On Oct 29, 2006, at 9:05 PM, Scott Kaplan wrote:

> I tried this on multiple browsers and on multiple machines. I have
> the browsers all setup to not cache anything.

Never trust a browser. Try using wget/curl.

> If I wait more than 1 minute between visiting the same dynamic
> page, I don't have the problem.

Are you testing locally? That sounds a lot like caching, either in
the browser or on a proxy server between you and the server ( a lot
of ISP's do that )

> If I visit a recent dynamic page within a minute, I have to hit
> refresh and only then do I get the most recent stuff.

That sounds exactly like browser caching.

> This is most annoying and not web safe. I am convinced that there
> is a
> configuration setting somewhere that is telling Apache to not
> interpret/compile any script that has been recently visited (within 1
> minute in my case). I searched through httpd.conf and I couldn't find
> anything.

If Apache were to blame, hitting refresh wouldn't have an effect.

On Oct 29, 2006, at 9:17 PM, Dondi M. Stroma wrote:
> We will need more details though; is this your own Perl handler, an
> Apache::Registry script, or something else?

Agreed. Also, what dynamic content is changing: sql content ?
dynamically generated stuff? or did you mean that you changed a
script, and you expected it to be different (then you get into
Apache::Reload and multiple server instances issues)

> How are you generating the response headers? Apache::Request?
> CGI.pm? Do you "use strict" and "use warnings"? Check for any
> "will not stay shared" warnings and make sure you don't use "my"
> variables in a subroutine that were declared outside of it.

from what he described, it doesn't sound like it could be scoping- -
but it never hurts to double check. make triple sure that you're
using warnings, and everything is scoped correctly.

there's a slim chance that you're experiencing this:

multiple apache children
poor variable scoping

request 1 is on pid X , everything happens fine
request 2 is on pid X , poor scoping gives same content as request 1
request 2 is on pid Y , everything happens fine ( new content is
generated )

it doesn't sound like that's happening, but its possible. usually
stuff like that manifests itself with Apache2::Reload issues.

> Finally, you may wish to try setting "Pragma" and "Cache-Control"
> headers.

+1 on that.