Re: What are the problems with IDML? (fwd)

MegaZone wrote:
> 
> Once upon a time Doug Donohoe shaped the electrons to say...
> >Which is what we have done (more or less) in our "debate" page:
> >       http://www.identify.com/welcome/idml_v_meta.html
> 
> Which makes meta look quite a bit longer - mostly because of the, IMHO
> unfair, line breaks after <META and between NAME and CONTENT.  That added
> a LOT of lines.
> 

I think this is a ridiculous argument (IMHO).  However, take a look
at this one, with the "unfair" line breaks removed :^)

	http://www.identify.com/welcome/megazone.html


> Otherwise, looks fine.
> 
> >However a couple of open questions:
> >       1) Which would you prefer to write as a user (after all,
> >          it's the user we're trying to help!)?
> 
> META - I perfer using tags that validate, AND it is a very very small
> amount of aditional work.  Besides, we're looking at having tools doing
> that, you put in the data, the tool spits out the tags.  I could write a
> Perl script to do it.
> 

What tools?  Are you referring to enhancements to existing
HTML editors?  I know most support META now, but are they going
to add support for the various META schemas that people use?
While we're on the subject, does anyone know of schemas that
define a data model? (I know of dublin core, but that just defines
the container, not the values).

BTW, we do have a tool for spitting out IDML (the IDML Generator)

	http://www.identify.com/welcome/generate.html


> >       2) Which would you prefer to parse as a robot writer [1]
> 
> Cacatenating tags together isn't hard at all.  As a robot writer I would
> rather set it up to parse META and then have an config/rules file that
> I can extend to use EVERYONE's tags.  That's a key - there is going to
> be more than one way to do this, and if they are all META based it would
> be simple to write the bot to pick what it wants using the same engine.
> With other tags, not so simple.

True.  Is there a specification or draft which describes what
these config/rules files look like?  I imagine that would be
complex to specify all the types of data validation one might
want to do.  Has anyone done any work in this area?

I'm sure you know, but I still feel that putting stuff that belongs
together in seperate META tags introduces more problems.  For example,
when writing HTML, I mess up tables far more often than <IMG>
tags because tables require beginning and ending tags.  In IMG,
the only thing I usally get wrong is forgetting a closing quote.
All I'm saying is that for a while longer, people are going
to be writing this stuff by hand -- and with copying & pasting and
moving stuff around, you are more likely to mess something up
when you have to keep 8 tags in a block versus one.  Perhaps this
will not be a problem when the HTML editors mature.  Still, are
the editors going to have support for all the various META schemas
out there?

Clearly, I'm biased in one way (maybe because that's the
way our robot already works? :-).  Nonetheless, I'm investigating
what it would take to adapt the crawler to accept tags in META
format.  If we do embark on that journey, we'll be able to
accurately say which way is easier (because we'll have done it
both ways).  If someone has experience using META for more than
just keywords and description, please let me know (and send
the URL!).

> 
> >    If you are going to have to change your robot anyway, why not
> >    consider using IDML format instead (after all an IDML tag
> >    is not syntactically different than an <IMG> tag)?
> 
> Because I would already need to muck with META for the systems in use NOW,
> it would just be another rule set for the bot.
> 



Cheers and thanks for your input,

-Doug


-- 


J. Douglas Donohoe
-------------------------------------------------------------------
Emerge Consulting			   Chief Technology Officer
415.328.6700			              http://www.emerge.com
donohoe@emerge.com			    http://www.identify.com

Received on Friday, 23 August 1996 12:29:08 UTC