topic Re: grep regular expression performance in Operating System - HP-UX

grep regular expression performance

Jim Wagner_2 — Mon, 12 Jul 2004 06:22:51 GMT

Hi,
Running OSF1 5.1 732:

the following egrep is designed to match a quoted string, possibly containing escaped quotes (\") or other escaped characters.
egrep '"([^\"]+|\\.)*"'
This works fine for small matches like
"abc"
but for longer matches like
"abc abc abc abc abc abc abc abc "
performance degrades severely.
On Sun and other platforms this works fine.

Same problem exists using regexec()
Has anyone seen this before, and is it fixed in a patch?
Alternatively, any workaround?

Re: grep regular expression performance

Jov — Tue, 13 Jul 2004 09:53:55 GMT

Hi,
Have you tried "awk", "gawk" or other *awk(s)? awk has better RE then grep|egrep.

Anyway, to answer your question, for strings like "abc abc abc abc abc abc abc abc", I would use egrep '^abc'. This will match the the first "abc" and move on. But I know this will be a very narrow solution.

Also "sed" has a "g" flag or tag which matches globally for the string, thus without this, it will only match the first instance. Give it a try.

Rdgs

Jov

Re: grep regular expression performance

Hein van den Heuvel — Tue, 13 Jul 2004 10:03:24 GMT

Wrong Forum.

OSF1 5.1 732 is a 4 year old Tru64 version nicknamed Zulu.

You probably want to retry after moving to a more recent (V5.1B?) baselevel, or at least with a recent patch kit.

fwiw,
Hein.

Re: grep regular expression performance

Jov — Tue, 13 Jul 2004 10:10:14 GMT

Hi Hein,

Don't be too harsh. HP-UX admins will experience similar problems if HP merge more Tru64 features/fails. ;)

Cheers

Jov

Re: grep regular expression performance

Jim Wagner_2 — Tue, 13 Jul 2004 10:52:49 GMT

[re-posted to Tru64 forum]