Thank you John. I'll give the test a try.
-----Original Message-----
From: John P. Rouillard [mailto:rou...@cs.umb.edu]
Sent: Friday, May 31, 2013 5:03 PM
To: Boyles, Gary P
Cc: Simple-evcorr-users@lists.sourceforge.net
Subject: Re: [Simple-evcorr-users] Question on SEC performance -- perlfunc (vs)
regex pattern.
In message
<eddf5556138a864f836192f6d101812a1788e...@fmsmsx108.amr.corp.intel.c
om>, "Boyles, Gary P" writes:
>I was wondering if anyone has done an performance testing on regex (vs)
>perlfunc "patterns"?
>
>I was thinking of replacing this....
>
>ptype=RegExp
>pattern=(\d+)\s+::\s+(\S+)\s+::\s+(\S+)\s+::\s+(\S+)\s+::\s+(\S+)\s+::\
>s+=
>(.*)\s+::\s+(\S+.*)\s+::\s+(.*)\s+::\s+(.*)\s+::\s+(.*)
>
>With this....
>
>ptype=perlfunc
>pattern=sub { my (@parseStr) = split/ :: /, $_[0]; if ($#parseStr > 4)
>= {return (@parseStr); } else {return (0); } }
>
>And was wondering what performance differences there might be.
>
>My events have this format... (input format) of ---
>
>$1 :: $2 :: $3 :: $4 :: $5 :: $6 :: $7 :: $8 :: $9
>:: $10
Which is faster depends on how complex the regexp is. I have had some regexps
that had to backtrack before they realized that they wouldn't apply to an
event. The perlfunc parsing was faster to determine failure. In the case where
the regexp was valid for the event I think it was close enough that I didn't
care, but I chose the perlfunc because it failed more quickly.
In your case I would expect the perlfunc to be faster but you really need to
test it. I think you can test using:
time sec -input test_data --notail -conf regexprule
then a run with:
time sec -input test_data --notail -conf perfuncrule
where each conf has one single rule structured as a perlfunc or regexp.
You need enough data in test_data that the work of compiling the perlfunc
doesn't skew the timing, I would shoot for enough data to make at least one run
take 5 minutes (but you coupld probably use a minute or so if you wanted). Run
each test 5 or so times and see which comes out ahead.
--
-- rouilj
John Rouillard
===========================================================================
My employers don't acknowledge my existence much less my opinions.
------------------------------------------------------------------------------
Get 100% visibility into Java/.NET code with AppDynamics Lite
It's a free troubleshooting tool designed for production
Get down to code-level detail for bottlenecks, with <2% overhead.
Download for free and get started troubleshooting in minutes.
http://p.sf.net/sfu/appdyn_d2d_ap2
_______________________________________________
Simple-evcorr-users mailing list
Simple-evcorr-users@lists.sourceforge.net
https://lists.sourceforge.net/lists/listinfo/simple-evcorr-users
