Re: PATCH: Compilation with LIBXML2 was broken


Subject: Re: PATCH: Compilation with LIBXML2 was broken
From: Thomas Fletcher (thomasf@qnx.com)
Date: Mon Feb 05 2001 - 11:25:12 CST


On Mon, 5 Feb 2001, Aaron Lehmann wrote:

> On Mon, Feb 05, 2001 at 09:02:00AM -0500, Thomas Fletcher wrote:
> > I personally think that this argument is totally flawed.
> > Are we planning on distributing both an expat version and
> > a libxml version of AbiWord?
>
> Excuse me? Support for each exists in all recent versions. They're
> options, not forked trees.

 Yes, I understand that. If they were in fact forks then it would
be terrible, the fact that they have both been integrated into AbiWord
is great. What I mean (I've already posted on this) is what library
are we going to bind to in a final release. If we bind to that
library, why not just junk the other one "which ever it may be".
My worry is that our distribution for dynamic objects becomes:

AbiWord_dynamic_expat_pspell
AbiWord_dynamic_expat_ispell
AbiWord_dynamic_libxml_pspell
AbiWord_dynamic_libxml_ispell

And that is with only two options. I like the idea of standardizing
on a library/technology and then working with it. If something better
comes along ... then move to it.
 
> > This would make maintenance
> > debugging and problem reporting much more complicated than
> > need be.
>
> XML parsers are not used all that often in the code. Both seem to work
> fine now and it shouldn't be very difficult to debug problems.

Yes .. while this discussion is about XML parsers in particular, I'm
talking about choices and options and distributing them in general.
 
> > My personal opinion on the matter is that we should
> > choose the best technology solution (size, speed, meets our
> > current and expected future needs) and then just stick to
> > that (I personally would choose expat since it has proven to
> > be very portable ... but I'm actually indifferent).
>
> Why don't we choose the best technological widget set/operating
> system?

That is not the point. If our standard distribution model is
going to be that you get GNOME + libxml or GTK + expat,
Windows + expat ... etc fine. I just would like to hear and
have it made clear what the dependancies are.

> Expat appears to be the standard on QNX, just as I beleive that
> libxml2 is more standard on free Unix implementations. This is
> analagous to one using Photon and the other using GTK, except the XML
> interfaces are waaay simpler and less utilized than the graphics and
> frontend calls to different platforms' toolkits.

Expat isn't "standard" on QNX, it is a convenience library which
is shipped with it. What do we tell our users about the dynamic
packages in terms of the components that they require to run? Do
we want to have a standard? Do we distributed dynamic libraries
along with the dynamic distribution or does the user have to figure
out what they need?

> It's not very important to support two XML parsers, but I do not see
> why this duality is a problem. Sure it's easier to inavertantly
> combust the tree, but the same is true of supporting multiple
> frontends.

This is not about the XML parser per se, it is about deciding
that we want to support standard components for all of AbiWord.

Thomas
-------------------------------------------------------------
Thomas (toe-mah) Fletcher QNX Software Systems
thomasf@qnx.com Neutrino Development Group
(613)-591-0931 http://www.qnx.com/~thomasf



This archive was generated by hypermail 2b25 : Mon Feb 05 2001 - 11:24:25 CST