Jeff Trawick wrote, On 11/30/09 04:42 PM: > sunanda menon wrote: >> Jeff Trawick wrote: >>> sunanda menon wrote:
>>>> Please review the changes made at >>>> http://cr.opensolaris.org/~sunandam/6857014/ and let me know your >>> >>> Is there a separate CR covering the cflags change from -x04 to -x02? >>> That should be called out in the webrev. >>> >>> That's the only thing I'm concerned about, but I guess finding/fixing >>> that issue can be prioritized/resourced independently from getting >>> the current failures resolved. >>> >>> +1 from me if the issue of -x02 can be clarified: >> >> I'll have to do this as a possible workaround for now as the issue has >> been taken up at http://bugs.mysql.com/bug.php?id=49091. >> -xO2 is a tried and tested optimization flag used by the mysql >> community and also in the mysql.com version sources. > > No disagreement here on -x02 and that separate tracking of the > optimization issue. > > What procedural steps associated with this particular check-in will > track the fact that the optimization was changed for a separate reason > from the upgrade and DTrace support refactoring? There are now three CRs on the Sun Studio 12 compiler indication problems with staying on -xO4 (two relate to MySQL 5.4.3, which has same code base as 5.1) - and even -xO3, since I've seen that question asked. While we are still trying to work around 6892501 with a "local" reduction to -xO2, there are strong indications we should drop to -xO2. This might also affect MySQL 5.0, looking into that also. Regarding Jeff's final question, I will file a separate high priority bug on this if we need to drop the optimization level to keep issues apart, and we must then include that in the webrev. Would that satisfy all procedural requirements? -- Lars
