Is there anyone in the QNX communitity who has version of
Apache ported to QNX 6.1 which they know works, and specifically
which supports CGI?
My attempts at downloading official source code has been a
complete disaster. The configure scripts have numerous bugs,
one of which I will document below. Even having hacked through
them, I find that the scripts do not place the code where I’ve
asked, and when I attempt to make what is provided, the compile
fails immediately in an undefinable manner.
The specific bug I have found is that in attempting to find
the largest shared memory segment it can use, the configure
script writes a program that uses open() instead of
shm_open() to create the shared memory segment. The
parameters it provides specifically open a file. That’s
right a file.
Anyone?
\
Mitchell Schoenbrun --------- maschoen@pobox.com
What you’re interested in, 1.3 or 2.0? There was port of 2.0 announced
earlier, check the newsgroups. I have good port of 1.3.12 with
optimisations and I have intention to update it to 1.3.20. Various other
things kept me distracted but I will release it RSN 
I would advice against using 2.0 at this point, unless you have specific
reasons (new functionality, etc). It won’t be any faster on QNX since
mixed ‘fork/threads’ model is not supported. It is not really faster
even on Unix, it only (allegedly) scales better, which is irrelevant
since noone is using QNX6 for such high-volume servers. My understanding
is that Apache team is going to scratch that ‘mixed’ execution model
anyway and replace it with ‘single listener multiply workers’ model for
release of 2.0.
Mitchell Schoenbrun wrote:
Is there anyone in the QNX communitity who has version of
Apache ported to QNX 6.1 which they know works, and specifically
which supports CGI?
My attempts at downloading official source code has been a
complete disaster. The configure scripts have numerous bugs,
one of which I will document below. Even having hacked through
them, I find that the scripts do not place the code where I’ve
asked, and when I attempt to make what is provided, the compile
fails immediately in an undefinable manner.
The specific bug I have found is that in attempting to find
the largest shared memory segment it can use, the configure
script writes a program that uses open() instead of
shm_open() to create the shared memory segment. The
parameters it provides specifically open a file. That’s
right a file.
Anyone?
–
Mitchell Schoenbrun --------- > maschoen@pobox.com
Previously, Igor Kovalenko wrote in comp.os.qnx:
What you’re interested in, 1.3 or 2.0? There was port of 2.0 announced
earlier, check the newsgroups. I have good port of 1.3.12 with
optimisations and I have intention to update it to 1.3.20. Various other
things kept me distracted but I will release it RSN >
I would advice against using 2.0 at this point, unless you have specific
reasons (new functionality, etc). It won’t be any faster on QNX since
mixed ‘fork/threads’ model is not supported. It is not really faster
even on Unix, it only (allegedly) scales better, which is irrelevant
since noone is using QNX6 for such high-volume servers. My understanding
is that Apache team is going to scratch that ‘mixed’ execution model
anyway and replace it with ‘single listener multiply workers’ model for
release of 2.0.
Thanks for replying Igor. I was working with the 2.0 port, but CGI
does not work with it. The port’er did not release source, and did
not return my email to him. I did eventually find your 1.3.12 port,
which works fine. I would like to get the source in case there are
problems. I’ll also be glad to test out 1.3.20 if you want.
Thanks,
Mitchell
Mitchell Schoenbrun --------- maschoen@pobox.com