On Thu, Nov 3, 2011 at 12:41, Delesley Hutchins <deles...@google.com> wrote: >> I wonder... how about disabling IPA-SRA when annotalysis is enabled? > > That would certainly make my life a lot easier -- IPA-SRA is not > really compatible with annotalysis. The only problem I can see is > that people may be unhappy if turning on annotalysis makes their code > runs slower. We need a good way to communicate the issue to users. > Is there a way to print a note that will be reported by the build > system, e.g. "Thread safety analysis enabled, some optimizations are > disabled?"
Certainly, you could call inform() with a diagnostic. Though, it's true that analysis should not affect optimization levels. I think I prefer making the analysis tolerate the optimizers than affecting code generation. > >> Also, why not turn OPT_Wthread_safety on by default? > > We certainly could. However, it generates extra warnings, which may > break the build for some users. Which would users prefer: knowing > that annotalysis is suppressing warnings, or having to update their > build files to pass in an extra flag to turn off the warnings? What is the effect of your patch on annotalysis warnings? Does it produce false negatives? If so, I think we should turn this warning by default. I'd rather break the build than let a false negative through. Diego.
