[gnugo-devel] Re: Regressions for 3.2 take 2x as long as for .34?

[Stephan Somogyi]

At 15:42 -0700 26.04.2002, <address@hidden> wrote:

> Are you sure?

The results are quite reproducible.

There are, however, multiple variable at play, which was why I didn't 
want to burden the entire list with the initial query. I am using a 
different toolchain than I was with my .33 and .34 tests, which could 
very well be causing the weirdness I'm seeing.

What I wanted to know is whether trying to isolate the problem was 
inherently futile because, eg, the regression load increased 
dramatically due to the inclusion of the HandTalk games at the last 
minute, or something along those lines.

This apparently not being the case, I have a few more tests to run 
and will hopefully be able to isolate the problem over the weekend.

s.