[Pidgin] #3798: portability patch

[Pidgin]

#3798: portability patch
---------------------------+------------------------------------------------
  Reporter:  pogma         |       Owner:  rlaager
      Type:  patch         |      Status:  new    
  Priority:  minor         |   Milestone:         
 Component:  unclassified  |     Version:  2.2.2  
Resolution:                |    Keywords:         
   Pending:  1             |  
---------------------------+------------------------------------------------
Changes (by pogma):

  * pending:  1 => 0

Comment:

 Replying to [comment:9 rlaager]:
 > I think parts of this are ready to commit and other parts can be dropped
 entirely:
 >
 > The libgcrypt guys don't seem interested in changing this. I don't like
 that they're calling abort() (instead of exit() like they were at the time
 of previous comments), but that's the way it is. That said, why are you
 compiling in SSL support for machines with no way of using it? I think I'd
 rather leave those changes out.

 For systems that do not have a native /dev/random, we rely on a pseudo
 random number generator PRNGD to provide some randomness, however, it is
 possible that the daemon is not running, and also possible that the pidgin
 user on those systems has no need of ssl. However, accepting/rejecting
 bits of the patch is entirely up to you.

 >
 > Not that it's likely to conflict with anything, but we should use
 something else instead of __use_fork__.

 Yes. Why I used __ is beyond me, I *know* that double underscore bracketed
 macros are reserved for the OS/compiler :(

 >
 > HAVE_STRUCT_SOCKADDR_SA_LEN has an extra underscore after HAVE in the
 configure script.

 Indeed it does, so the code always assumed it was missing on our systems
 for that build. Oh well, proves that it works :-)

 >
 > The changes to gtkmain.c should be obsoleted by newer GStreamer
 versions, as the comment in gtkmain.c says now.

 Ok.

 >
 >
 > What real name would you like me to use for COPYRIGHT (if you're not
 already listed)? What e-mail address should I use for the commit?

 pogma at thewrittenword.com. My real name is Peter O'Gorman, but the
 copyright belongs to my employer, The Written Word, Inc.

 >
 > The config.h at the top of every file seems like something that needs
 more investigation. Do you know why that was required?

 I believe that it was because none of our AIX systems would compile pidgin
 otherwise. config.h includes macros for, for example, large file support,
 that must be set before any system headers are included.

 Note that the requirement to have config.h as the first include in all
 source files is listed here:
 http://www.gnu.org/software/autoconf/manual/autoconf.html#Configuration-
 Headers "The package should `#include' the configuration header file
 before any other header files, to prevent inconsistencies in declarations
 (for example, if it redefines const)."

 If I recall correctly, pidgin #includes some headers, then ends up
 #including config.h and that caused xlc to stop the build with an error
 when another system header was included.

 Thank you very much for looking into this, I had thought it forgotten.

-- 
Ticket URL: <http://developer.pidgin.im/ticket/3798#comment:10>
Pidgin <http://pidgin.im>
Pidgin