OASIS Mailing List ArchivesView the OASIS mailing list archive below
or browse/search using MarkMail.


Help: OASIS Mailing Lists Help | MarkMail Help



   RE: [xml-dev] There is a meaning, but it's not in the data alone

[ Lists Home | Date Index | Thread Index ]

From: Jeff Lowery [mailto:jlowery@scenicsoft.com]

>> DOCTYPES are fine for DTDs as long as you accept 
>> extensibility in any instance that includes the 
>> DOCTYPE definition and/or own the system reference. 
>> A PUBLIC id specifies the owner organinization and 
>> the system ID tells you where to locate your version 
>> of that.  

>Well, since I don't use DTDs myself, it may be possible that I'm missing
>some subtleties. I get your drift, though. 

And they tell you what the root of the tree is just 
in case it isn't the first thing in the file (it's 
an old SGML hack that XML avoids).

>> Why doesn't that work?  Are you really 
>> looking for someone to make your choices for you?

>But I don't want to use DTDs. You can't make me.

YES I CAN!!! i just don't feel like it today.  But since 
the message was "Doctypes? I don't think so", all I 
am saying is they work fine for DTDs given you trust 
the sender or you own what is at the location identified 
by the system id.

>> Levels.  Agreements are usually layered if negotiated. 
>> Blind exchanges should not be the way the web works. 

>Ummm... not quite with ya there. Certainly, I can ignore or passively store
>information I don't understand, for later regurgitation. Not that I do that
>on this list.

If you do that, then the only agreement in place is with yourself.  No 
problem with that.  A DTD enables extensibility in the internal 
subset.  Where would the agreement not to do that go?  In past 
cases, in the contract or cited spec.  It could be in a comment, 
but just citing the W3C spec won't work.  The NIC folks miss the 
subtle points like that.  Anyone who blindly cites a PUBLIC ID 
or authority misses points like that.  On the other hand,  
it works reasonably well because the humans are reasonable 
critters and handle the details.  One only has to be paranoid 
about the internal subset if one is really worried that one 
or two documents will screw over the whole process.  I can 
make a case for that, but it is the paranoid case.
>> And if you plan to send the semantics with the message, 
>> Java and PDF are there to serve your every need. :-)

>Oh, c'mon. 

Yes, it kills the fun of the movie if someone tells 
you the ending.  We went over these discussions endlessly 
in the MID project; it came down to package the semantics 
(design a programming language) or send the data and trust 
the documents (design a data language).  The solution where 
you point off to some other document just treats data as a 
yet another dereferenceble resource.  It doesn't solve much 
that can't be done packaging more data or just sending 
an executable.  <? ?> are just more data made syntactically 
opaque to the application language so your file can carry 
another message to another application (yourSystem).  That's 
why they are the basis for archforms.  Tell me the difference 
between an arch form and a load of attributes on the root.
>> Why is this debate still being held every year?

>It makes me wonder why you even bother, Len. :-\

Instructions on popsicle sticks.  We all learn semantics 
somewhere and Huntsville doesn't have big automated 
road signs, so the LA PUBLIC systems are out of the question.



News | XML in Industry | Calendar | XML Registry
Marketplace | Resources | MyXML.org | Sponsors | Privacy Statement

Copyright 2001 XML.org. This site is hosted by OASIS