Mike,
I've not heard of this. Have you looked on http://bugs.php.net? there seem to be several with segmentation faults and Apache httpd 2x and mentioning sessions, it may give you some ideas. The manual says that a session is not actually written until a $_SESSION variable is registered, even though session_start() is called. Do you still get the error if you register a variable e.g. $_SESSION['myvar'] = 'abc'; after session_start()?
I assume you are using Apache httpd in prefork rather than worker mode? (httpd 2 + php not yet prime time?).
Out of interest, from the ptt-bin part of your url, are you running this as cgi rather than a module?
I feel there must be something unusual about your set up, the fact that you are getting this fault. I've been doing a php development with php 4.3.3, Apache httpd 2.0.47 on Fedora Core 1 using sessions and not come across this. I've also used earlier versions without problems. I can understand the frustration, but if you feel that dropping php because of this is the right course of action then the phrase "throwing the baby out with the bath water" comes into my mind.
Sorry that I can't put my finger on the specific solution you need, but I hope that I have suggested some useful avenues for you to look into.
Let us know how you get on. HTH Chris
I've fixed the problem (well, I think), but I haven't figured out what it was. I'm working on creating a simplest case. I know that it has something to do with making a reference to a session variable, as in:
$loc=&$_SESSION['local'];
And then subsequently changing it, in some specific way under some specific circumstances that I haven't isolated yet.
Answering your questions:
1. I'm using apache in it's default mode (so pre-fork or child, whatever it does out of the box)
2. php in module mode. (The reason I have "php-bin" is because I want only that directory to execute php scripts, not the whole site- is that possible to do? I haven't looked at how yet in the httpd.conf, I think it'll be easy. I'm using php only for applications, not the "inline html" stuff, I don't like that.)
3:
I feel there must be something unusual about your set up, the fact that you are getting this [...] can understand the frustration, but if you feel that dropping php because of this is the right course of action then the phrase "throwing the baby out with the bath water" comes into my mind.
There's nothing unusual about my setup, it ended up being something unusual in my code (although not a "mistake" and nothing forbidden by the manual as far as I know (?)) and likely a php bug (aren't all segfaults bugs, by nature? I guess excepting infinate loops and dividing by zero etc...). And as for the baby and the bathwater- would you put code that had this problem onto a site, if you couldn't fix it? or would you switch to something else? (What I said was "I need to insure that this doesn't keep happening on the production site or we have to bail out of php"- you really don't agree with that sentence?)
Thanks Chris, I'll try to make an isolated case for the sake of the php project and report back. thanks,
mike
