* Sumeet Singh <[EMAIL PROTECTED]> wrote:I understand how a sub-request is different from an internal-redirect. However, my point was that of recursion. If you have any light to shed on the original question I asked (that whether it is legal to invoke ap_internal_redirect from an output filter) I would really appreciate that.
[internal redirects in output filters]
Yes, I also find it unsafe. However, note that some standard modules do that too. For example mod_include runs sub-requests from within its output filters. In my opinion this shouldn't be allowed, because with the wrong configuration a web-admin could cause the server to go into recursion. But since mod_include is a standard module, I guess what it does is considered legal.
Subrequests and internal redirects are entirely different things. A subrequest can be done from almost everywhere and affects just the data flow (if at all). An internal redirect affects the current _control flow_.
nd
-rgds,
ss
p.s.
And by the way, I don't agree that sub-requests are *entirely different* from internal-redirects. For example if you did a subrequest-uri-lookup followed by a ap_run_sub_req you would still end-up running all the stages (except post-read-request) and generating data and sending it down the output filter stack. Same thing is done by internal-redirect. There are subtle differences like the state of r->main/r->next/r->prev, filters inherited by the new request, running or not running of post_read_request and quick-handler stages, but finally both are recursions. And both the above generate data going down the output filter stack. After calling ap_internal_redirect you are supposed return immediately with a OK/DONE, but what if you didn't - then you would essentially be doing what people do with subrequests.
If you wish to continue this discussion, please start a separate thread.
