[Pacemaker] Pacemaker 1.0 and compiler optimization

Rainer Weikusat rweikusat at mobileactivedefense.com
Thu Sep 29 13:42:09 UTC 2011


Andrew Beekhof <andrew at beekhof.net> writes:
> On Tue, Sep 27, 2011 at 1:47 AM, Rainer Weikusat
> <rweikusat at mobileactivedefense.com> wrote:
>> Is there a specific reason why compiler optimization is disabled (line
>> 1323 in configure.ac, '-ggdb3 -O0') when building pacemaker?
>
> Something we inherited from heartbeat I guess - possibly so that stack
> traces were sane.
> Not sure we should be forcing it though

I actually have an opinion on that, namely, when (most of) this code
has never been tested with optimization enabled, which is likely the
case, enabling optimization for anything even remotely 'production'
would be a very unwise descision. But doing this for some kind of more
experimental release should be considered: Leaving the issue of faster
execution speed, especially for subroutine calls, aside, compiling the
1.0.9.1 based sources of the current Debian package with -02 will
reduce the total size of the generated binaries by about 12.17%
(1856.35K[*] vs 1630.35K).

	[*] 1024
        
        




More information about the Pacemaker mailing list