Skip to main content

[Date Prev][Date Next][Thread Prev][Thread Next][Date Index][Thread Index] [List Home]
[cdt-patch] Applied [HEAD] Fw: Single scanner patch


Thanks Dave.  

JohnC
www.eclipse.org/cdt

----- Forwarded by John Camelon/Ottawa/IBM on 06/04/2004 10:32 AM -----

David Daoust/Ottawa/IBM wrote on 06/04/2004 09:56:51 AM:

> Sure,  Here it is.  (I also removed my test harness from the patch)

>
>  - Dave

>
> [attachment "SubScanner.txt" deleted by John Camelon/Ottawa/IBM]

>
> John Camelon/Ottawa/IBM

> 06/04/2004 09:50 AM
>
> To

>
> David Daoust/Ottawa/IBM@IBMCA

>
> cc

>
> Douglas Schaefer/Ottawa/IBM@IBMCA

>
> Subject

>
> Re: Single scanner patch

>
> Dave, can you send the entire patch again, so I can see everything at once?

> Thanks
>
> JohnC
> www.eclipse.org/cdt

>
> David Daoust/Ottawa/IBM wrote on 06/04/2004 06:49:54 AM:
>
> > Here is the updated file

> > [attachment "ScannerContextTopString.java" deleted by John
> > Camelon/Ottawa/IBM]

> >
> > In my tests (of just the scanner) the memory allocated has dropped
> > from 49M to 36M.

> >
> > For the same parse I get:

> >
> > New :  5.2s  143M

> >
> > Old:: 5.6s 156M

> >
> >  - Dave

> >
> > Douglas Schaefer/Ottawa/IBM

> > 06/03/2004 11:11 PM
> >
> > To

> >
> > David Daoust/Ottawa/IBM@IBMCA

> >
> > cc

> >
> > John Camelon/Ottawa/IBM@IBMCA

> >
> > Subject

> >
> > Re: Single scanner patch

> >
> > Well, the time looks great. You took me from 8.9 to 8.1 seconds.
> > But, ouch, you've grown memory by 8 Meg from 156 to 164. It looks to
> > be all in char[]'s so my work might blend well with this.

> >
> > I'd prefer you address the macro issues before you submit, though.
> > This could be explaining the perf improvement ... without iostream
> > you only went from 4.7 to 4.5.

> >
> > Doug Schaefer, IBM's Eclipse CDT Architect
> > Ottawa (Palladium), Ontario, Canada

> >
> > David Daoust/Ottawa/IBM

> > 06/03/2004 05:09 PM
> >
> > To

> >
> > John Camelon/Ottawa/IBM@IBMCA

> >
> > cc

> >
> > Douglas Schaefer/Ottawa/IBM@IBMCA

> >
> > Subject

> >
> > Single scanner patch

> >
> > John,

> >  Here is a patch to reduce the scanner count to 1.
> >
> >  Note that there are three JUnit failures

> >   3-  looks like they were there before.
> >   1.  Looks like a bad test case (off by one in the offset).
> >
> >  There were a bunch of problems masked by the subscanners -- I
> > talked with andrew, but if you include iostream, there are a bunch
> > of bad characters and macro problems with ... arguments.

> >
> >  This patch includes my test class.  There is some extra code in
> > Scanner.evaluateExpression that we can discuss.

> >
> > [attachment "SubScanner.txt" deleted by David Daoust/Ottawa/IBM]

> >  - Dave

Back to the top