On Sun, 2005-07-10 at 17:18 -0500, Jason L Tibbitts III wrote: > >>>>> "NM" == Nicholas Miell <nmiell comcast net> writes: > > NM> And the answer to that question is: no, there isn't any way, > NM> besides noticing that something dies/behaves erratically/etc. with > NM> NPTL but not with LinuxThreads. > > I guess I just didn't understand the nature of the issue. Is the > situation simply that LinuxThreads and NPTL have the same API but > subtly different semantics? If so, then my question was pointless, > and only proper testing will show whether there are compatibility > issues. LinuxThreads and NPTL not only have the same API (which is a posix standard) but also the same ABI. In fact, by default RHL9 and later (including all Fedora Core releases and RHEL3 and RHEL4) use NPTL by default for *all* thread using apps, even when they were built against linuxthreads. You had to go through some hoops to make apps NOT use NPTL. Now the "breakage" people talk about is applications depending on bugs in linuxthreads that got fixed in NPTL.
Attachment:
signature.asc
Description: This is a digitally signed message part