Re: [abc] Debug flags

From: Oege de Moor <Oege.de.Moor@comlab.ox.ac.uk>
Date: Sat Mar 04 2006 - 12:27:28 GMT

Eric,

Sorry, we discussed these issues in the local abc meeting, I should
have stated it, obviously.

Re: memory usage in Figure 6 (Fig:safeEnumSpace). This is the same
type of graph we had in our oopsla '05 paper: a measurement
after gc for every 100 iterations, with the dots connected by
a line to show the trend. There should be 10 such lines in this
figure (so it probably needs to become a 2-column one).

Re: Times in Figure 7 (Fig:further).
columns:

  TM default settings (so leak detection and indexing on)
        This column cannot be filled in until indexing is
        ready.

  TMNoLeak: leak detection off, so everything a strong ref
            a flag for this behaviour still has to be added
            indexing on, however, so this cannot be filled in
            until indexing is ready

  TMNoIndex: indexing off, again flag to be added for this
             once indexing is ready (hopefully soon), this
             will coincide with the current default behaviour

Cheers,

-Oege

On Fri, 3 Mar 2006, Eric Bodden wrote:

> Hello.
>
> Regarding table "Fig:further":
>
> Could you please tell (or better even write into the tex file) me what
> exact debug flags ought to be used to fill up the table? For instance I
> found a flag useIndexing, which presumably enabled the new indexing.
> However, I found no such flag for turning off leak detection. Also I am
> not sure whether useNFA should be enabled. I think it is essential to
> have this 100% clear before we start over with the final benchmark runs.
>
> Eric
>
> --
> Eric Bodden
> Sable Research Group, McGill University
> Montreal, Canada
>
>
>
Received on Sat Mar 04 12:27:29 2006

This archive was generated by hypermail 2.1.8 : Tue Mar 06 2007 - 16:13:27 GMT