This is the mail archive of the
gcc@gcc.gnu.org
mailing list for the GCC project.
Re: Code Bloat g++
- To: pfeifer at dbai dot tuwien dot ac dot at (Gerald Pfeifer)
- Subject: Re: Code Bloat g++
- From: Joe Buck <jbuck at possibly dot synopsys dot com>
- Date: Thu, 17 Feb 2000 07:54:23 -0800 (PST)
- Cc: martin at loewis dot home dot cs dot tu-berlin dot de (Martin v. Loewis), NEELAKANTH dot NADGIR at sun dot com, gcc at gcc dot gnu dot org
>
> On Tue, 15 Feb 2000, Martin v. Loewis wrote:
> > Anyway, as Alexandre points out: this is debugging information, so
> > what is the problem?
Argh! I am astounded at how many C++ compiler developers make such comments.
I am familiar with a commercial project that is at a crisis point because
of the massive, massive size of debug information generated by a
well-known proprietary C++ compiler (literally gigabytes of debug
information for a single medium-to-large application, more than triple the
size of the previous release). Please let's not make g++ suck just as
badly.
Debug information is critical for the software development process.
It needs to be compact, meaning that measures need to be taken to avoid
redundancy.
Cool optimizations are sexier, but in real life developers will be doing
a lot more compiles with -g than with -O.
Gerald writes:
> When compiling template-heavy C++ code with -g, GCC generates huuuuge
> .s files which take 10+ minutes to assemble with Sun as, even on a fast
> UltraSPARC.
> It might be that GNU as is faster, but the size of .s files with debugging
> information on sparc-sun-solaris2.6 alone is prohibitive.
Does -fsquangle help?