Lists Home |
Date Index |
Rick Jelliffe wrote:
> From: "Robin Berjon" <firstname.lastname@example.org>
>>However if you look at the broadcast industry, there are cases in which
>>the application does a lot more than just throw some XML around and yet
>>using textual XML would be a serious hog (in fact impossible).
> I kinda don't get it. XML is a web technology, and HTTP allows compression.
> If people don't turn on compression on their webservers, that is their problem.
> If people want more efficiency, they could write serializers that generate
> compressed data directly, and the reverse at the parser end.
Hmmm, well I don't know what you're not getting because what you're describing
is exactly what we're doing :) When HTTP is used BiM can be used as a simple
content coding. When the XML is not already in textual form (but instead in SAX,
DOM, etc) we write serializers that generate binary infosets directly.
I was simply pointing out the fact that XML's verbosity is an acknowledged
problem in a number of spaces, that it is perceived to have sufficiently
interesting features to look into using it nevertheless, and that solutions that
address the vrebosity exist.
> I still have never seen any numbers that suggest that the cost of parsing
> XML is not trivial compared to the cost of object creation.
On what platforms have you seen the numbers you did see?
> Especially when it is quite possible to write light-weight parsers that omit
> most WF checks. If you can trust that your data is WF, you don't *need* to
> check for name-correctness, for example. You can just tokenize
> using space separators: this is appropriate for tightly-coupled systems or
> routing systems, for example.
Yes, but unfortunately that's not always the case as some systems have a wide
variety of content creators.
> XML was designed for this: perhaps the emphasis on Draconian error-handling
> for WF and validity has obscured that XML was designed also to be useful for
> the Desparate Perl Hacker
True, but for that to be generally admitted we need to wait for Perl to take
over the world ;)
Robin Berjon <email@example.com>
Research Engineer, Expway
7FC0 6F5F D864 EFB8 08CE 8E74 58E6 D5DB 4889 2488