Re: [Oorexx-devel] Discussion item

[Rick M. <obj...@gm...>]

Parse still is optimized for speed...if you only measure the performance of
a single parse instruction.  However, there are side effects, such as the
impact on memory usage caused the numbers of short term objects that get
created.  The big problem with the "parse var x a x" paradigm is it uses a
very expensive method to keep track of a single piece of information (the
current parsing position) to be reused on the next parse instruction.  I
might have to dust off my parser class I used at the last Rexx Symposium
again.  That class encapsulates the keeping of the markers for you so you
can reuse the information.

Rick

On Dec 7, 2007 8:34 AM, Sahananda (Jon) Wolfers <sah...@wi...>
wrote:

> thanks Rick, Lee,
>
> Yes, it was very easy & I'm very impressed, especially by the various
> match methods - I reckon I'll be making use of them in the future.
> I only heard this year at the symposium that parse was optimized for
> speed, so for the last 8 months I've been trying to get into the mindset of
> using parse wherever I could and now I've found out that that doesn't count
> for ooRexx I'm going to have to retrain myself.
>
> I had always thought that code read better if you referred to the
> attributes by self~attributeName and wondered why anyone ever did anything
> different - now I know.
>
> Lee, thanks for checking it in for me - I'm a duffer with diff and find
> the patches way the most difficult part of doing the development.
>
> thank you both,
>
> Jon
>
>
> On 07/12/2007, Rick McGuire <obj...@gm...> wrote:
> >
> > See, only a few tweaks needed :-)  I took a look at your changes and
> > they looked really good to me.
> >
> > Rick
> >
> > On Dec 7, 2007 7:45 AM, Lee Peedin < le...@sa... > wrote:
> >
> > > Sahananda (Jon) Wolfers wrote:
> > > >     lee wrote --Required 1.578000 Seconds To Read File With 5177
> > > Rows &
> > > >     6 Columns Per Row
> > > >
> > > >
> > > > Ok,
> > > >
> > > > I've made the changes suggested by Rick & some rudimentary testing
> > > > suggests that it still works.
> > > > Lee, If you get a chance, could you run your timings against the
> > > > attached version & see what performance improvement is gained
> > > please.
> > > >
> > > > thanks,
> > > >
> > > > Jon
> > >
> > > Jon,
> > > I re-ran the test this morning 3 times with the old cls and averaged
> > > the time.  I then did the same
> > > thing with the new cls. Your new version results in a 32.51% decrease
> > > in time - well worth the time
> > > you put into it!
> > >
> > > Since I now have commit rights, I am going to check in this new
> > > version.  We can always revert back
> > > if something unforeseen is discovered.
> > >
> > > Thanks for taking the time to develop and enhance this class file.
> > >
> > > Lee
> > >
> >
>
> -------------------------------------------------------------------------
> SF.Net email is sponsored by:
> Check out the new SourceForge.net Marketplace.
> It's the best place to buy or sell services for
> just about anything Open Source.
> http://sourceforge.net/services/buy/index.php
> _______________________________________________
> Oorexx-devel mailing list
> Oor...@li...
> https://lists.sourceforge.net/lists/listinfo/oorexx-devel
>
>