Rice@SUMEX-AIM.STANFORD.EDU (James Rice) (12/05/89)
Well, I have a reasonable version of GNU pathnames now (I think). My next message will be the code itself, which is rather long (mainly because of copying 10 zillion page long TI functions. The code should, in principle, allow the easy definition of a fake version number generating type of pathname for any existing type of pathname for a file system that does not understand version numbers. I've implemented three of them so far fs:gnu-ucb-pathname - a gnu version of fs:unix-ucb-pathname fs:gnu-unix-pathname - a gnu version of fs:unix-pathname fs:gnu-mac-pathname - a gnu version of fs:mac-pathname Note: because all of the versioning is done on the client, the server needs no modification. Known misfeature/gray area: on the Mac pathname version I haven't managed to get the correct renaming to happen in the unwind protect if you abort a with-open-file. I keep getting mysterious invalid NFS reply errors. This is left as an exercise to the reader. Rice - yes, there are bound to be bugs, but hopefully this will get us most of the way. P.S. This all runs on 6.0. You'll have to do a little recoding if you don't have a working CLOS. P.S.S. If you don't have a TI source license please close your eyes.