[Retros] Fw: Aw: A new program to solve fairy proof games

[François Labelle]

Hi Bernd,

I'm glad to hear that Jacobi works quickly in some of your tests, and 
I'm sorry to hear that it cooked one of your problems!

> by now, I have noticed that Jacobi can solve proof games in Losing 
> Chess (although it is not mentioned in the list of conditions);
>

For those wondering, LosingChess exists as a condition name in the text 
box only. It is implemented as NonRoyalKings + WhiteMustCapture + 
BlackMustCapture. I designed it this way so that someone could use 
NonRoyalKings on its own, or in combination with other conditions. The 
drawback is that LosingChess is hard to discover this way. Another 
example of condition which might not be obvious is FunctionaryChess (= 
Provocation + Ultra). I will list these hidden conditions in some 
upcoming documentation.

I plan to eventually add SuperCirce and some other fairy conditions. It 
just takes time to add each condition and Popeye has a head start of 
many years. SuperCirce is different from the Circe variants that I have 
so far because it offers a choice of square.

andrew buchanan wrote:
> Once I had discovered I had to unclick the mysterious "Flood 
> Protection" button

Once again, the documentation is missing. Flood protection should be 
clicked except for counting problems. What it does is pause the search 
when it reaches 100 solutions. This is useful because Jacobi could 
search for 1 hour without finding any solution, and then suddenly 
"flood" the output box with 1,000,000 solutions, which could freeze your 
browser.

> My number 1 engine wish currently (and I am not sure how feasible it 
> is for Jacobi), would be to support help retractors for helpmates. 
> Typically one or two moves are undone, and then forward play occurs. 
> It would not be necessary to determine exactly which retractions are 
> retroanalytically viable: that can be done manually. Instead, simply 
> to try all unmoves without regard to global legality. This would be 
> useful even for orthodox problems, but should not be excluded from 
> Jacobi on that account.

I think it should be feasible to add unmoves as a stipulation, that way 
it would be possible to combine it with forward play in the same 
problem. Maybe for orthodox only at first in case it requires separate 
work for each fairy condition.

There are many things that I could add to Jacobi (fairy conditions, 
backward play, neutral pieces...) and maybe I'll add them all eventually 
but it takes time. I'll add them in the order I feel like, but it's 
always good to hear what people are wishing for.

> I think that "Load problem" would be more helpful an instruction than 
> the oracular "Set problem from text". I also enjoyed musing the 
> subtleties of the double when clauses in "When you change the problem, 
> the solver will be updated with the new stipulations/conditions when 
> you click on the /Start/ button." but in the end did not manage to 
> solve that one.

Sorry for the strangeness, it's just that I'm planning ahead for when 
there will be a graphical editor. The text box, the graphical editor, 
and the solver will each be showing the same chess problem, but they 
will end up out-of-sync while someone is editing. That's why I need "Set 
problem from text" to copy the problem from the text box to the 
graphical editor, and "Start" to copy the problem from the graphical 
editor to the solver. So maybe wait to see if it makes sense when my 
vision is implemented. Meanwhile I'll try to come up with a sentence 
that doesn't contain two whens. :)

     François

-------------- next part --------------
An HTML attachment was scrubbed...
URL: <https://pairlist1.pair.net/pipermail/retros/attachments/20171101/89771791/attachment-0001.html>