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

[Final comments]Are we losing users to Gentoo?



Ok guys (and girls??(oh do i ever pray for girls)) 

I approached this list with the _IDEA_ of doing an i686 distro because
the main arguments that I AM HEARING (yes, in irc and the web) is that
they really believe that compiling all libs on your actual machine gives
you so much performance (enough to not use debian, and to use gentoo).

This is even reflected in the debian community, see some of this thread
before I got involved:
http://lists.debian.org/debian-devel/2002/debian-devel-200212/msg00374.html

Next, I tried to devise a solution to this whole problem that debian
could use. No, debian is not gentoo. Is gentoo adding to world entropy
in a negative fashion? probably a little. If they were smart, they could
have simply written a version of apt/dpkg in python to work very much
like apt-source...(but that is not the focus of this thread...). 

So anyway, I'm going to respond to the last comment on this thread:

On Mon, 2002-12-23 at 20:14, Scott James Remnant wrote:
> A quick google brings up lots of papers about the performance
> differences between different C compilers, but nothing about the
> benefits of compiling with different processor options using the same
> version as gcc.
> 
> Has anyone got some real-world evidence that this makes a difference? 
> I'm thinking about things like how many pages/second an Apache server +
> mod_perl can serve when compiled generically over when compiled with the
> "perfect" optimisations for the processor.
> 
> Scott

Well, it does. Users like the above link have seen improvements in
graphics apps. Moreover, in this theoretical i686 distrobution, libc
would be done in 686, which means that just from that, every package
benefits.

Next, there has been talk about this in the debian community before. I
googled, and here are some results:

http://www.geocrawler.com/archives/3/210/1999/9/0/2678785/ argues that
not everyone has a 686 chip now. Well  its three years since then, and
even MORE OF US DO. OK?!?!?!!?!??????????

Next, http://www.debianplanet.org/node.php?id=210 this link from 2001
discusses the actual performance that other platforms that have "a >
i386" release.

Finally, just another similar thread: 
http://lists.debian.org/debian-mentors/2002/debian-mentors-200210/msg00138.html


So thats that. Debian has been conservative on this issue. Too
conservative in my opinion. Discussions have been had on this topic for
over 3 years, as I have shown. Tracing back though THIS email thread, I
find many of the arguments weak, because they are simply reactions to
this suggestions. The two strongest reactions are: 
 -not everyone has > i386, AND i think this email does a good job of
knocking this bullshit and nowadays even less warranted conservatism
down.
 -that this will increase the size of debian mirrors.

To this second argument, I will give a response: who the hell makes you
think that every mirror has to take on i686??? A few servers would start
out serving this i686 distro, and it would be evident that these are in
great demand, and hence more i386 mirrors can switch over until the
demand is met. (See http://www.debian.org/mirror/ftpmirror , its a
document that addresses policy/practice on this issue).

So SOMEONE PLEASE step out there and start working on an i686 distro. It
will take a minimal time to port. And it will progress THE WORLD!!!




Reply to: