george@mnetor.UUCP (George Hart) (04/26/89)
In article <642@mitisft.Convergent.COM> kemnitz@mitisft.Convergent.COM (Gregory Kemnitz) writes: > >o Everything on the X tape should be POSIX compliant > >I just got through the arduous adventure of porting X to a System V system. >(Fortunately, we had a socket implementation). Are there any plans by anyone >in MIT or somewhere to make Xlib POSIX-based? (I realize that this probably >won't be necessary for about a year or eighteen months, but EVERYONE's **IX >(and VMS???) should be POSIX-compliant by then.) From the limited experience around here, IPC is the biggest problem area in porting the libraries, particularly to systems that do not have support for byte counts (BytesReadable()) and timeouts. I agree with the sentiment about POSIX-compliance. Alas, it is not clear what "POSIX-compliant" means in the context of IPC. POSIX 1003.1 does not define networking functionality. Unfortunately, the POSIX networking subcommittee (1003.8?) doesn't appear to be far enough along to give reliable direction. (If I am wrong, *please* correct me). On the other hand.... Given the adoption of streams by AT&T/SUN and OSF, and more importantly the adoption of TLI (Transport Level Interface) by AT&T/Sun and X/OPEN (which calls it XTI), it may still be appropriate for the Consortium to invest the effort in streams/TLI with POSIX compliance as the longer term goal. Anything is possible, but it is hard to believe that given the support for streams and TLI (extant and promised), that the POSIX networking subcommittee would exclude them from any proposed standard. -- Regards.....George Hart, Computer X Canada Ltd. UUCP: {utzoo,uunet}!mnetor!george BELL: (416)475-8980