Re: test results on Cygwin

2010-05-02 Thread Simon Josefsson
Bruno Haible writes: > Simon Josefsson wrote: >> > Thanks, it should likely solve the problem too -- I'm running a Windows >> > cygwin build to test it now. >> >> Yep: >> >> http://autobuild.josefsson.org/gnulib/log-20100425225612508.txt >> ... >> There is a bunch of printf-failures in ther

Re: test results on Cygwin

2010-05-02 Thread Bruno Haible
Simon Josefsson wrote: > > Thanks, it should likely solve the problem too -- I'm running a Windows > > cygwin build to test it now. > > Yep: > > http://autobuild.josefsson.org/gnulib/log-20100425225612508.txt > ... > There is a bunch of printf-failures in there that I don't recall, but > othe

Re: mbsinit problem with inlined original

2010-05-02 Thread Michael Goffioul
On Sun, May 2, 2010 at 8:27 PM, Bruno Haible wrote: > Hi, > > Michael Goffioul wrote: >> When trying to compile gnulib (from octave), it fails at compiling >> mbsinit.c. It seems this is due to the fact that on my system (MSVC) >> mbsinit is declared inlined in wchar.h. The configure script fails

wcwidth: declare also when it's aliased

2010-05-02 Thread Bruno Haible
Inside libunistring, I got this test compilation failure on mingw: depbase=`echo test-wcwidth.o | sed 's|[^/]*$|.deps/&|;s|\.o$||'`;\ gcc-3 -mno-cygwin -DHAVE_CONFIG_H -I. -I. -I../lib -I.. -DIN_LIBUNISTRING_GNULIB_TESTS=1 -I. -I. -I.. -I./.. -I../lib -I./../lib -I/usr/local/mingw/i

Re: mbsinit problem with inlined original

2010-05-02 Thread Bruno Haible
Hi, Michael Goffioul wrote: > When trying to compile gnulib (from octave), it fails at compiling > mbsinit.c. It seems this is due to the fact that on my system (MSVC) > mbsinit is declared inlined in wchar.h. The configure script fails to > detect the system mbsinit, because of link error, and th

Re: install-reloc error on Debian-hurd and Debian-kfreebsd

2010-05-02 Thread Sylvain Beucler
Apparently the builds went fine this time :) https://buildd.debian.org/pkg.cgi?pkg=freedink Thanks again! - Sylvain On Tue, Apr 20, 2010 at 08:18:07PM +0200, Sylvain Beucler wrote: > Hi Bruno, > > Thanks, I'll make a new release with an updated gnulib and see how > this goes :) > > - Sylvain

mbsinit problem with inlined original

2010-05-02 Thread Michael Goffioul
When trying to compile gnulib (from octave), it fails at compiling mbsinit.c. It seems this is due to the fact that on my system (MSVC) mbsinit is declared inlined in wchar.h. The configure script fails to detect the system mbsinit, because of link error, and the compiler fails to compile the repla

Re: Minor bug in sc_program_name of maintainer-makefile.

2010-05-02 Thread Jim Meyering
James Youngman wrote: > Minor bug: sc_program_name complains about programs not calling > set_program_name even if they're not linked with gnulib and the > progname module wasn't imported. Yeah, this test is useful for coreutils, with its 100+ programs but less so for others. I don't see an easy,

Minor bug in sc_program_name of maintainer-makefile.

2010-05-02 Thread James Youngman
Minor bug: sc_program_name complains about programs not calling set_program_name even if they're not linked with gnulib and the progname module wasn't imported. James.

Re: testdir without unportable tests?

2010-05-02 Thread Bruno Haible
A week ago I committed this: > 2010-04-25 Bruno Haible > > gnulib-tool: Don't include hairy tests of dependencies in testdirs. > * gnulib-tool (func_usage): Document that --with-*-tests options apply > also to --create-testdir. > (func_acceptable): Don't consider the sta

Re: ws2tcpip.h include bug

2010-05-02 Thread Ozkan Sezer
On Sun, May 2, 2010 at 1:14 PM, Bruno Haible wrote: > Hi, > > Ozkan Sezer wrote: >> It is mingw-w64 (http://mingw-w64.sourceforge.net/) where >> the provided ws2tcpip.h header doesn't include winsock2.h. >> The mingw32 header does include it, for convenience as it >> seems > > It is not "for conve

Re: test results on Cygwin

2010-05-02 Thread Bruno Haible
Hi, On 2010-04-05 I reported: > A simplified test case is attached. It crashes when optimized with -O2 > and works fine without optimization: > ... > The same behaviour also occurs when the test case uses the gnulib replacement > of mbsrtowcs. Therefore I think it's not a bug in Cygwin's mbsrtowcs

Re: ws2tcpip.h include bug

2010-05-02 Thread Bruno Haible
Hi, Ozkan Sezer wrote: > It is mingw-w64 (http://mingw-w64.sourceforge.net/) where > the provided ws2tcpip.h header doesn't include winsock2.h. > The mingw32 header does include it, for convenience as it > seems It is not "for convenience". It is for correctness. 20 years ago it was usual that y

Re: ws2tcpip.h include bug

2010-05-02 Thread Ozkan Sezer
On Sun, May 2, 2010 at 11:11 AM, Bruno Haible wrote: > Hi, > > Ozkan Sezer wrote: >> The config header checks for windows' ws2tcpip.h have a little bug where >> they do not include winsock2.h before ws2tcpip.h: ws2tcpip.h requires >> winsock2.h to be included before, a plain AC_CHECK_HEADERS([ws2t

Re: ws2tcpip.h include bug

2010-05-02 Thread Bruno Haible
Hi, Ozkan Sezer wrote: > The config header checks for windows' ws2tcpip.h have a little bug where > they do not include winsock2.h before ws2tcpip.h: ws2tcpip.h requires > winsock2.h to be included before, a plain AC_CHECK_HEADERS([ws2tcpip.h]) > like in current m4/sys_socket_h.m4 may fail to dete