RE: [xml] Port of libxml for MPE/iX

-----Original Message-----
From: Daniel Veillard [mailto:veillard redhat com]
Sent: Friday, January 18, 2002 1:59 PM
To: Henke, Markus
Cc: 'xml gnome org'
Subject: Re: [xml] Port of libxml for MPE/iX

On Fri, Jan 18, 2002 at 01:41:08PM +0100, Henke, Markus wrote:

although i'm "...apparently don't dare looking at the API,
code, docs or answers..." (sic!)

 Okay, that was a bit harsh. Sorry about that but re-sending
the same issue just got on my nerve.

OK, let's forget it!  8)

And I don't want to change the decision to move all APIs to
UTF8 (except the very few ones dedicated to encoding support).

Clear, i never wanted to ask for an API _change_ (as i guessed,
there where some misconceptions, maybe it's my poor english...8)
I'm shure you have good & weighty reasons for your design

  Well the simplest is to send patches for those changes
and ask for feedback on those.

Oh, i've done so. But the arbitrative changes rather
concern the 'trio' code, i'm not shure what has become of it:

The changes to libxml a rather minor, that is the use of 'strdup'

and several missing 'static' modifier, but that's rather
AFAIK you've confirmed the patches/proposals (havn't had the
time to look at the CVS snapshots or version 2.4.13 by now).

If really there is some
patches which can't be integrated or special instructions
needed for that platform then one can add a directory for
those like we did for VMS and MacOs.

OK. BTW, seems that there are some parallels between VMS and
MPE WRT the build process of libxml...

Unfortunately 'make install' fails, i havn't a clue why
(there is a posting WRT that question).

  Hum, where, I don't remember.

No problem, i do :)

Additionally there's is a serious problem concerning the
MPE 'cc' compiler that won't return unresolved external
symbols as an error at link time, but i hope i can
provide a workaround soon (for the moment we've to
manually edit the generated 'config.h').

  I see so basically configure tests just don't work for that


That could be taken care from the script
or by keeping config.h.MPEiX somewhere with appropriate

Yep, i could provide that config.h.MPEiX, but setting the
right conditionals in config.h won't be sufficient since
the 'configure' skript also sets a 'NEED_TRIO' conditional
that influences the link process in 'make'.
Actualy i've modified the 'configure' skript to hardcoded
enable trio support:

    echo "$ac_t""no" 1>&6

+ # ORDAT/HEA/19.12.2001
+ # If compiling with HP C/iX on MPE

  echo $ac_n "checking for gethostent""... $ac_c" 1>&6

I'm shure that this is a ugly hack, maybe one could make
this more elegant (conditional)...

I'll also give an detailed description of the steps that
are necessary to build libxml2 on MPE if someone wants
to do so (which is rather unlikely, but who knows...).

  Just send them, and it may end up as a specific subdir for
that system.
OK, i'll write down the things i've done to build libxml
on MPE and send it to the list. Maybe some steps already
are or will be obsolet, we'll see...


Ciao, Markus

Mit freundlichen Gruessen - Kind regards
Markus Henke

________________________Addressed by:________________________
 ORDAT GmbH & Co. KG  -  Serversystems / eCom 
 Dipl.-Inf. (FH) Markus Henke  Fon: +49 (641) 7941-0
 Rathenaustr. 1                Fax: +49 (641) 7941-132
 35394 Gießen                  mailto:markus henke ordat com
              ...this behavior is by design...

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