Re: [gentoo-dev] Re: Gentoo peer-to-peer

[Jean-Michel Smith]

On Wednesday 24 July 2002 04:13 am, Alexander Gretencord wrote:
> On Tuesday 23 July 2002 23:14, Charles Lacour wrote:
> > I'm not sure there's any need for USE flags to identify the binary. If
> > two different people compile a package to identical binaries, what USE
> > flags they had are completely irrelevant. (Someone might compile a
> > package under another flavor of Linux, for example, in which case there
> > ARE no USE flags.)
>
> Well it's pretty obvious that different USE flags should result in
> different binaries but you need the USE flags so you know what's in the
> package. If I have a -X in my USE flags I surely wouldn't want to install a
> vim binary package that has X support in it.
>
> Should everyone be able to submit binaries ? I surely wouldn't trust them.

I really think, before we even consider distributing pre-compiled binaries, it 
makes far more sense to get the existing 'emerge rsync' and source tarballs 
working with a good P2P system.  This would fix the existing bottle necks 
with emerging packages and improve performance, and would be relatively 
straightforward.  All we need to make it work are GPG-signed ebuilds.

Binaries open a whole new can of worms, including subtle incompatabilities not 
just due to differing USE flags, but also do to (slightly) differing 
libraries, differing CPU and architecture settings in /etc/make.conf, etc. 
etc.

Frankly, I have no interest in downloading any binaries whatsoever (and doing 
so really strikes me as akin to having widespread, repeated sex with no 
protection in terms of the opportunity for infection by a trojan or Linux 
specific worm or virus), but P2P as a means to download source tarballs would 
be a great boon to Gentoo, and a hell of a lot easier to impliment than the 
distribution of binaries being described here.  Being able to emerge sync via 
P2P would be even cooler (but probably more difficult to impliment).

My 2 cents, for what it is worth

Jean.