About "must understand":

--On Thursday, November 11, 2004 11:01 AM -0800 Tim Bray <[EMAIL PROTECTED]> 
wrote:
>
> Questions for Joe:
> 1. Why do you think this is hard to implement?
> 2. Assuming I can persuade you that it's cheap, would you be OK with it
> even given the lack of use in SOAP?

Speaking as "Joe for a day": We parse no formats with compulsory
extensions. It just doesn't exist for PDF, MS Office, OpenOffice,
WordPerfect, whatever. You want to change it, you rev the version.

Implementing this require a new result from parsing: it is a legal
document, but we are not allowed to tell you what is in it.

Feeds are in a funny place between API and document. But I think
they are more document than API. APIs have "must understand", but
documents don't.

What does a search engine do when faced with a "must understand"
element? Throw away the whole thing and log it? Ignore the spec
and salvage what we can? Which one results in fewer support calls?

wunder
--
Walter Underwood
Principal Architect, Verity

Reply via email to