The next stable version

[Jakson A. Aquino]

Hello!

I wrote a script that compares the statistical results of
the current cvs version of statist with the results of the
version that was in the cvs in the first half of 2005. I
have found two differences. One was a bug in the cvs version
that was fixed yesterday (u-test). Now, the only difference
between the current version of statist at the cvs and the
old stable version is one bug in the old version at "Partial
linear correlation" when we run this function with five
variables.  Thanks to valgrind this bug was fixed in
statist-1.3.1.

The script might not catch all differences in the
statistical results between the two versions. Thus, everyone
is invited to do new tests. Anyway it seems that the current
cvs version is, regarding statistics, at least as accurate
as the old version. In addition, the current version doesn't
consider that user errors are fatal errors: it crashes less,
and quits abruptly less. Thus it might be considered the new
stable version.

The script can be found here:

http://intevation.de/cgi-bin/viewcvs-statist.cgi/statist/tools/run_comparison.sh

Best regards,

Jakson

-------------- next part --------------
A non-text attachment was scrubbed...
Name: not available
Type: application/pgp-signature
Size: 189 bytes
Desc: Digital signature
Url : http://www.intevation.de/pipermail/statist-list/attachments/20060111/dca7a915/attachment.bin