[Bug 298810] Performance issues with sed in base

https://bugs.freebsd.org/bugzilla/show_bug.cgi?id=298810

Tim Chase <freebsd@tim.thechases.com> changed:

           What    |Removed                     |Added
----------------------------------------------------------------------------
                 CC|                            |freebsd@tim.thechases.com

--- Comment #1 from Tim Chase <freebsd@tim.thechases.com> ---
For the record, this data appears to be "combined log format"

  https://httpd.apache.org/docs/2.4/logs.html#combined

rather than "common log format"

  https://httpd.apache.org/docs/2.4/logs.html#common

and the sed statement emits all the user-agent strings (the contents of the
last quoted field).

I snagged some publicly-available "combined log format" data from

 
https://github.com/elastic/examples/blob/master/Common%20Data%20Formats/apache_logs/apache_logs

(beware line 8899 has an unterminated quote that needs to be added in) and can
confirm the long runtime (that data took ~27s here).

If I change the sed invocation to

  sed 's/"$//;s/.*"//'

it runs in a tiny fraction of the time (0.825s here).

This has the strong feeling of a backtracking issue, FWIW.

-- 
You are receiving this mail because:
You are the assignee for the bug.