Date:	Fri, 11 Jun 1999 10:08:28 +0300 (WET)
From:	Andris Pavenis <pavenis AT ieva01 DOT lanet DOT lv>
To:	Chris Matrakidis <cmatraki AT ee DOT ucl DOT ac DOT uk>
cc:	djgpp-workers AT delorie DOT com
Subject:	Re: egcs-19990602 (gcc-2.95 prerelease) binaries for testing
In-Reply-To:	<199906110159.CAA09987@melchard.ee.ucl.ac.uk>
Message-ID:	<Pine.A41.4.05.9906110947250.45982-100000@ieva01.lanet.lv>
MIME-Version:	1.0
Reply-To:	djgpp-workers AT delorie DOT com
X-Mailing-List:	djgpp-workers AT delorie DOT com
X-Unsubscribes-To:	listserv AT delorie DOT com

On Fri, 11 Jun 1999, Chris Matrakidis wrote:

> > I have built 2 June snapshot of egcs for DJGPP and made binaries of 
> > C and C++ compilers are available for testing.
> > 
> > http://www.lanet.lv/~pavenis/gcc295snap.zip  - C compiler
> > http://www.lanet.lv/~pavenis/gpp295snap.zip  - C++ compiler + libstdcxx.a
> > http://www.lanet.lv/~pavenis/lgp295snap.zip   - libgpp.a built with this compiler
> > 
> > Please test them (I'll do the same).
> > 
> 
> I did some quick tests of the C compiler compiling a moderate size project,
> and everything seems to work fine, with the generated code beeing impresively
> quick compared to gcc 2.8.1 and egcs 1.1.2.
> 
> Some minor points from this test:
> 
> 1) Why is libc_p.a required when profiling? From the spec file:
> 
> *lib:
> %{!shared:%{g*:-lg} %{!p:%{!pg:-lc}}%{p:-lc_p}%{pg:-lc_p}}

Thanks.

I myself have not tested profiling. Seems that after all changes in djgpp
related configuration files not only by me definition of LIB_SPEC has
disappeared from djgpp.h. Perhaps I should put it back

> 
> 2) Shouldn't the directory of the compiler be ${DJDIR}/lib/gcc-lib/djgpp/2.95
> instead of ${DJDIR/lib/gcc-lib/djgpp/gcc-2.95 for consistency with version
> 2.8.1?

It's gcc-2.95 and I don't see any special reason to change it. For
egcs-1.1.X I had to modify version string as 2 dots there are not
acceptable. With gcc-2.95 there are no such problems so I think there is 
no need to change anything. Also in Linux I'm getting directory 
/usr/lib/gcc-lib/i586-pc-linux-gnu/gcc-2.95/  

> 
> Finally, wrt the alignement requirements discussion, this version generates
> several .p2align 5 statements in the .S file. Therefore, I continue to suggest
> that the default section alignment should be 32 bytes. 
> 

I have not looked at these things much. Maybe it's best to ask related
questions in egcs mailing list (egcs AT egcs DOT cygnus DOT com). I'll see answer
also there.

Anyway I think that defaults for i[3456]86 should not be so bad.

Andris

- Raw text -