[XML-DEV Mailing List Archive Home] [By Thread] [By Date] [Recent Entries] [Reply To This Message]

Re: version numbers and infosets

• To: xml-dev@l...
• Subject: Re: version numbers and infosets
• From: Richard Tobin <richard@c...>
• Date: Tue, 23 Jul 2002 13:26:16 +0100 (BST)
• Cc:
• In-reply-to: <000a01c23240$39958070$6401a8c0@pcukmka>
• Organization: HCRC, University of Edinburgh

In article <000a01c23240$39958070$6401a8c0@pcukmka> you write:

>The problem is, no one knows. Deep inside some application running an
>online banking system is a bit of code that says printf("1.O") where it
>should have said printf("1.0"). You issue your erratum, a new release of
>a parser is produced to conform to it, and hey presto, the online
>banking system goes down and no-one knows why.

The unamended spec allows parsers to reject 1.O.  You might get a new
release of your parser that does that anyway.

>This stuff is mission-critical and not to be tampered with so lightly.

If my bank plugs in new XML parsers without testing them, I'm moving my
money back into the shoebox under my bed.

But as I said, I've been persuaded that we shouldn't require 1.0
parsers to reject documents labelled other than 1.0.  What I would
really appreciate is some feedback on the possibility of changing the
terminology so that such documents are "not 1.0 well-formed, but
acceptable by 1.0 parsers" rather than "1.0 well-formed, but
rejectable by 1.0 parsers".

-- Richard

-- 
Spam filter: to mail me from a .com/.net site, put my surname in the headers.

FreeBSD rules!

• Follow-Ups:
  • Re: version numbers and infosets
    • From: Elliotte Rusty Harold <elharo@m...>

• References:
  • RE: version numbers and infosets
    • From: "Michael Kay" <michael.h.kay@n...>

• Prev by Date: RE: version numbers and infosets
• Next by Date: Marketing ...
• Previous by thread: RE: version numbers and infosets
• Next by thread: Re: version numbers and infosets
• Index(es):
  • Date
  • Thread