How to make regexes faster? (Python v. OmniMark)

[John Machin]

"Donn Cave" <donn at drizzle.com> wrote in message news:<1019278636.737636 at yasure>...
> Quoth sjmachin at lexicon.net (John Machin):
> ...
> | Another practical perspective:
> |
> | When people mention excessive run-time to me, I ask them to rank the
> | severity of their problem on this scale:
> | 1. Job doesn't finish before they get back to their desk with a fresh
> | mug of coffee.
> | 2. Job has (exclusive access to database or excessive impact on
> | on-line response time) and doesn't finish by sun-up.
> | 3. Job doesn't finish before the weekly cold back-up is due.
> 
> How would you score a CGI program on a busy server in that scale?
> Or for that matter any program that runs on a busy server.  And
> I drink enough coffee, without refilling the cup every time I run
> some software that makes me wait.

The scale is for batch processes, and was tendered in context of the
OP's
comparison of two methods of performing a batch process, the longer of
which took 0.57 SECONDS.

> 
> I'm really delighted to hear about performance improvements in file
> object input in recent Python versions (in a message that would
> have been in this thread if Tim Peters used a newsreader instead
> posting to the list.)  Mostly because it's a good thing in itself,
> assuming the numbers are there, but also because it shows that
> someone must have felt that performance was important enough to
> work pretty hard on it - that's a healthy sign.

You can become delighted *earlier* and more comprehensively if,
instead of waiting for sporadic messages, you read the "What's New in
Python x.y" summary when a new version is released. In this particular
case:

http://www.amk.ca/python/2.1/index.html#SECTION0001500000000000000000