On Oct 25, 2004, at 4:33 PM, Geoffrey Young wrote:
let's give david a chance to investigate - either to fix or, if a quick fix
isn't obvious, revert the behavior.
if david doesn't respond by, say, wednesday, we (you or I) should feel free
to just revert the change. maybe david is on vacation or something, or just
temporarily behind in his emails.
No, just hacking. Let's see...oh, I get it. I changed it so that it ignored $RedirectOK if LWP was installed. That's not necessarily a good idea, given the goofy way in which this module is written. I've applied a fix to only let LWP handle the call to redirect_ok() if a an array reference was passed to user_agent( requests_redirectable => []). Ugh, that is so ugly!
But this doesn't seem to help t/ssl/basicauth.t. But even if I roll back the changes that test still fails, so I'm inclined to think that it's failing for some other reason.
You know, I'm inclined to remove that stupid $RedirectOK global variable, because you can't tell whether it was set by the user of Apache::TestRequest. This makes it difficult to decide whether or not to let LWP handle the call to redirect_ok(). What say you all to my changing this to a class method, say RedirectOK()? Then it can be smarter about who is doing what to whom. It would require I change the tests that rely on it, but they don't appear to be too many:
% grep -r RedirectOK t t/modules/alias.t: local $Apache::TestRequest::RedirectOK = 0; t/modules/alias.t: local $Apache::TestRequest::RedirectOK = 0; t/ssl/proxy.t: local $Apache::TestRequest::RedirectOK = 0;
Thoughts?
Regards,
David
