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

Re: RFD : libg++/gcc/egcs upgrades needed for libc6



jdassen@wi.leidenuniv.nl wrote:

> Both egcs's cpp and gcc 2.8's cpp have problems that show up with
> imake-generated Makefiles (#17307). These will have to be solved before we
> move away from gcc 2.7.2.3 as the standard compiler.

So I've heard.  Unfortunately, there's not a lot I can do about this directly.
I'll browse the egcs list archives and see if anybody has mentioned this.  For
all I know, this is a feature, not a bug.  :-P


> I did the non-maintainer release of 2.8.0 in experimental.  I agree that in
> the current situation, egcc is the preferred compiler. I do think however we
> should have a gcc 2.8.0 package available, and encourage maintainers to test
> their packages against it. Given the problems, experimental is probably the
> best location.

I've just now uploaded the new egcs package, which provides our standard
gcc/g77 compilers.  Remember that gcc 2.8.0 is the second-class citizen now;
take that into account when making a new experimental package.  You'll have to
rename the binaries in /usr/bin (gcc28 perhaps?) and set the /etc/alternatives
priority lower than 20.  (Take note of bug #17468.)


> I'd like to encourage you to build gcc and egcs fully bootstrapped (#17154)
> and with an additional debian/rules target for running the testsuite (gcc
> doesn't have an integrated one yet; egcs does, but requires a "build of the
> day" dejagnu).

The m68k people would probably lynch me.  I don't see much value in building
the stage 3 compiler; and since the right things happen even without an
explicit "make bootstrap" (f771 and libstdc++ are both compiled with the
just-built gcc rather than the system gcc) I won't double the build time by
doing so.

Since I don't have (and have no intention of installing) dejagnus on my system,
I couldn't test any "test" target I added to debian/rules.  If you want to run
the test, just cd to builddir and type "make test" or whatever.  Is that really
so hard?

--Galen



--
TO UNSUBSCRIBE FROM THIS MAILING LIST: e-mail the word "unsubscribe" to
debian-devel-request@lists.debian.org . 
Trouble?  e-mail to templin@bucknell.edu .


Reply to: