Re: The libgcc.so for glibc initiative

[Mark Kettenis <kettenis at wins dot uva dot nl>]

   Date: Thu, 13 Jul 2000 15:44:18 -0700
   From: "H . J . Lu" <hjl@lucon.org>

   On Thu, Jul 13, 2000 at 05:13:50PM -0500, Mark Brown wrote:
   > 
   > 
   > "H . J . Lu" wrote:
   > > > While libgcc.so may be a per-platform issue, I do not see why we cannot
   > > > work with/educate the appropriate gcc maintainer on this; as long as
   > > > part of the education includes why we need a libgcc that meets *all*
   > > > of the glibc needs.
   > > 
   > > Last time when I tried it, they didn't wany any solution which only
   > > works for glibc. They want a solution which works on all targets gcc
   > > supports. That is why I said the solution they chose would fall far
   > > short for glibc.

What's wrong with requesting a solution that isn't Linux/glibc
specific for a problem that isn't Linux/glibc specific?

   > I had a discussion on this point with another gcc SC member,
   > David Edelsohn. He is of the opinion that looking at libgcc as a
   > per-platform item is a valid viewpoint, and agrees that maybe
   > there needs to be a version done for gibc. He's not the only one
   > who feels that way -- don't know if he represents the majority or
   > not.

   I am very glad to hear that.

Looking at libgcc as a per-platform item may be a valid viewpoint, but
certainly not the only one.  There's a huge part of the current libgcc
that's pretty much per-architecture.  There's also a lot of C++
support code for which I cannot find a reason not to share among
platforms.

   > 
   > this leads me to believe that if we approach this in a non-confrontational
   > matter we might get what we need in this way.

   The main objection comes from Jeff Law. Here is his typical view on glibc
   and Linux in general:

   http://gcc.gnu.org/ml/gcc/2000-07/msg00376.html

And what's wrong with that?  `All the world's not Linux' sounds like a
good attitude for a compiler that already works on a lot of platforms
and should be portable to others without much effort.

   He doesn't start thinking this way just a few days ago. I learned it years
   ago. And I think he will object anything I propose for glibc/Linux. Maybe
   someone else besides me and Ulrich should push the issue with a different
   approach. However, we cannot just wait here for something to happen in
   the meantime.

Please keep your personal grudge against Jeff out of this HJ.

Mark