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

[Peter Murray-Rust <peter@ursus.demon.co.uk>]

At 11:13 PM 6/8/00 +0200, Eric van der Vlist wrote:
>Peter,
>
>I fully agree with what you've written, and I'd like to add some few
>comments...
>
>Peter Murray-Rust wrote:
>
>> Some people commented that it was a pity that high-quality code like XT
>> should lapse. Having written tools myself, I sympathise, but there is also
>> the realisation that - at a certain point - energies are often spent better
>> elsewhere. It is particularly valuable that we are seeing virtual
>> communities who are looking to pick up on the efforts of individuals or
>> companies - xml.apache.org is an excellent and valuable example.
>
>There is a vision of open source developers as people who must support
>forever any line of code they have delivered.
>
>We'll discourage everyone considering developing open source if we
>wanted to enforce this assumption !
>
>An open source developer may, on the contrary, feel relatively free to
>leave what he has developed : if his development was useful enough,
>other developers will probably maintain it. If it was not, why should he
>bother ?

Agreed. For those unfamiliar with the OpenSource effort it's worth reading
Eric Raymond's "Homesteading the Noosphere" - I assume available at
http://www.opensource.org. ER makes it clear that the *implicit* ethos is
that the author of OpenSource has control over the code until s/he releases
it. Apart from blatant theft and misappropriation the difficulties come
when the release is implicit and disorganised. JJC has done precisely the
right thing in identifying his commitment to XT.

>
>> <snip/>
>>         What we are starting to see is the pattern of an individual
building an
>> early implementation and then a group, often with CVS, taking on the
>> development phase. I'd urge XML-DEVers with time and enthusiasm to consider
>> how they might help in communal efforts.
>
>See you on 4xt then (nothing won't happen without your help).

I assume you mean you (plural) - mailing a subsection of the list
membership. I personally support and applaud 4xt but have to ration my
activities.

As a "middle rank" OpenSource developer (CBMT [collaborative biomolecular
tools], costwish, JUMBO, VHG, CML...) and - through XML-DEV - catalyst of
others I shall briefly indulge myself and hope the following is useful. At
least it gets it in the archives and highlights the pluses and minuses:

in 1993/4 Henry Rzepa and I realised that SGML was required for the 2nd
generation web - HTML was spectacular, extraordinary valuable and fragile.
I started learning SGML and - if the archives are still kept - you can see
my naive and semi-literate beginnings on comp.text.sgml "How do you include
a file in another?" (typical newbie question). For our own work we needed
graphics, and the only reasonably choice at that time was tcl/tk.
Simultaneously I discovered:
	- sgmls from JJC (which produces ESIS - an alternative serialisation of
structured documents)
	- CoST from Klaus Harbo and later taken up by Joe English [an excellent
example of ER's protocol]. CoST took ESIS and applied queries and
transformations; essentially a mixture of XSLT, XPath, and SAX.

	I needed graphical output - for chemistry and for the tree - and therefore
wrote costwish (wish is the basic tcl/tk shell), which was a C++ and tcl
library to extend CoST. It was primarily used by me and colleagues, but
also was used by a linguistics professor and some terminologists.


costwish was a 1995 proof of concept as advanced as what I currently have
in 2000. It lapsed because there had to be a different *.exe for each
platform and in 1995 you could not expect people in a vertical domain to
compile and install C++ code, and I couldn't - as a part-time voluntary
activity - maintain different exes for machines I didn't possess. So the
pre-java  era drastically limited *unfunded* OpenSource. Computer Science
was probably the only area it flourished. [We had really only one example
in chemistry - RasMOL - a molecular viewer and certainly some of that was
effectively funded by the author's employer.]

When I discovered Java (early 1996) I rewrote the CBMT library and costwish
in Java. Effectively these ended up as JUMBO. JUMBO is often labelled as a
CML browser; in fact it is not; it is [the first] XML browser, based on a
DOM-like structure. [It pre-dated the DOM]. I also had to write all the
widget stuff (trees, events, layout) in AWT1.02 - those who know what that
means may sympathise. Jon Bosak very bravely demonstrated JUMBO at WWW6 in
Santa Clara, where XML made its first big public impression.

JUMBO1 couldn't survive because of AWT1.02, so I rewrote JUMBO2 as Swing
0.* [*=horrible early release]. It was designed to allow both display *and
editing* of XML, including in DTD-specific ways. It will validate content
on entry, including non-textual input, and is therefore AFAIK ahead of most
efforts in DTD/Schema-driven editors. [Again, because it was ahead of the
Schema spec I had to invent my own].

There have been a lot of downloads of JUMBO and I regularly get book
publishers asking for permission to include it on CDROMs. I have hoped that
it might generate interest from other OpenSource people, but this hasn't
materialized. I have had sporadic offers of help, which when followed up,
lapse. Example: "I am in Newfoundland; We are building an XML system and
have decided to use JUMBO. We have allocated 2 programmers to this", "Great
- how can I [PMR] help?" [null reply]. I have now become accustomed to
offers of help where I never get a followup - why do these people do it?


I have launched several memes (*with working prototypes or specifications*)
into cyberspace and reckon that:
	- 10% take off pretty quickly. Chemical/MIME, XML-DEV are in this category
	- 40% are developed in parallel (and often better)
	- 30% are ahead of their time (for whatever reason). The difficult thing
is to know whether to keep hanging on in. JUMBO, CML and VHG are in this
category. CML and VHG are starting to take off. "Their time" includes
technology, culture, expectations, etc. SAX, DDML/XMLSChema are examples on
this list where the first efforts were ahead of their time.
	- 20% are half-baked and deserve all the inattention they get

A necessary (but not sufficient condition) is that the launcher has to
stick with the meme and work extremely hard for no apparent or immediate
reward. There is a limit to what an individual can do. Most have some day
job which requires their top priority, so hanging on is a lot of effort. My
current day-job is being run down, so I have to put additional energies
into earning a future living - probably partly in XML - and I am sure that
others know the same feeling. 

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? 

Clearly the culture is essential, and here I am extremely grateful that
many others have shown that OpenSource is economically valuable. TeX, GNU,
Linux, tcl, perl, apache all have sizeable industries associated with them.
It is less clear in verticals (such as chemistry, healthcare...) but at
least I can point and say - "they" show it can be successful.  So what
helps an OpenSource project make it?


	- the goal must be very clear. The attraction of g++, Xerces, Xalan, FOP
is that they are all tied to specs.  A more general concept like "XML
editor" is tougher to define.
	- There needs to be some permanency. The key point is the transition from
individual to communal group (JJC->x4t, James Tauber->FOP). The
organisation has to be credible
	- It has to serve an unfulfilled need. Thus XT filled a vacuum when it
first came out - the only OpenSource XSLT processor. Xalan is now also in
that market. I personally believe it is very important to have more than
one OpenSource offering in a given category because they will reveal
inconsistencies (in specs, interpretation and implementation). And the
implementers need all our support.
	- it helps if it is likely to have an extended future - partly because of
the need for OpenSource offerings and partly because it can be included in
commercial products.
	- it has to have an early offering that is portable/distributable and does
something useful. FOP does this; it doesn't do everything I want but it
does something I can't do by other means. [I will write in support of FOP
elsewhere].
	- the community should support the effort *as OpenSource*, even if they
themselves are unlikely to participate. Some members may feel FOP is
misguided or unlikely to succeed but I hope they will avoid demotivating
the current FOPers.

Personally I would like to see an OpenSource in Schema-driven, graphical
XML editors, including non-textual material. I've mentioned this before on
this list. If - in any way - JUMBO can act as a catalyst - fine. I'll
probably keep mentioning it. If Apache, Mozilla or 4xt or whoever have
independent ideas, great - I don't care how it happens. But at present I am
frustrated in not having that functionality!

	P.

	 
>
>Thanks
>
>Eric
>-- 
>------------------------------------------------------------------------
>Eric van der Vlist       Dyomedea                    http://dyomedea.com
>http://xmlfr.org         http://4xt.org              http://ducotede.com
>------------------------------------------------------------------------
>
>***************************************************************************
>This is xml-dev, the mailing list for XML developers.
>To unsubscribe, mailto:majordomo@xml.orgBODY=unsubscribe%20xml-dev
>List archives are available at http://xml.org/archives/xml-dev/
>***************************************************************************
> 

***************************************************************************
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/
***************************************************************************