About merging egcs and pgcc...

[Hannu Koivisto <Hannu dot Koivisto at vvf dot fi>]

Greetings,

Just a thought that me and a friend of mine started wondering:
is there some particular reason why merging pgcc improvements
into egcs hasn't been considered? (Of course, it may be that
I've missed such discussion as I have followed this list only
for two months or so.) Are they perhaps implemented in some
fashion that isn't suitable considering the future plans of
egcs? Or perhaps pgcc team wants to work with a separate version
and not together with other egcs developers? 

After all, egcs is an experimental compiler and it's pretty much
on the bleeding edge in terms of C++ features/standard but Ix86
performance could be better, as pgcc has shown. Or perhaps egcs
team is waiting to fully utilize haifa scheduler etc themselves
or simply wants to do things in a different way (?) Although it
doesn't seem so, perhaps there's a hidden overlap I'm not aware
of. 

The reason I'm asking this is that, well, egcs-project, in my
opinion, has a lot of potential. As a consequence, there is a
lot of support interest for it. For example, I'm personally
looking for possibilities to provide mirror site for at least
Mr. Khan's Win32-port (I saw him asking for a mirror in his
www-page) and of course for everything else in case I'm allowed
to. And if I'll learn enough about egcs internals, adding
Codeview debuginfo support might be an interesting project to
get support for different debuggers and tools like Purify in
Win32 platform.

Having just a single The Experimental Compiler :) would clarify
this picture, in terms of maintaining installations (for example
I'm now keeping both egcs and pgcc installed on a same machine,
which isn't that hard but not very convenient either especially
when compiled-with-either-one libstdc++ is shared between them
(also shared files in general complicate keeping these both
compilers under Debian package system a bit)) and reporting bugs
(you always have to examine carefully that you don't submit
pgcc-specific problem to egcs developers and vice versa). In
addition, it would be sort of a mental stimulus to know that you
are supporting The kick-ass development compiler :)

I'd like to emphasize that I'm asking this just because I'm
curious -- I know egcs developers are already busy enough and
besides, I don't enough egcs internals enough to be in a
position to suggest anything.

Just wondering,

//Hannu