On Fri, 15 Jul 2005, Eric Boutilier wrote: > And I still think calling POSIX "a key UNIX/Linux industry standard" is > a very good reflection of reality. > > > I actually fail to see a Linux industry standard in POSIX; > > Well at a minimum, it's certainly key to much of the development of > POSIX-like Linux distro standards and other Linux standards such as > those coming from LSB and OSDL. When Linux distro developers and > LSB/OSDL/etc. explicitely do different than POSIX, the existence of a > high-quality "precedent" standard is still key to the decision-making > process. > > > the Linux > > industry standard is some kind of compatibility with RHEL. > > I have to admit to knowing almost nothing about Ret Hat Inc's position > (in theory or practice) on POSIX.
Linux (in the broad sense of Linux -- kernel and glibc and userspace) follows POSIX when it suits or isn't difficult but ignores it when someone in a position to say so thinks POSIX is saying something stupid about the matter at hand. Basically, it's treated as an important guideline to respect if possible, but not as a sacrosanct standard which must be adhered to at all costs. But then, the LSB standards for Linux are in practice treated about the same way by most distro developers, so.... ;-) And Linux users mostly seem happy with that. How many of you actually set POSIX_ME_HARDER (or the newer, slightly less offensive POSIXLY_CORRECT) on your Linux boxes?[1] Or have ever even seen a production Linux box with that set? Linux users seem quite happy to choose ease-of-use over compliance.... later, chris [1] Setting that for many GNU utils gets them closer to POSIX compliance and farther from what GNU regards as sane usability configuration. For example, GNU coreutils (df, du, etc.) assumes 1k block sizes for output unless that variable is set, in which case it uses the 512-byte block size dictated by POSIX _______________________________________________ opensolaris-discuss mailing list opensolaris-discuss@opensolaris.org
