Re: [PATCH] fix build failure with getline.c

-BEGIN PGP SIGNED MESSAGE- Hash: SHA1 According to Robert Millan on 1/1/2010 4:24 PM: >> Aren't you using the gnulib stdio module, or is the code not working >> somehow? >> > So no, the stdio module wasn't being included. Maybe you want to consider > removing this dependency, I think othe

Re: [PATCH] fix build failure with getline.c

On Fri, Jan 01, 2010 at 11:45:19PM +0100, Simon Josefsson wrote: > Robert Millan writes: > > > When importing getline.c into GNU GRUB, it failed to build due to undefined > > ssize_t. Here's a fix (I believe including is portable > > enough, > > please let me know if I'm wrong). > > I don't y

Re: [PATCH] fix build failure with getline.c

Eric Blake writes: > Which platform was this failure on? Perhaps it resulted from > Bruno's attempts to reduce namespace pollution on glibc machines, coupled > with an older glibc that didn't yet expose everything required by POSIX 2008? Debian Lenny with glibc 2.7. -- Robert Millan "Be the

Re: [PATCH] fix build failure with getline.c

-BEGIN PGP SIGNED MESSAGE- Hash: SHA1 According to Simon Josefsson on 1/1/2010 3:43 PM: >> with an older glibc that didn't yet expose everything required by POSIX 2008? > > Is there really a requirement for stdio.h to define ssize_t? Glibc's > stdio.h doesn't define ssize_t: Yes, POSIX

Re: [PATCH] fix build failure with getline.c

Robert Millan writes: > When importing getline.c into GNU GRUB, it failed to build due to undefined > ssize_t. Here's a fix (I believe including is portable enough, > please let me know if I'm wrong). I don't your this patch is the right solution -- the gnulib getline module depends on the std

Re: [PATCH] fix build failure with getline.c

Eric Blake writes: > According to Robert Millan on 1/1/2010 5:56 AM: >> When importing getline.c into GNU GRUB, it failed to build due to undefined >> ssize_t. Here's a fix (I believe including is portable enough, >> please let me know if I'm wrong). > > Hmm. The prototype for getline lives in

Re: warn-on-use preview

Hi Eric, > I'm glad you liked my approach at poisoning those identifiers. Sure! I didn't know how to do it; you found the approach :-) > By the way, is there any reason we don't have an fpclassify > module yet? It's rarely used in math function algorithms: Hardly any algorithm needs to disting

Re: link-warning usage improvements

Eric Blake wrote: > >> + headers: make check of system header explicit > >> + * m4/netdb_h.m4 (gl_HEADER_NETDB): Don't exploit knowledge of > >> + gl_CHECK_NEXT_HEADER internals, but call AC_CHECK_HEADERS_ONCE > >> + ourselves. > > > > This looks fine. > > Pushed (after rebasing to pick up the

Re: link-warning usage improvements

-BEGIN PGP SIGNED MESSAGE- Hash: SHA1 According to Bruno Haible on 1/1/2010 5:47 AM: > Eric Blake wrote: >> +2009-12-31 Eric Blake >> + >> +headers: make check of system header explicit >> +* m4/netdb_h.m4 (gl_HEADER_NETDB): Don't exploit knowledge of >> +gl_CHECK_NEXT_HEADE

Re: [PATCH] update nearly all FSF copyright year lists to include 2010

Eric Blake wrote: > According to Jim Meyering on 1/1/2010 2:38 AM: >> +++ b/ChangeLog >> @@ -1,5 +1,9 @@ >> @@ -9296,7 +9300,7 @@ >>$ ./src/ls --version | head -n3 >>ls (GNU coreutils) 7.1-dirty >>Packaged by Gentoo (patchset 1.6) >> - Copyright (C) 2009 Free Software F

Re: warn-on-use preview

-BEGIN PGP SIGNED MESSAGE- Hash: SHA1 According to Bruno Haible on 1/1/2010 7:11 AM: > The recommendation how to deal with variables like 'environ': > > static inline char ***rpl_environ (void) { return &environ; } > # define environ (*rpl_environ ()) > > will not work when a program

Re: warn-on-use preview

Hi Eric, Eric Blake wrote: > I won't push this until it's had a bit longer for reviews. Yes, for reviewing 4 patches like this you need to give me a day at least. > [1/4] warn-on-use: new module The recommendation how to deal with variables like 'environ': static inline char ***rpl_env

Re: [PATCH] fix build failure with getline.c

-BEGIN PGP SIGNED MESSAGE- Hash: SHA1 According to Robert Millan on 1/1/2010 5:56 AM: > When importing getline.c into GNU GRUB, it failed to build due to undefined > ssize_t. Here's a fix (I believe including is portable enough, > please let me know if I'm wrong). Hmm. The prototype fo

Re: [PATCH] update nearly all FSF copyright year lists to include 2010

-BEGIN PGP SIGNED MESSAGE- Hash: SHA1 According to Jim Meyering on 1/1/2010 2:38 AM: > +++ b/ChangeLog > @@ -1,5 +1,9 @@ > @@ -9296,7 +9300,7 @@ > $ ./src/ls --version | head -n3 > ls (GNU coreutils) 7.1-dirty > Packaged by Gentoo (patchset 1.6) > - Copyright

[PATCH] fix build failure with getline.c

When importing getline.c into GNU GRUB, it failed to build due to undefined ssize_t. Here's a fix (I believe including is portable enough, please let me know if I'm wrong). -- Robert Millan "Be the change you want to see in the world" -- Gandhi 2010-01-01 Robert Millan * lib/getline.c:

Re: link-warning usage improvements

Eric Blake wrote: > +2009-12-31 Eric Blake > + > + headers: make check of system header explicit > + * m4/netdb_h.m4 (gl_HEADER_NETDB): Don't exploit knowledge of > + gl_CHECK_NEXT_HEADER internals, but call AC_CHECK_HEADERS_ONCE > + ourselves. > + * m4/search_h.m4 (gl_SEARCH

[PATCH] test-update-copyright: don't let envvar setting cause test failure

Without this patch, and with my left-over setting, export UPDATE_COPYRIGHT_MAX_LINE_LENGTH=79 the update-copyright test would fail: FAIL: test-update-copyright.sh (exit: 1) --- - 2010-01-01 11:23:31.412799557 +0100 +++ update-copyright.test-ex.1 2010-01-01 11:23:31.400435448 +0100

Re: localename and thread locale

Hi, I wrote on 2009-12-26: > (gl_locale_name_thread_unsafe): New function, extracted from > gl_locale_name_thread. This leads to gcc warnings localename.c:2602: warning: `gl_locale_name_thread_unsafe' defined but not used Fixed as follows: 2010-01-01 Bruno Haible loc

FYI, version-etc.c's enum now says 2010

The rest will be updated shortly.