Up Prev Next
From: kimball at stsci dot edu (Timothy Kimball)
Subject: Re: DBlib vs CTlib, please give me a clue!
Date: Oct 10 1997 4:33PM
: I certainly agree that for short, simple queries, your bottleneck
: is likely to be the overhead of fork(), exec(), and loading all
: your perl code.
: But I'm wondering if anyone has done comparisons between mod_perl
: and FASTCGI. So far I haven't tried either one. On this mailing list
: I've heard mod_perl mentioned much more than FastCgi, but when I
: looked into FastCgi, it seemed like a more flexible and general
: technology. If any one has tried both, I'd be interested in hearing
: about their experiences.
Can't answer to that one, not having tried FastCGI (yet). It is more
general, in the sense that it can be done not just with Perl, but with
C or Java (if you're so inclined ;).
Since both FastCGI and mod_perl run pre-compiled Perl, it's not likely
that one is going to have a blinding speed advantage over the other.
: One specific concern I have is handling development. During
: development you _want_ to reload the perl code every time, because
: you're constantly changing it. My impression is that with mod_perl,
: once your code has been read by the server, the only straightforward
: way to load a new version is to restart httpd, and that doesn't
: sound like something I want to do every five minutes. Am I wrong
: about this?...
Yes, you are. :) I'm working on a mod_perl script right now. Changes
to the script are vested at the next invocation, when mod_perl detects
that the script or any modules it loads have changed (modification date
later than what mod_perl remembers). No need to restart the httpd
daemon. I believe the same is true for FastCGI.
Tim Kimball · Data Systems Division ¦ firstname.lastname@example.org · 410-338-4417
Space Telescope Science Institute ¦ http://www.stsci.edu/~kimball/
3700 San Martin Drive ¦ http://archive.stsci.edu/
Baltimore MD 21218 USA ¦ http://faxafloi.stsci.edu:4547/