From: William H. N. <wil...@ai...> - 2000-05-17 21:36:00
|
On Wed, May 17, 2000 at 07:23:18PM +0200, Raymond A. Wiker wrote: > I've just applied my patches for FreeBSD 4.0 to SBCL-0.6.4. The > patches applied cleanly, and I got a working sbcl on my first compile > attempt. I can live with that :-) I'm happy to hear this, because I've just applied the 4.0 upgrade to my FreeBSD 4.0 system, and now even after frustrating work to clean it up, it's still brain-damaged (X not working, Ethernet not working..) and I'm sufficiently frustrated with my attempts to make it work that I'm inclined to make my office a FreeBSD-free zone soon, and I'm certainly not in the mood to enjoy spending a lot of time trying to get the port to work.. > [..problems with UNIX interface..] I haven't actually done much work with the Unix interface other than commenting out stuff which seemed not to be used. I did make some mistakes, commenting out too much stuff, so if EINTR or some other symbol is needed but is commented out, it's probably another such mistake on my part, and the quick fix is to uncomment it. As a longer-term solution, I think Daniel Barlow's approach sounds promising for a lot of stuff, and I'd also like to write some C wrapper functions for other stuff. My guess is that the idea of writing new software to parse #include files is probably impractical -- there are a lot of nasty, confusing things in existing #include files, and the situation could well get worse rather than better as time goes on. > mk:defsystem has quite a few feature-tests on :cltl2, which is > predefined for CMUCL. Should this also be put on the features-list for > SBCL, or would it be better to change defsystem.lisp to use :ansi-cl > or :sbcl directly? I think it was Daniel Barlow who pointed out that the feature :CLTL2 isn't really supposed to be consistent with ANSI Common Lisp. I seem to dimly remember that that's why I took it out of the features list in SBCL. Alas, my development log is on the other end of a brain-damaged Ethernet connection, so I can't check very well right now, but I tentatively agree with Daniel that SBCL is doing this right, and should not change, and that instead portable library code should be rewritten to depend on some other feature tag. -- William Harold Newman <wil...@ai...> software consultant PGP key fingerprint 85 CE 1C BA 79 8D 51 8C B9 25 FB EE E0 C3 E5 7C |