Re: [gnugo-devel] auto generated regression tests

[Gunnar Farneback]

Dan wrote:
> Then since we have a field called dragon_status that is probably
> not a good name for this gtp function. 

No, we don't. We have fields called owl_status, matcher_status, and
status.

> After my patch the gtp function matcher_status can be used to obtain the
> matcher status and since this is the status used consistently throughout
> the engine that seems to me to be the best way to write these tests.

But the name matcher_status only makes sense with respect to some
internal engine variables, which may change with time. In fact the
original purpose of the matcher_status field was to be a status value
suitable for use when running the matcher. Later it has expanded its
role.

If it is a problem that there's a naming mismatch between the internal
names and the gtp names, I'd rather change the former than replace the
established gtp function dragon_status with something called
matcher_status.

I propose the following change of field names in struct dragon_data:

status -> crude_status
matcher_status -> status

to better reflect their current use. Then the meaning of the different
status field becomes:

crude_status:
Status estimate obtained from static analysis of eye spaces,
surrounding moyo, and escape potential.

owl_status:
Status computed by life and death reading, i.e. the owl code.

status:
Best estimate available. Currently usually owl_status if the owl code
has been run, but could also be modified by the semeai module.

/Gunnar