> I see. > > Well is the LSB going to force distribution change? Do we > want to run the > risk of breaking existing code? > > This sounds like a policy decision.
Yeah. There's a little more to it that I just deduced. glibc tried to fix this a long time ago, the fix was in 2.2.5 (went in June 2001), although it wasn't quite correct. The 2.2.5 fix was to return "base + incr" which would be invalid if "base + incr" was too big - didn't account for bounds truncation. The extra LSB fix was to return the result of getpriority instead, thus making sure it returns what the niceness actually got set to, instead of something that might be more than that. I think that fix shouldn't be controversial but I don't see it in the glibc cvs. An extra "got it wrong" part was that apparently this fix didn't propagate out to a Linux build, whether that was intentional or not, I don't know. That problem was corrected by the other bit of the nice patch, which went into glibc's cvs tree on March 2 of this year. The policy decision is pending. Mats
