[thelist] good practice META tags?

Keith Davis cache at dowebs.com
Thu May 24 17:46:08 CDT 2001

deke wrote:

> HTTP 1.0 specifies "pragma" as a header to be sent by clients
> when requesting a fresh copy of a page, not by servers.
> Can you name *any* circumstance in which the illegal "pragma"
> header works and these headers don't?
> <META HTTP-EQUIV="Expires" CONTENT="Thu, 01 Dec 1994 16:00:00 GMT">
> <META HTTP-EQUIV="Cache-Control" CONTENT="no-cache">

Yeah. I can't see how you would use a meta tag to tell the browser to
keep a certain image, or an external javascript, or an embedded cgi
response out of the cache while caching the rest of the page. But,
sending the Pragma: no-cache header along with those page components
will do the trick. 

You can run across Pragma: no-cache occasionally in CPAN modules and
from what I've read it has been used by Perl programmers for years as
the preferred method of keeping content and/or full pages out of the
cache. (Maybe that's because we're always looking for the least verbose
way to do something.)

The way I understand it, the browser starts to read a page from cache,
discovers it has expired, requests (via pragma?) the lastModifiedDate
and/or fileSize from the server, and then may continue to display the
page from its cache if no change has been made in the interval. Which
means that the server has to return instructions on that pragma request,
and a pragma header originating from the server is simply preempting the
request. Appearantly "no-cache" is one of the responses that the browser

Delivering dynamic components to realestate on static web pages is
becoming a very large industry. Managing what gets cached in the mix is
a problem in this area that has many solutions, each with many
proponents. Can you offer a URL that points to the standard you quote??
Or better, what kind of data is exchanged in a pragma request/response?


More information about the thelist mailing list