This is the mail archive of the
gcc-patches@gcc.gnu.org
mailing list for the GCC project.
Re: [v3] --enable-pch
- From: Richard Earnshaw <rearnsha at arm dot com>
- To: Benjamin Kosnik <bkoz at redhat dot com>
- Cc: gcc-patches at gcc dot gnu dot org, roger at eyesopen dot com, Richard dot Earnshaw at arm dot com
- Date: Mon, 07 Jul 2003 13:32:22 +0100
- Subject: Re: [v3] --enable-pch
- Organization: ARM Ltd.
- Reply-to: Richard dot Earnshaw at arm dot com
>
> Roger was the last in a long line of people who've politely asked me
> to do this.
>
> This allows --disable-pch to turn off attempts to use pre compiled
> headers in C++ runtime land, as should have been done from the very
> beginning. I'd hoped that the persistent failures with PCH and -fPIC
> etc etc would motivate quick patches: this hasn't happened, and it's
> unfair to ask people who are trying to keep ports above water to do
> hacky things to testsuite_flags.in to work around this. Note, that PCH
> is still enabled by default: however, once it's apparent that it's not
> working, this flag will allow people to side-step the problem
> gracefully.
>
> Recently I've had to move back to pre-PCH g++'s do do debugging, so
> this was starting to be a pain in the neck for me as well.
>
Yay! That cuts testing time ('time make -k check') on my poor
memory-starved shark (32M) by more than 20%!
user system elapsed
2003/06/28 28971.91 16151.89 82516.12 (PCH enabled)
2003/07/02 40920.70 15537.30 64758.78 (PCH disabled)
More importantly, the timeouts during testing are gone, so the test
results look much more solid:
Before:
=== libstdc++-v3 Summary ===
# of expected passes 1477
# of unexpected failures 13
After:
=== libstdc++-v3 Summary ===
# of expected passes 1523
# of unexpected failures 2
Note, that machine doesn't do anything other than testing, so the cut in
elapsed time is a real improvement in the compiler, not just a side-effect
of other activity.
R.