Vincent Arkesteijn <vincent@...> writes:
> If it would have been easier for you if I had sent these separately,
> please let me know.
From my perspective, it would have been. Perhaps I should explain
why: when there are multiple independent changes in one patch, I have
to think about each of them: whether any changes need touching up,
whether I should apply them when we're in the middle of a code freeze,
and so on.
The way I typically keep track of sbcl development stuff is to tag
messages in my sbcl-devel buffer until they're dealt with (currently
105 tagged messages, *sigh*). Dealing with them can mean applying or
fixing whatever the patch or issue is; replying to a question; or
scribbling something into BUGS or some more distributed place of
record than my mail files. So in many ways not only would independent
patches be appreciated, separate messages are more convenient (at
least for me; I don't actually know how other sbcl developers work)
each dealing with one issue.
With that said:
> 2. contrib/sb-aclrepl/sb-aclrepl.texinfo
> small typo
> 3. auto-generated (not anymore?) manual for sb-bsd-sockets
> Internet address as returned by for instance get-host-by-name are
> lists nowadays. make-inet-address still returns vectors, so call them
> sequences in the documentation.
> 4. contrib/sb-grovel/def-to-lisp.lisp
> FIXME has been handled.
> 5. contrib/sb-grovel/sb-grovel.texinfo
> packages that include groveled stuff now need to use sb-alien.
> 6. doc/manual/ffi.texinfo
> make the documentation of alien array type specifiers match its
> load-1-function and load-function were probably typos for
> ...-foreign, so change that to load-shared-object
I applied these, thanks. Whoever is looking at the various
sb-bsd-sockets issues may wish to repatch the sb-bsd-sockets
documentation for whatever it is decided to do with get-host-by-name.
> 1. make-config.sh
> 'whoami' isn't normally in the path on Solaris, so add /usr/ucb
> (also, whoami isn't posix, so one might want to change it to 'id
> -un'. On SunOS, you would then need /usr/xpg4/bin/id, and not
> %F as a format specifier ('date +%F') is a GNU extension, and (as you
> no doubt could guess) isn't supported on SunOS. Change to %Y-%m-%d.
> 7. src/code/sparc-vm.lisp
> implement get-machine-version for SunOS.
> (Does it strike anyone else as odd that these functions are in
> hardware-specific source files? Finding out the machine version
> depends more on the os than on the hardware, I would say.)
I didn't apply these two, because changing stuff like this in the
middle of a code freeze is a little worrying. In case you are
motivated to do a little more on this, I'd suggest reworking the first
to be posixly correct if possible (that is, using id(1), in a portable
way in as much as you can), and to examine whether there is any saving
in moving machine-version to an os-specific file in the second
All that notwithstanding, thank you very much for raising the issues
and for providing patches :-)
http://www-jcsu.jesus.cam.ac.uk/~csr21/ +44 1223 510 299/+44 7729 383 757
(set-pprint-dispatch 'number (lambda (s o) (declare (special b)) (format s b)))
(defvar b "~&Just another Lisp hacker~%") (pprint #36rJesusCollegeCambridge)