Hunting for performance regressions in libstdc++

Paolo Carlini pcarlini@suse.de
Fri Jun 17 13:03:00 GMT 2005


Howard Hinnant wrote:

> Simply doing a before and after check still allows you to conform to
> the C++03 standard, while not causing a significant performance
> regression against your customers.  Why I care:  I'm currently
> counseling a customer in exactly this position.

Agreed: as long as we have *both* the checks we are fine. Now that we
are close to the final wording this makes sense: before Lillehammer I
had no idea what shape would take the final resolution and having both
checks it's a rather more invasive change. Also, we had complexity
concerns, which you reported in the paper.

Thanks,
Paolo.



More information about the Libstdc++ mailing list