httpd-dev mailing list archives

Site index · List index
Message view « Date » · « Thread »
Top « Date » · « Thread »
From Rob Hartill <hart...@ooo.lanl.gov>
Subject Re: logging
Date Thu, 25 May 1995 10:34:33 GMT

If Apache is going to dump raw info to a logging process, why are
we even discussing what format the logging process is going to use ?

All we need to write is a logging process which writes the current CLF..
we could easily do that with a spawned Apache process which can be
replaced by a config setting. 

I see no reason why we need to design a new CLF.. let the people 
pick their own. Some people want easy to read logs, others want 
logs which are easy to analyze. Some want Referer and User 
Agent, others want to filter out images, and some are quiet happy
with CLF. We can't write a new CLF which'll make them all happy.

We can dump a pile of example logging processes into the support 
directory. Some in C, some in perl...
I think Randy said "it'll need to be in C", I think perl is far better
suited to transform raw data into formatted data.

As for replacing logging processes at runtime, Apache can easily be
made to reload the logging process after a SIGHUP, and that process
can do whatever it likes, e.g. spawn specialised subprocesses to do other
tricks if that's what people want.

What I'd like to have at the end of this is system which lets me write
a perl script to format a log (or logs) in whatever format *I* choose.
I don't need a CLF or CLF2 because I'm not interested in all the garbage
info that they (might) contain.

So can we forget about how best to escape characters or delimit fields,
and just give people an infinitely configurable logging system with
*no* restrictions built in.


--
Rob Hartill                           
http://nqcd.lanl.gov/~hartill/

Mime
View raw message