bug#21114: [PATCH 1/4] Fix the rule to check for new signals and errors
David Michael skribis: > * .gitignore (/libguile/*.NEW): New pattern. > * libguile/Makefile.am (chknew-E chknew-SIG): Remove the line > continuation after the targets, and include numbers in the > recipe's signal/error regexp to catch names like E2BIG. > --- > > > This patch series makes a few updates to the generation of the lists of > signal and error signals. This first one makes the chknew targets work > again. GNU Make says there is no such target without this. Indeed. Applied in 9a4215b, thanks. Ludo’.
bug#21115: [PATCH 2/4] Add new Linux error definitions
David Michael skribis: > * libguile/cpp-E.syms (EHWPOISON, ERFKILL): New definitions. Pushed as ead362f.
bug#21116: [PATCH 3/4] Add Hurd signal and error definitions
David Michael skribis: > * libguile/cpp-E.syms (EAUTH, EBACKGROUND): New definitions. > (EBADRPC, ED, EDIED, EFTYPE, EGRATUITOUS, EGREGIOUS): Likewise. > (EIEIO, ENEEDAUTH, EPROCLIM, EPROCUNAVAIL): Likewise. > (EPROGMISMATCH, EPROGUNAVAIL, ERPCMISMATCH): Likewise. > * libguile/cpp-SIG.syms (SIGEMT, SIGEV_MAX_SIZE): Likewise. > (SIGEV_PAD_SIZE, SIGINFO, SIGLOST): Likewise. Pushed as 1be3063. > #define ED _HURD_ERRNO (102)/* ? */ > #define EDIED _HURD_ERRNO (101)/* Translator died */ > #define EFTYPE _HURD_ERRNO (79)/* Inappropriate file type or > format */ > #define EGRATUITOUS _HURD_ERRNO (105)/* Gratuitous error */ > #define EGREGIOUS _HURD_ERRNO (103)/* You really blew it this > time */ > #define EIEIO _HURD_ERRNO (104)/* Computer bought the farm */ Hurd humour for the Guilers! :-) Ludo’.
bug#21117: [PATCH 4/4] Remove sigevent structure value definitions
David Michael skribis: > * libguile/Makefile.am (chknew-E chknew-SIG): Anchor the end of the > signal/error name pattern to only match alphanumeric symbols. > * libguile/cpp-SIG.syms (SIGEV_MAX_SIZE, SIGEV_NONE): Remove. > (SIGEV_PAD_SIZE, SIGEV_SIGNAL, SIGEV_THREAD): Likewise. > (SIGEV_THREAD_ID): Likewise. > --- > > > I'm assuming those sigevent structure values weren't intended to have > their definitions included with signals. This patch removes them and > adds stricter name matching to the regexp. Good catch. Pushed as c3f95ec. Thanks! Ludo’.
bug#21425: [PATCH] Remove needless mkstemp() check
Kouhei Sutou skribis: > There is mkstemp() check by AC_REPLACE_FUNCS() in > configure.ac. If mkstemp() isn't found, mkstemp() is defined > by gnulib-tool. > > But mkstemp() is always defined in libguile/mkstemp.c. If we > define mkstemp() by gnulib-tool, mkstemp() is defined > twice. It caused a link error. > > The situation is caused when we cross-compile Guile by MinGW > because mkstemp() doesn't exist on Windows. Indeed. I pushed as similar patch as efd8a43. Thank you! Ludo’.
bug#21221: guile-2.2 socket ports used with display does not send utf8 !
Linas Vepstas skribis: > The following simple client-server program fails for me. > For simplicity, for the server, just use netcat listening on port : > > $ nc -l > > In a guile shell, try this: > (setlocale LC_ALL "") > (define sss (socket PF_INET SOCK_STREAM 0)) > (set-port-encoding! sss "utf-8") > (connect sss AF_INET (inet-pton AF_INET "127.0.0.1") ) > (set-port-encoding! sss "utf-8") > (display "SmålandSmåland\n" sss) > (close-port sss) I can confirm this bug with commit 00884bb. Guile 2.0.11ish is unaffected though. Ludo’.
bug#21373: compile for development version from git fails on OpenBSD
Jamil Egdemir skribis: > I was using the flex that came with OpenBSD 5.6 which was 2.5.4 so I > pulled down and installed flex-2.5.39. This got me past these errors > though I'm not sure if this is associated with differences between the > OpenBSD brand of flex or just a versioning issue. If it's a versioning > issue then it probably make sense to put something in configure.ac like > AC_PROG_LEX to make sure we have a flex that's a bit more recent than > 2.5.4 as suggested here: We cannot use AC_PROG_LEX because Flex is not a dependency when building from a tarball. AC_PROG_LEX would look for flex/lex when running ‘configure’, even though it’s not actually needed. Maybe we could specify a minimum version number in ‘HACKING’ though? I’m not sure which one. Thanks, Ludo’.
