Unless there's further information I can't act on this. For example, I no clue what a malformed header is.

Ok. The story so far.

Installed on NT4/SP4 & Apache and PHP 4.0.4.

I have had to comment out the crypt part in coreAPI.php as there is no crypt function in the WIN32 version of php. The error message I get is in the apache error log and is "[Fri Jan 05 15:56:20 2001] [error] [client 172.18.1.245] malformed header from script. Bad header= INTO mantis_bug_table: c:/program files/apache group/apache/cgi-bin/php.exe".

I have found that the error seems to be produced in report_add.php where it inserts into the mantis_bug_table.

I forgot to add that data is added to mantis_bug_text_table so this parts works but data is not added to the mantis_bug_table.

This is with mysql? Check the $query on line 80 and see if the $result is not 0.

I have now done the following:-

1. Upgraded NT4 to SP6a (Workstation version)
   
   
2. Upgraded to win32 apache 1.3.14
   
   
3. Upgraded to php 4.0.4
   
   
4. MySQL is at server 3.23.22-beta
   
   
5. Re-installed mantis 0.13.1 from scratch and commented out the crypt function call in password check routine.
   
   

I no longer get the malformed header error but get other errors. What happens is the php.exe will Dr Watson on the server when I use some functions in mantis and not others. An example is:-

Select Report Bug from the menu and php.exe will Dr Watson on the server with an access violation (0xc0000005), Address (0x7800d5ca).

I am certain that the is something in the report bug script that the version of php I am using does not like. If anyone can think of anything I need to check please let me know.

I have keystone running on the same setup and all seems well.

I really don't have any idea, sorry. Try checking all the results for validity. For example, $t_id may not be valid (although it shouldn't be a cause for an error). The only vaguely odd thing in that file is the $query = "select LAST_INSERT_ID()"; line which may only work on mysql (but you have that).

Also, crypt() is running fine on my test setup with win2k, 4.0.3pl1, mysql (same ver), and apache (same).

I have uncommented crypt and it works. This must have been a problem in 4.0.2 (which I was using) and not in later versions.

I am fairly certain the rest of my problems is a problem with php & mantis so I will get php 4.03pl1 and see if it makes a difference.

Well, I'd start commenting chunks of the code and see what is causing the errors. If you can narrow it down some more maybe we can figure out what is wrong and how to fix it.

Ok. In report_bug_page.php3.

If I add a line with <? exit; ?> after line 58 it does not dr watson. If I put this after line 71 it does and I get a 500 error from the server. If I remove the exit and comment out the print_field_option_list(" repro...") I still get a dr watson error but the page is displayed once I have pressed OK on the dr watson screen.

I have played around some more and I am now even more confused than I was before.

I have two systems:-

1. The server (the NT4 WS) running apache etc. with mantis on it.
   
   
2. Another PC running win95/osr2 which I use as a client.
   
   

Using IE4 on the client and I get the results described above but if I use Netscape 4.7 I get different errors. An example with netscape is that I can add a bug and then a bugnote. If I then delete the bugnote I get a dr watson. When I look a the bug the bugnote is deleted.

There seems to be some problem with my setup when php terminates in that it does not terminate cleanly and dr watson's.

Hrm, maybe try closing dr. watson and see if windows gives you a GPF?

Whoop... After checking php.net ( http://bugs.php.net/bugs.php - filter on open bugs that cause reproduceable crashes ) it seems Dr. Watson and php.exe do not get along. Eek, after reading that page I wonder about PHP 4.x stability.

Yes, I had a look around php.net but couldn't find anything exactly the same.

I have downgraded php to 3.0.17 and all seems ok on my system. I will give it a work out and get back to you with my results.

The problem seems to be a known issue with PHP 4 and Dr. Watson. Keep checking php.net for any updates.