About URL\'s and paths

[pinup57]

Hi

Having struggled for about a month trying to integrate FULLY the huge potential of this script in a CMS, I'd like to communicate my feelings about this.

I think there's "a whole lot of shaking going on " regarding all the url's and paths, not only in PFC itself, but also in xajax.

Basically, all is well and fine as long as we have the chat in the index page of a website. But as soon as we start to try integrating it in another context, especially when there are query strings involved, things start to malfunction.... It might work for basic functionalities, but as soon as we start adding more complicated stuff, like querying databases to set nicks and metadata, there's trouble...

My question is: why are there so many "we'll calculate this url from the 'this_path parameter' " all over the place in the core files (and not only in pfc, but also in xajax core files). Most of the time, imho, this gives totally erronous results. I know some of these, in pfc, have been corrected by Kerphi in the next release, but still there are issues with this.

I went through every single line of code trying to solve my problem, and the more I get into this, the more I get convinced that there should be some serious simplification here.

I honestly think that most issues (if not all) in pfc are about to be solved by Kerphi, but still there are issues regarding xajax, who relies rather heavily on _SERVER queries to set paths and URL's, which might lead to malfuctions when the script is integrated in a CMS, where every URL is index.php?querystring. (Like "Ok, so the server says this script was called from index.php, so lets try to find if there was a querystring attached, if so, append it and calculate from there..." and it might happen that the server returns a bunch of html instead of the expected response to a php request) We finish to have instances of objects in different namespaces, and many features just don't work.

Mind, although I'm a confirmed C++/Pascal coder, I'm a beginner in php stuff so it's possible I'm totally wrong here.

But it's my intuition that all the path and url stuff should be seriously simplified and centralised!!!

Just my 0.02,
Dirk

PS I like this script very much, it's a helluva job that's been done, thanks Stéphane! Please don't get me wrong... Your stuff rocks! Just trying to get the most out of it......

[lordmundi]

man... i'm glad i ran across this post. I have been fighting with integration of this script for about a week now, with many late nights.

and at first, i couldn't understand how the simple demos could work so well with so little code, but mine was broke and it was just as simple. But I am starting to see exactly what the previous poster was talking about... as soon as I move the instantiation/constructor out of the main php file used as the page, things start to break. And it's really bad because it looks like it affects these XAJAX paths which I don't really understand, and therefore can't troubleshoot easily.

if the product wasn't so good feature wise (as evident in the demos) I would have dropped it almost immediately... but I really want to get it working.

just wanted to vote with the previous poster that some of the simplification in the code to autodetect paths and things have made it almost impossible to integrate into a website when the chat is not instantiated in the index page.

FG

[phpfreechat]

I think you should have a look to demo21. It explains how to hardcode urls in order to override automatic url detection.
For example: server_script_url parameter is very usefull to integrate the chat other systems like CMS.