Re: [PATCH] socklen: do not depend on sys_socket

2011-03-17 Thread Paul Eggert
On 03/17/2011 08:29 PM, Bruno Haible wrote: >> Emacs is not intended to be portable to MingW and Cygwin. > Huh? How does this statement match the "GNU Emacs FAQ For MS Windows" [1], > section 2.4? Ah, sorry, I was mistaken. I don't understand how Emacs is built on MS-DOS or MS-Windows (nor do I u

Re: [PATCH] socklen: do not depend on sys_socket

2011-03-17 Thread Bruno Haible
Paul Eggert wrote: > Emacs is not intended to be portable to MingW and Cygwin. Huh? How does this statement match the "GNU Emacs FAQ For MS Windows" [1], section 2.4? Bruno [1] http://www.gnu.org/software/emacs/windows/faq.html -- In memoriam Matthias Kraus

[PATCH] socklen: do not depend on sys_socket

2011-03-17 Thread Paul Eggert
Here's another Emacs-related patch I'm proposing for gnulib to break an unnecessary dependency, in this case a circular one. While trying to modify Emacs to use gnulib's socklen module, I discovered a circular dependency: socklen depends on sys_socket and vice versa. Emacs can use socklen,

Re: -Wundef warning in striconveh.h

2011-03-17 Thread Andy Wingo
On Thu 17 Mar 2011 17:15, Paul Eggert writes: > On 03/17/2011 07:46 AM, Andy Wingo wrote: >> Can you consider changing this one header file? It does not sound like >> an issue on which it is fruitful to be stubborn. > > Well, there is a bit of stubbornness on both sides here, no? :-) Hehe, poi

Re: -Wundef warning in striconveh.h

2011-03-17 Thread Paul Eggert
On 03/17/2011 07:46 AM, Andy Wingo wrote: Can you consider changing this one header file? It does not sound like an issue on which it is fruitful to be stubborn. Well, there is a bit of stubbornness on both sides here, no? :-) But I tend to agree: for headers meant to be included by other co

Re: -Wundef warning in striconveh.h

2011-03-17 Thread Andy Wingo
Hi Eric, On Thu 17 Mar 2011 15:05, Eric Blake writes: > On 03/17/2011 04:20 AM, Andy Wingo wrote: >> striconveh.h:22 and striconveh.h:34 use #if when they should use #ifdef, >> or something. > > Not a bug in gnulib, > >> >> Building Guile a contributor got this warning: > > but in guile. Gnuli

Re: -Wundef warning in striconveh.h

2011-03-17 Thread Eric Blake
On 03/17/2011 04:20 AM, Andy Wingo wrote: > Hello, > > striconveh.h:22 and striconveh.h:34 use #if when they should use #ifdef, > or something. Not a bug in gnulib, > > Building Guile a contributor got this warning: but in guile. Gnulib specifically documents that -Wundef is incompatible with

-Wundef warning in striconveh.h

2011-03-17 Thread Andy Wingo
Hello, striconveh.h:22 and striconveh.h:34 use #if when they should use #ifdef, or something. Building Guile a contributor got this warning: CC libguile_2.0_la-bytevectors.lo In file included from bytevectors.c:41:0: ../lib/striconveh.h:22:5: warning: "HAVE_ICONV" is not defined [-Wundef]

Re: Support for options in valgrind-tests

2011-03-17 Thread Reuben Thomas
On 17 March 2011 06:27, Ralf Wildenhues wrote: > I suggest you guys check out the parallel-tests driver from recent > Automake.  It provides makefile variables for specifying a command > to run specific tests, plus options to that.  That can be used for > valgrind. It would certainly be nice if t

Re: Making releases

2011-03-17 Thread Reuben Thomas
On 17 March 2011 07:25, Jim Meyering wrote: > > reuben-stable: >        $(MAKE) stable >        $(MAKE) upload >        $(MAKE) send-announcement Nice. > But you'll want a delay between the upload and the actual announcement. > At least when uploading to ftp.gnu.org, you have to wait for a while

Re: Making releases

2011-03-17 Thread Jim Meyering
Reuben Thomas wrote: > On 16 March 2011 19:59, Jim Meyering wrote: >> Reuben Thomas wrote: >>> >>> OK, so an extra target is needed. I used to use "make release". This >>> would seem to make sense to cover uploading and announcing the >>> release. >> >> There's already a target (or three) for that