Re: OO and scripting

Subject: Re: OO and scripting
From: David LeBlanc <whisper@xxxxxxxxxxxxx>
Date: Tue, 11 May 1999 18:15:33 -0700
At 05:39 PM 5/11/99 -0500, you wrote:
>Dave LeBlanc said:
>> I personally think <script> ought to be in XML itself - I can imagine 
>> using it to allow a document to convey (via it's DTD for example) how 
>> a receiving processor is supposed to manipulate it. Very object 
>> oriented.
>I don't think that your idea will take hold.
Based on the existing implementation (of XSL in IE 5) on the market today
it already has done so. Further, I think that various ideas circulating re:
RDF, CDF and the schema/data proposals, tend to argue towards an OO trend.

>I have been told that Tim Berners-Lee was late for a keynote at one of the
>early WWW conferences because he was in a big argument with Alan Kay (of
>Smalltalk fame). Alan was trying to convince him that the Web was totally
>broken and should instead be based on intelligent objects floating between
>servers, sort of like RMI or a big, distributed smalltalk engine.
I guess that Berners-Lee and Kaye had a difference of opinion similar to
what I think you and I have.. what is a document, passive or active? OO was
decried at it's inception too.. and today's dominant languages are, to some
extent or another, OO (admittedly, languages like VB, Pearl and Python give
it more lip service then true OO).

>I wasn't there but I see this story as incredibly important because it
>demonstrates the two different views of the world. On the one side is
>Alan, data and code are the same -- mix'em up. On the other side is Tim,
>data is data, code works on data. Code can be viewed as data but should
>not be mixed with the primary data.
Well, in response, a knife might be an object best used for slicing things,
but wouldn't it be aweful if it was somehow constrained such that it could
not be used in an emergency as a can opener?

>The two views also existed in the document processing world. Most smart
>people bet on cool stuff like Postscript, display Postscript, TeX and
>LaTeX. A few people said that mixing data and processing (especially
>algorithmic processing) shortens the lifetime of your data. Most others
>figured that as long as it displays correctly it does what it was supposed
>to do and the documents demonstrably displayed correctly.
As I understand it, TeX/LaTeX are essentially markup notations - isn't any
markup better then none?

>Now the separationists we've won. Postscript is likely doomed, display
>Postscript will hopefully go to hell with it, macro-embedding in Word is
>now widely frowned upon and downright difficult, TeX is used increasingly
>as an output format and those who use it for input are increasingly
>informed about the dangers of doing so (but may have good reasons anyhow).
I'm sorry, but I missed something here.. what danger?
Seems to me that while the "separationists" might be in the ascendent in
the XML reccommendation writing world for now, that the opposing view is
alive and well and creating agents - which may very well use XML or a
derivative as a protocol for communication.

Realistically, I doubt either view is superior to the other. My bias is
towards a peer view of the world where documents can be active/intelligent.

>The maintainers and editors of the XML specification all come from this
>separationist background. It will be a long, long time before they agree
>to add something to the XML specification that takes us back to the
>"object oriented" view of the world. OO is great for building software
>systems that manipulate information but it is not a good basis for the
>information models themselves.
If the maintainers and editors of the XML specification (should we say that
with the same hushed reverence once reserved for the high priests of the
mainframe?) don't build something that people will use then what is the
point? As for their agreement, it is absolutely unecessary... XML is a
recommendation, not a law of god (err, excuse me, should that be a law of
Python?). So far as I know, the only broadly available easily obtainable
implementation (the one in IE 5) contains a <script> tag. Perhaps the
market has already spoken.

>More information:
> Paul Prescod  - ISOGEN Consulting Engineer speaking for only himself

In the end, XSL isn't going to be what is in whatever the committee
reccommends, it will be, to use an euphemism, what "works as coded". The
brief history of HTML suggests that implementors will add whatever
convenience features they desire (for good or bad, it's true). To
paraphrase, specifications rarely surive first contact with the users.

I'd like to say that i'm learning from this dialog (thank you!) and, again,
intend no offense towards or disrespect of anyone.


Dave LeBlanc

 XSL-List info and archive:

Current Thread