Re: [xml] autoconf patches to 2.4.2 (determine socklen_t type)

On Thu, Aug 16, 2001 at 04:20:15PM -0400, Daniel Veillard wrote:
On Thu, Aug 16, 2001 at 03:05:05PM -0500, xml thewrittenword com wrote:
The current autoconf code to determine the socklen_t type is icky.
Relying on a return error from the compiler doesn't seem so robust to
me (in all fairness though the patch below requires the compiler to be
smart enough to error out when two prototype declarations don't
match). So, I suggest the attached patch. NOTE: this depends on
autoconf 2.52 (2.50 will probably work). I could get it working under
2.13 if need be. You will need to regenerate I am trying
to get the same socklen_t detection into rsync (which switched to
autoconf 2.50 hence the dependency). This patch has been tested on
Solaris 2.5.1, 2.6, 7, 8, IRIX 6.2, 6.5, Tru64 UNIX 4.0D, 5.0A, 5.1,
HP-UX 10.20, 11.00, and AIX 4.3.2 with success.

  Is it broken ? I.e. does the current detection fails ? If yes
then what architectures ?

  basically, I have learnt the hard way that:
    - requiring the latest auto* is a path doomed to failure
      for example the latest libtool happily break my setup here.
    - socklen_t detection have been a PITA so far
 I'm sorry, but my attitude w.r.t. the autoconf scripts are:
   "If it ain't broken, don't fix it"

 So the first question to answer is:
   - is it broken ?
   - if yes, what arch ?
   - then how can we fix it with no added requirement to the
     auto*/libtool stuff ?

 Honnestly I do think that the auto* is a pile of crap (multi thousand
lines of perl stuff, nuff' said) and the less I have to mess with it the
better I feel. This is "icky" and "not robust" anyway from the framework
itself, so I don't really consider aestetic questions at that level, but
just "does it work ?".

Nothing is broken. I was just offering it as something I thought was

albert chin (china thewrittenword com)

[Date Prev][Date Next]   [Thread Prev][Thread Next]   [Thread Index] [Date Index] [Author Index]