Why XML?

Why XML?

[note: paragraphs reformatted to fit on slide]

To: "Pete Cordell" <pete@tech-know-ware.com>, <confctrl@ISI.EDU>
Subject: RE: [sdp-ng] Encoding SDPng messages using UMF
From: "Christian Huitema" <huitema@windows.microsoft.com>
Date: Fri, 8 Jun 2001 09:39:27 -0700
Sender: owner-confctrl@ISI.EDU
Thread-Index: AcDvP1amHX72K047Suy+kTYzoya+iAA+DJxA
Thread-Topic: [sdp-ng] Encoding SDPng messages using UMF

 

If, at this date and time, you want to not use XML, then you need an extremely strong case. XML is well understood, there are many support tools, and many more are in development. The W3C is producing a schema description language which is considered adequate for many business applications, many of which are way more complex than SDP.

The talks about ASN.1 are just that -- talks. The only possible advantage of ASN.1 is the size of the messages, but even that is debatable. On the other hand, the cost is very well known: you need specialized parsers and libraries, you cannot easily use text tools for debugging or monitoring purposes, and the syntax is hard to understand and a pain to extend. Most of the proponents of ASN.1 actually propose some variation of it, which is even worse, since it would require even more specific tools.

The main inconvenient of XML is that it can be bulky. I am not convinced that this is an actual problem: SDP is used for describing multimedia sessions, that normally last a few minutes and carry at a minimum several tens of kilobytes of media; the media stream dwarfs the signaling stream by orders of magnitude. If it is an actual problem, then we can indeed use compression. In fact, we can safely assume that other applications will be hurt before us, and that we will get generic XML compression tools sooner or later. All in all, that should not be a big problem.

Let’s not be silly. Just pick XML.

-- Christian Huitema

http://bmrc.berkeley.edu/mhonarc/openmash-developers/msg00315.html Links to an external site.


Transcript

[slide204] So the first question is, why XML? And Christian Huitema, one of the people behind IPv6 said basically, hey, it's maybe inconvenient, but it's really easy to read, and it's much better than ASN.1. ASN.1 is a machine description for machine readable messages for binary, and ASN.1 is used to specify H.323.