June 20th 2005 #
random, June 20th 2005
Interesting Times Are Here, or at least in SBCL CVS HEAD, which has recently seen a number of interesting changes in addition to the usual torrent of bugfixes:
- Pthread based threads on x86 instead of building directly on the clone syscall.
- Dynamic-extent support on x86-64, Alpha, and PPC for &REST lists, lists, and closures. (x86 had these already, and also boasts dynamic-extent vectors which the other ports still lack.)
- eucJP support is also present, thanks to Niimi Sathoshi. The moral: if your favorite external format is not yet supported, write a patch.
- Callback support exists, even though the interface is not yet exported. If your application uses or would use callbacks, please comment on the interface issues on sbcl-devel.
- Beginnings of an internals manual now live under doc/internals/.
- More efficient structure representation has been provided by David Lichteblau, saving a word or two of memory per allocated structure and removing one indirection from raw slot references.
...just to mention my favorites. Unfortunately current HEAD has also suffered some damage in the process and is apparently not quite up the usual standard on x86: Christophe Rhodes reports CLX non-workingness with the (p)threaded 0.9.1.59.
Not Quite A Retraction, but I guess I'll have to soften my position and clarify my standing regarding Lispwire and Franz: (i) it seems that the legalese is going to be toned down and (ii) Franz is indeed is quite active in reaching out to the lisp community at large, and has admirable open source projects of its own (like AllegroServe), but I still feel that the reaching out bit is directed more towards the segment of the community represented by ALU (which doesn't include yours truely) and less towards those segments characterized by, say, CLiki, common-lisp.net, and various open source lisp implementations and project. Nothing wrong with that, though.
I wonder if I've grown too fond of italics.