xml-dev - Re: XT, OpenSource and altruistic effort

[Rick JELLIFFE <ricko@geotempo.com>]

Peter Murray-Rust wrote:

> I have discoursed at some length because I think there is a real issue here
> for XML-DEV. It is clear that high-quality OpenSource projects require a
> virtual collaborative group - what is the best way of promoting this?

> So what helps an OpenSource project make it?

Actually, I wonder if it is true that high-quality projects require
collaborative programming.  I would say quite the reverse: we need to be
able to reduce collaboration interdependencies to a minium: we need to
have technologies layered enough that a motivated and talented
individual programmer can finish off a layer in less than a month or two
(full-time).

In other words, what I am learning with Schematron is that the initiator
of a project has to provide adequate proof-of-concept and implementation
and infrastructure, which can be used as a base so that others can
submit enhancements.  If the kinds of enhancements expected can be
distilled into  general interfaces, then the collaboration is one of
collaborative publishing (e.g. CNET for perl) rather than of
collaborative programming per se.

If we have a look at the larger, collaborative projects, it seems that
they benefit from large sponsors (e.g. Apache, Mozilla); having two
programmers on a payroll is a good way to promote collaborative
programming.  Otherwise, most programmers will try to solve specific
problems only. I think the bite-sized chunking aspect is very important.

When a software package gets to a certain size, then bug fixes and
non-extension enhancements represent quite a workload on the original
developer. I think Dave Ragget's tidy passed that stage sometime last
year, and perhaps James feels the same about XT.  (It is also important
to note, for the people who complain that W3C specs are inferior to IETF
specs because IETF requires working implementations, that most W3C specs
have (99%) working implementations available before going to
Recommendation: look at XML, XSLT, etc. It seems that XML Schemas will
definitely follow this: that is what Candidate Recommendation period is
for.)

One area where Open Source projects have been week has been
internationalization. Sometimes it seems that Open Source has been a
disaster for i18n. Part of this has been that LINUX and the standard C
libraries have been so poor, but part of it is because adding reasonable
internationalization has tended to move a project beyond the scope of a
single programmer's effort. If we look at Emacs, it is really good for
providing a package mechanism to let the single man-month programmer or
weekend hacker build a tool. But it has taken years for the MULE
(MULtilingual Extensions) version of Emacs to get grafted into the
mainstream release. 

Fortunately, things have changed rapidly: the two biggest technological
contributions, I would say, are Java and C++'s IBM's International
Classes for Unicode. If anyone is approaching a C++ Open Source project,
I would strongly recommend they start by adopting the ICU International
Classes. The URL is
http://oss.software.ibm.com/icu/
(If you need more info on Unicode, Tony Graham's book is excellent.)

Rick Jelliffe

***************************************************************************
This is xml-dev, the mailing list for XML developers.
To unsubscribe, mailto:majordomo@xml.org&BODY=unsubscribe%20xml-dev
List archives are available at http://xml.org/archives/xml-dev/
***************************************************************************