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

RE: Managing Innovation

• To: David Megginson <dmeggin@a...>,"'xml-dev@l...'" <xml-dev@l...>
• Subject: RE: Managing Innovation
• From: Elliotte Rusty Harold <elharo@m...>
• Date: Fri, 3 Oct 2003 21:56:02 -0400
• In-reply-to: <16252.39742.367882.848298@m...>
• References: <15725CF6AFE2F34DB8A5B4770B7334EE03F9ED39@h...><1065129219.3933.11.camel@z...><16252.39742.367882.848298@m...>

At 5:40 PM -0400 10/2/03, David Megginson wrote:


>I understand that argument well, but in this case, we're not talking
>about limiting the overall length of XML documents.  Let's try not to
>imagine only the present, but the future as well -- can anyone make a
>reasonable case for an XML element or attribute name longer than 4096
>characters (for example)?


I would not be surprised to see such a thing pop out of some program 
that generates element names using an algorithm. Programs that create 
programs are notorious for finding bugs in compilers that had 
implicit assumptions about what code would look like, often involving 
length of variable names or maximum line length. I wouldn't be 
surprised to see programs that generate XML overflow a few boundaries 
in XML parsers.

There are definitely parts of most APIs that are length limited. For 
instance,  SAX can't handle a comment, an attribute value, or a 
processing instruction  longer than 2.1 gigabytes. Think, for 
example, of a Base-64 encoded movie shoved in an attribute value. 
Completely legal for XML, but SAX can't handle it.
-- 

   Elliotte Rusty Harold
   elharo@m...
   Processing XML with Java (Addison-Wesley, 2002)
   http://www.cafeconleche.org/books/xmljava
   http://www.amazon.com/exec/obidos/ISBN%3D0201771861/cafeaulaitA

• References:
  • RE: Managing Innovation
    • From: "Bullard, Claude L (Len)" <clbullar@i...>
  • RE: Managing Innovation
    • From: Rick Marshall <rjm@z...>
  • RE: Managing Innovation
    • From: David Megginson <dmeggin@a...>

• Prev by Date: Re: Fwd: [e-lang] Protocol implementation errors
• Next by Date: stupid deterministic content model question
• Previous by thread: RE: Managing Innovation
• Next by thread: RE: Managing Innovation
• Index(es):
  • Date
  • Thread