OT: Why planet gnome doesn't tell all the truth? ... strikes back.

[Brian J. Tarricone]

-----BEGIN PGP SIGNED MESSAGE-----
Hash: SHA1

On 2/21/2006 1:08 PM, Xavier Otazu wrote:

>            XFCE         KDE        GNOME
>           =======     =======     ======
> 
>           0:01.54     0:01.63     0:01.57
>           0:01.53     0:01.65     0:01.58
> rxvt      0:01.53     0:01.67     0:01.56
>           0:01.55     0:01.64     0:01.56
>           0:01.54     0:01.68     0:01.52
>        
>           0:08.01     0:08.01     0:08.08
>           0:08.00     0:07.98     0:08.14
> xterm     0:07.97     0:07.99     0:08.05
>           0:07.90     0:07.96     0:08.04
>           0:07.96     0:07.97     0:08.09
>        
>           0:07.36     0:07.35     0:07.78
>           0:07.23     0:07.36     0:07.85
> gterm     0:07.42     0:07.56     0:07.96
>           0:07.43     0:07.30     0:07.95
>           0:07.48     0:07.26     0:07.75
> 
>           0:06.84     0:06.54     0:06.74
>           0:06.61     0:06.58     0:06.80
> konsole   0:06.85     0:06.60     0:06.75
>           0:06.74     0:06.61     0:06.90
>           0:06.68     0:06.59     0:06.81
> 
>           0:04.28     0:04.22     0:04.38
>           0:04.21     0:04.10     0:04.38
> Terminal  0:04.28     0:04.11     0:04.36
>           0:04.38     0:04.16     0:04.37
>           0:04.27     0:04.18     0:04.40

Not to be mean, but these numbers are basically meaningless without some
kind of error margin.  Can you honestly say that the state of your
system was *exactly* the same for each trial, with the exception of the
desktop environment running?  I'd think not.

And how about methodology?  Did you run the test multiple times on one
terminal, then switch to another?  Or did you run it once on each
terminal, and then repeat several times?  This stuff matters.  I'd argue
that the latter test is more "valid", though still questionably useful.

You can make some correlation between different terminals running on the
same DE, but the cross-DE performance numbers are very suspect and IMHO
useless.  At most, they show that the terminals run about the same on
each DE, as you'd expect.  A fraction of a second isn't enough to be
statistically significant, anyway.

Regardless, as your subject prefix indicates, this is still very
off-topic, and I don't think of general interest to this list.  Maybe
the -dev list, but I think we've beaten this topic to death already.

	-brian

-----BEGIN PGP SIGNATURE-----
Version: GnuPG v1.4.1 (MingW32)

iD8DBQFD+4K76XyW6VEeAnsRAhWlAJ998/f5Nnz10DHxXv8pBXQhYRdZ5wCg5ncv
Ie+BSDyQfRS1t2Dimk3pm0o=
=sFcO
-----END PGP SIGNATURE-----