[JDEV] Re: [jadmin] [Fwd: jabberd porting questions/comments]

[Peter Saint-Andre]

Hi Doug,

Thanks for working to port jabberd to HP-UX, that's cool. The "build
manager" (among his many other roles) is Thomas Muldowney, a.k.a. temas,
so I'm cc'ing him on this message to make sure he sees your
contribution and cross-posting to the JDEV list to flush out any other
HP-UXers out there. :)

Peter

On Wed, 13 Jun 2001, Doug Claar wrote:

> Wow! There are so many ways to communicate! I hope this is the right one...
> I am trying to port jabberd 1.4.1 to hp-ux. I've pursued both the gmake/gcc
> method and the make/ansi c methods, and have run into problems with both.
> 
> Once/if I get it running, I am willing to feed back my changes if folks
> are interested. (I've never done that before, so I guess I get to learn
> something new!)
> 
> Problems I've found in the non-gmake/gcc path:
> 
> The makefiles are gnumake specific, while this is not really necessary.
> I'd like to change them to work with standard make. This would entail
> modifying all the := lines, and replacing the ifeq() stuff in
> jabberd/Makefile. Pretty minor stuff.
> 
> The c files are not ansi c compatible. The first problem I ran into is
> that anonymous structures are not an ansi feature. Thus, the call:
> 
>    jutil_error(pkt, (terror){502, "Unable to resolve hostname."});
> 
> is invalid. This can be fixed by either a utility function:
> 
>    terror mk_terror(const int code, const char *msg)
> 
> and rewriting the above as:
> 
>    jutil_error(pkt, mk_terror(502, "Unable to resolve hostname."));
> 
> Or else rewriting jutil_error to have 3 arguments.
> 
> The next problem is that jabberd/single.h is missing a backslash on
> the "Usage:" line.
> 
> The next problem, which I haven't solved yet, is that the mio.c
> routine _mio_write_dump wants to add to a void pointer, and:
> 
>    cc: "mio.c", line 295: error 1539: Cannot do arithmetic with pointers
>       to objects of unknown size.
> 
> I think I just need to cast to a char *, but I haven't studied the
> function enough to be sure...
> 
> 
> The gmake/gcc path had problems with interpreting a warning that
> __va_alist was redefined (in stdargs.h and stdlib.h) as an error,
> and so configure punted early. I got past that (not sure how, when
> I added -xv to the lower level configure script, it worked...),
> but then I ran into a problem with pth_acmac.h understanding a define:
>    #define pth_sksize_sigstack(skaddr,sksize) ((sksize)')
>    cpp: "pth_acmac.h", line 38: error 4007: Unterminated string or
>       character constant.
> 
> ==Doug Claar
> _______________________________________________
> jadmin mailing list
> jadmin at jabber.org
> http://mailman.jabber.org/listinfo/jadmin
>