Re[2]: [cobalt-users] Newer Apache for cobalt Raq2 (was Raq3)

[Pierre Chopot <pierre@xxxxxxxxxxx>]

Hello List,

Wednesday, April 25, 2001, 3:36:46 AM, you wrote:

SW> "Jay Summers" <jay@xxxxxxxxxxxxxxxxxxxx> wrote:
>> > "Geir Halvorsen" <geir@xxxxxxxxxxxxxx> wrote:
>> >> Will there be a newer Apache version for our Raq3?
>> >> Seems that things are going very slow at this point.
>> >> Or can i upgrade it myself without to much problems.
>> >>
>> >> Any sugetions anyone?
>> >> Need at least 1.3.13 or newer!
>> >
>> > There is a newer Apache, you just have to compile it yourself.  Since
SW> the
>> > RaQ3 is no longer being sold I doubt Cobalt is going to release an
SW> upgraded
>> > Apache as a rpm or pkg file.  The latest stable Apache is 1.3.19 and
SW> I've
>> > compiled it on several RaQ2s so if it can compile on a RaQ2 it can
>> > definitely be compiled on a RaQ3.
>>
>> I just recently compiled Apache 1.3.19 for my Raq2 with the latest PHP4.
>> Everything went fine except when I replaced the httpd binary with the
SW> newer
>> one and restarted apache I got connection errors. So I tailed the error
SW> logs
>> and noticed that it seemed to think it was still the old apache binary
>> running? Is there a scoreboard file or something that I have to clear out
>> before the new binary takes over? Or, is there something different to do
SW> at
>> compile time with the new Apache 1.3.19 release?

SW> I've been able to compile PHP4.03pl1 into Apache 1.3.19 on a RaQ2, but not
SW> PHP4.04pl1.  Perhaps it's b/c of the many PHP options I compiled - I'm
SW> waiting until PHP4.05 is released (should be in the next week or two) before
SW> I try upgrading on a RaQ2.  When compiling Apache, stop after 'make'.  Do

I just finished compiling PHP 4.0.4pl1 into Apache 1.3.19 as a static
module on a Qube2. I had the same problem but Oliver Hingst on the Qube list
pointed out a nasty bug in PHP for Mips. See http://bugs.php.net bug ID 9827.
The very latest release of PHP is already a little bit different from
the described bug (you have to change 10000.0 with 1.0 three times
instead of 1 in the source of crypt.c ... better read the problem
description). I did write a small howto for the Qube2 but I was
waiting a few day to check for stability.

HTH

SW> --
SW> Steve Werby

-- 
Regards,
 Pierre                          
 pierre@xxxxxxxxxxx