doesn't think so.
To clarify that: being the universal OS means to me that we have a good way to
use Debian for every situation. It doesn't mean that every user thinks Debian
is the best solution for them.
Thanks,
Bas
-----BEGIN PGP SIGNATURE-----
Version: GnuPG v1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=TxnM
-----END PGP SIGNATURE-----
Reply to: