[Top][All Lists]

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

Re: XML idea

From: Nicola Pero
Subject: Re: XML idea
Date: Wed, 7 Jan 2004 15:17:30 +0000 (GMT)

> > > Well, they are free to send patches that implement all this stuff. The
> > > patches may or may not be accepted; if not, they can work on something
> > > like PortabilityKit, or even fork.
> > 
> > If I write an implementation of NSXMLParser will you refuse to include
> > it in -base ? 'cos I am not going to bother starting if so...
> From what I've seen of its interface, and because I don't think that
> anything XML belongs in -base, I would most likely argue against its
> inclusion. However, the final call is Adam's, not mine.

The issue is a bit more complex for me, and the need for standards play an
important role.

Generally, I think that we should try to keep our core libraries down to a
small, good, well-designed and hopefully standard set of APIs common with
other OpenStep systems.

In this light, GSXML is not a particularly strong candidate for inclusion
in gnustep-base: its API is totally non-standard (and the design is not
terrific - it's just a wrapper around libxml2), and there are major users
of gnustep who use other XML libraries/APIs with gnustep - some use
libxml2 directly, some use OGo/Helge's XML libraries, some use Marcel's
ones.  The existence of all those widespread alternatives make the case 
for including GSXML in gnustep-base weaker.

Said that, it would be very nice if OpenStep systems had a good,
well-designed standard API for doing basic XML things such as SAX parsing,
and it would be in our interest to help establish such a standard and
support it in -base.

Judging by the emails, NSXMLParser seems to be doing good in terms of
acceptance, but not very good in terms of design and quality.  If
NSXMLParser had a good well-done design, I think we'd definitely look at
implementing it into gnustep-base.  Just think how much easier life would
become. :-)

Given it's not that good, I'm not sure ... maybe we should hope a better
design comes out, or be pro-active and offer Apple a hand with the design
(would they be interested in our help ?) in order to establish a good
standard - we've got some very good designers and programmers (and some of
them have already written entire XML APIs and frameworks) who might be
very much interested in having a good NSXMLParser and so interested in
discussing it and its details.

Maybe we could look at NSXMLParser, figure out what's wrong, and produce
an API spec (and an implementation wrapping libxml2 or expat) trying to
talk to some Apple people and see if they'd read it later and consider it
for their system as well.

... don't know, just brainstorming.

reply via email to

[Prev in Thread] Current Thread [Next in Thread]