Subject: Re: next prerelease available
From: Dale E Martin (dmartin@cliftonlabs.com)
Date: Mon May 14 2001 - 08:34:13 MDT
> This is a good place to be. The RH compiler breaks too many things to
> worry about. Anyone generating (C++) binaries for RH7 will find they
> don't run any where else, not even any other RH releases. IMHO, there is
> no reason to go out of your way to make certain that those binaries are
> possible to generate in the first place...
OK, I think this is the route to go too. Even supporting these two
compilers appears to be a bit of a nightmare - specifically, there are
incompatible changes in libstdc++ that are going to require either heinous
macros or configure tests to deal with...
Also, g++-3.0, at least in prerelease form, is _really_ slow. My Celeron
533 machine (256M of RAM) can compile SAVANT in around 36 minutes with g++
2.95.2... With the version 3.0 prerelease I have, it's taking more like 96
minutes, and that's not even a complete build!
> As a practical matter, RH7 (almost certainly) shipped with older
> libstdc++ versions included for compatibility, so if the binaries were
> build elsewhere, it should be possible to run them even if it requires
> LD_PRELOAD tricks (but certaintly they changed the version numbers, i
> hope!).
Right, I'm more concerned about source/build compatibility. Currently
there are link problems with our system on that platform, which are hard to
debug via email exchange.
> I will compile up this prerelease tonight, on a G4 PPC box running
> SuSE6.4.
How do you like Linux on PPC? I've been tempted to try it occasionally.
Later,
Dale
-- Dale E. Martin, Clifton Labs, Inc. Senior Computer Engineer dmartin@cliftonlabs.com http://www.cliftonlabs.com pgp key available
This archive was generated by hypermail 2b25 : Mon Mar 18 2002 - 13:00:02 MST