Another plug for XML Catalogs (and caching)

The W3C systems group posted a blog entry the other day about the caching of DTDs and schemas. The failure of some XML software to use caches wisely is causing unbelievable amounts of traffic on the W3C site: in some cases, the same IP address is requesting the same DTD file hundreds and thousands of times in the space of a few hours.

The blog has good pointers to resources about using HTTP caching well, and about XML Catalogs.

I’ve said it before, and I’ll say it again: every piece of software that works with XML ought to use XML Catalogs. By all means allow the user to turn it off, but support it, and turn it on by default. The main reason is: it makes the life of your users easier. And the kind of problem discussed by the systeam blog post is one more reason.

One thought on “Another plug for XML Catalogs (and caching)

  1. An even better solution, in many cases, is to ignore the DOCTYPE declaration completely. After all, usually you don’t want to check whether the document follows the DTD it claims to follow, but whether it follows a particular DTD. In these cases, the solution is to include a copy of that DTD in your software and to always validate against that DTD. This removes the whole issue completely (in those cases where this is what you want).

    Otherwise you run the risk of your RSS reader blowing up on an XHTML document that validates perfectly against the XHTML DTD…

Comments are closed.