Re: [Plib-users] need help compiling plib

[Steve B. <sjb...@ai...>]

Sebastian Ude wrote:
> 
> On Sun, 21 Apr 2002, sjb...@ai... (Steve Baker) wrote:
> > Date: Sun, 21 Apr 2002 18:48:23 -0500
> > To: ud...@ha...
> > From: sjb...@ai... (Steve Baker)
> > CC: pli...@li...
> > Reply-To: sjb...@ai...
> > Subject: Re: [Plib-users] need help compiling plib
> 
> [...]
> 
> > > The source file should include "sys/types.h", but it doesn't. (To the
> > > original author: typedef'ing "socklen_t" to be "int" is not a solution.
> >
> > There are problems with some OS's (I think IRIX is one of them) with
> > those things.  I don't think older Unixen define socklen_t - so you have
> > to use an 'int'.
> 
> Okay. The problem is, however, that by not including "sys/types.h", the
> current code even typedef's "socklen_t" to be "int" on machines that *do*
> define socklen_t, which can be fatal when it comes to pointers to those
> types.
> 
> As the manuals of standard-conforming UNIX systems do, I really suggest
> including "sys/types.h" in addition to "sys/socket.h". It is a standard
> header file - does IRIX miss it, too ? If yes, just add a check for it in
> configure.in.

sys/types.h exists - but the socklen_t type does not.

Clearly we should be including that header - but we may need conditional
compilation to use socklen_t only if it's defined and fall back on an 'int'
instead. To be honest though, it's hard to imagine a socket system whose
internal structure was more than 2^32 bytes long!

----------------------------- Steve Baker -------------------------------
Mail : <sjb...@ai...>   WorkMail: <sj...@li...>
URLs : http://www.sjbaker.org
       http://plib.sf.net http://tuxaqfh.sf.net http://tuxkart.sf.net
       http://prettypoly.sf.net http://freeglut.sf.net
       http://toobular.sf.net   http://lodestone.sf.net