Did Damian say Christmas of what year?  ;-)

Paul E Wilt 
Principal Software Engineer
_____________________________________________________________________
XanEdu, Inc. (division of Proquest Information and Learning)
http://www.xanedu.com  mailto:[EMAIL PROTECTED] 
300 North Zeeb Rd      Phone: (734) 302-6545  (800) 218-5971 x6545
Ann Arbor, MI 48106    Fax:   (734) 975-6440
_____________________________________________________________________



-----Original Message-----
From: Nadim Khemir [mailto:[EMAIL PROTECTED]] 
Sent: Thursday, December 06, 2001 2:35 AM
To: 'Garrett Goebel'
Cc: [EMAIL PROTECTED]
Subject: RE: Inline not viable with this speed


Thanks for the documentation link, and NO I did not read it.

Also, did you read the section in the documentation titled: "Precompiling
parsers"? Not going to help if you're tweaking the grammar, but otherwise it
cuts down the startup time.
But as damian says:
Note however that ...., so whilst precompilation has an effect on set-up
speed, it has no effect on parsing speed. RecDescent 2.0 will address that
problem.
And I think its more a parsing time problem than a parser build time
problem. Anyhow pre-compiling the parser seems like a good idea specially if
one plans to have many Inline section  in different packages (or does the
same parser get reuse ?? does anyone know?)

I really think that Inline will not get used to its full potential (that is
writing inline code, generating modules that can be installed is fine)
before this problem get fixed. I can think of cases where I want to generate
C code at run time and waiting for compilation that spends 99% in the parser
is just overkill.

One simple solution, rather inelegant compare with using Parse::Recdescent,
would be to write the parser in Yacc. But let wait and see what
Parse::Recdescent 2.0 brings for Christmas.

Nadim

 -----Original Message-----
From:   Garrett Goebel [mailto:[EMAIL PROTECTED]]
Sent:   Wednesday, December 05, 2001 5:32 PM
To:     'Nadim Khemir'; [EMAIL PROTECTED]
Subject:        RE: Inline not viable with this speed

From: Nadim Khemir [mailto:[EMAIL PROTECTED]]
>
> Problems: There is a big problem with inline, SPEED.
> The main problem is the use of Parse::Recdescent.

Damian Conway the author of Parse::Recdescent is currently planning
Parse::FastDescent. Which according to his webpage
(http://www.yetanother.org/damian/b+w/projects.html) will be:

> Successor to Parse::RecDescent: faster, more powerful, easier to use.


>From tail of the tutorial documentation for Parse::RecDescent... Damian
writes:

> version 2.00 (which should be in the CPAN for Christmas)
> will feature a considerable speed boost.


Also, did you read the section in the documentation titled: "Precompiling
parsers"? Not going to help if you're tweaking the grammar, but otherwise it
cuts down the startup time.

Reply via email to