Serializations and data structures (was Re: Topic Maps onSQL)

Lars Marius Garshol larsga at ifi.uio.no
Wed Nov 25 17:22:22 GMT 1998


* Paul Prescod
>
>So your fundamental argument seems to be that character data is simple and
>thus more reliable and long-living. I agree. That's why I use XML.

Then we agree here, at least. :)

[skipping other stuff we agree on]

* Lars Marius Garshol
>
> Can you really guarantee [agreement on interpretation]
> for something as complex as groves for decades? There'll be no
> disagreement on the actual sequence of bytes in the files, but their
> interpretation in terms of the abstract grove is another thing entirely.

* Paul Prescod
>
> Disagreement on the actual sequence of bytes in the files is irrelevant.
> If I can't get the moral equivalent of the same grove, ESIS or 
> SAX events out of it that the creators intended then I am working with
> different data (at the logical level) than they are. The fact that I
> have the same bytes is not very comforting if the software that 
> processes it fundamentally misunderstands it. [...]
>
> XML software can't work with serializations. It must work with the
> data model! If that doesn't survive, all is lost. If the data model
> a wishy, washy and implicit like that in the SGML and XML
> specs, then the data is in danger of mild logical corruption as people
> come to understand it differently. "What did Tim mean by this? And 
> this?" If it is a well-defined formalism then that danger is much 
> smaller (eliminated?).

Well, you can't very well transmit the data model itself between computers
or down through time. In other words: you must have the serialization
syntax. The question is: is an explicit specification of the data model 
of any use?

Tim seems to be saying no, claiming that it is 'ephemeral'.

You + Len + Eliot seem to be claiming that without it the chance that we
don't interpret the syntax the same way is simply too great. 

As the argument stands it looks like checkmate to me. Why not turn the
issue around slightly: what would be difficult to do without the data
model? I can think of one thing: interpreting parts of XPointer 
unambiguously (the string and span stuff, for example) as it and XML 
stand today is near-impossible. A data model would hopefully resolve this.

--Lars M.



xml-dev: A list for W3C XML Developers. To post, mailto:xml-dev at ic.ac.uk
Archived as: http://www.lists.ic.ac.uk/hypermail/xml-dev/
To (un)subscribe, mailto:majordomo at ic.ac.uk the following message;
(un)subscribe xml-dev
To subscribe to the digests, mailto:majordomo at ic.ac.uk the following message;
subscribe xml-dev-digest
List coordinator, Henry Rzepa (mailto:rzepa at ic.ac.uk)




More information about the Xml-dev mailing list