no more ChessGUI

[hgm]

"deterioration of standards" makes no sense here because Arena was a Chess960-UCI pioneer that set a standard at a time when no other standard existed to deteriorate.

The standards are mutually incompatible, so it is very undesirable that new engines should continue to be produced for both. It means noob users would be forced to manually alter settings that they don't understand and don't know exist, rather than things just working. I am not talking about the past. What matters is the future. Any increase of the minority type reduses the chances that things will work without user intervention. That is deterioration...

Engines from that era still deserve to be supported on the GUI side.

I never implied differently. But production of new ones should be thoroughly discouraged. Supporting them like they are equivalent send the wrong signal, and is thus evil...

[Modern Times]

I. But production of new ones should be thoroughly discouraged. Supporting them like they are equivalent send the wrong signal, and is thus evil...

I disagree - up to the engine author to decide how he or she wants to code their engine. And great that a GUI supports both standards. Both standards are valid.

If I take your logic, then most people would say that UCI has superseded WB, so producing WB engines is evil and GUIs shouldn't support them anymore...

[Alexander Schmidt]

I am sure we talked about that. You don't want to support an ancient standard set by the Arena team and engine authors for the first UCI/Chess960 engines, that's fine.

I think you are mixing up the FEN problem (WB engines) with the castling problem now (UCI engines).

So I consider actions that encourage deteroriation of standards highly immoral.

The problem is that there are several opinions on what the standard is or was. For WB it is the same problem as with UCI. Arena supported the first Chess960/WB engines and later someone set a new standard. But anyway most WB engines work under Arena.

SMK set a new standard for his protocol without cooperating with the Arena team, that's fine as well. The Arena team don't want to support the new standard introduced by SMK that's also fine.

No, that is not fine. It is stubborn and detrimental. They obviously lost that battle, and get stuck in the stone age by not adapting to present-day reality.

Arena is a freeware and sparetime project, so it is up to them what to support and what not. I would never come up with the idea to blame someone not to support something. It is possible to run engines with polyglot under Arena, not so user friendly like in Winboard, but it works the same way.

[Teemu Pudas]

The standards are mutually incompatible, so it is very undesirable that new engines should continue to be produced for both. It means noob users would be forced to manually alter settings that they don't understand and don't know exist, rather than things just working.

Any UCI engine that has the UCI_Chess960 option supports Shredder Chess960. Any other UCI engine either supports Arena Chess960 or doesn't support Chess960 at all.

So if the interface were to automatically pick the appropriate protocol depending on the existence of a UCI_Chess960 option, all any noob would need to know is that an engine that crashes on castling (or the moment it's given a Chess960 FEN) doesn't support Chess960. (This is another thing that's broken about Arena's protocol: the only way to find out if an engine supports it is to crash it if it doesn't).

[hgm]

OK, that is an interesting observation. So to use Polyglot as ACI adapter instead of UCI adapter you would need an option that unconditionally reports it can play Chess960 to the GUI, but lets it depend on the presence of a UCI_Chess960 option how it will encode castling moves and FEN castling rights.

Since this is basically a dialect of UCI for a specific variant, it might be more logical to not run the engine through Polyglot but through UCI2WB. The latter reports to the GUI what it plays, and WinBoard is normally configured to call it with the variant name in the command line. So if it detects Chess or Chess960 there (because WinBoard was already set to one of these variants when the engine was loaded), it could be made to report it plays variants normal and fischerandom to the GUI. And when receiving the comand variant fischerandom it could then switch to a mode where it substitutes any move O-O or O-O-O received from the GUI by the KxR moves as derived from the position FEN.

Users could then simply tick the UCCI/USI checkbox for the engine, rather than the UCI checkbox. (It is, after all, not a UCI engine...)

[hgm]

I disagree - up to the engine author to decide how he or she wants to code their engine. And great that a GUI supports both standards. Both standards are valid.

That would throw us back in the 70s, where each Chess program had its ow unique interface, and engine-engine matches were only possible when operators were typing the moves one did into the other. Do you realize that none of the achievements of the CCRL would have been bossible, if people did what you propose ("decide how he or she wants to code their engine")? I can hardly believe you seriously mean that.

If I take your logic, then most people would say that UCI has superseded WB, so producing WB engines is evil and GUIs shouldn't support them anymore...

Well, most engines are actually WB, and new WB engines are produced monthly. So indeed, UCI can be considered evil.

BTW, before the popularity argument can start to carry any weight one of course has to worry first about quality. If something is intrinsically broken, is is not acceptable as a standard. People would prefer things to be non-compliant and work over them to be compliant and fail any time. So your claim that "both standards are valid" is just wrong. The "Arena standard" is NOT acceptable as a standard, because it is intrinsically broken:

* It is not possible to set up positions with correct castling rights when both Rooks are on the same side of the King.
* It is not possible for the GUI to know whether an engine can play Chess960 at all.

If you want to carry over that argument to WB vs UCI, you would quickly come to the conclusion that UCI is not really acceptable as a communication standard, due to the many deficiencies (e.g. no draw handling).

[hgm]

Arena is a freeware and sparetime project, so it is up to them what to support and what not. I would never come up with the idea to blame someone not to support something. It is possible to run engines with polyglot under Arena, not so user friendly like in Winboard, but it works the same way.

Well, so I am more demanding than you. If someone blows up an airplane, the fact that he did it for free as a sparetime project does not make it OK. We all share the same world, and if someone engages in actions that that make life worse for everyone I would disapprove of those actions even if they came for free.

[Modern Times]

I disagree - up to the engine author to decide how he or she wants to code their engine. And great that a GUI supports both standards. Both standards are valid.

That would throw us back in the 70s, where each Chess program had its ow unique interface, and engine-engine matches were only possible when operators were typing the moves one did into the other. Do you realize that none of the achievements of the CCRL would have been bossible, if people did what you propose ("decide how he or she wants to code their engine")? I can hardly believe you seriously mean that.

Of course I seriously mean that - if an engine author wants to use the Arena chess960 standard, why shouldn't they ? There are two concurrent standards, Arena and Shredder in simplistic terms. The author has a free choice of what to use. Both are valid, both are current. Same as WB and UCI are both current protocols. UCI is the newer of the two, and has been adopted by the vast majority of commercial engines, but authors can still choose WB if they want to, and some do of course.

[JoshPettus]

Why the heck would you want that choice? The whole point of having standards is so we can have engines that run on many GUIs as possible. Why would anyone choose a quirky "standard" that pigeonholes them to one lousy UI?

You are valuing freedom of protocol (which in it self doesn't matter) above freedom to use your engine in many places.


To put this in hardware terms, I thank god that there are standards. I would hate to go back to the day when I could only order hardware from only the manufacturer of my computer. When I had to make a decision to lock myself into a single manufacturer in the hopes that they would continue on that line of parts for as long as possible.

[Alexander Schmidt]

Arena is a freeware and sparetime project, so it is up to them what to support and what not. I would never come up with the idea to blame someone not to support something. It is possible to run engines with polyglot under Arena, not so user friendly like in Winboard, but it works the same way.

Well, so I am more demanding than you. If someone blows up an airplane, the fact that he did it for free as a sparetime project does not make it OK. We all share the same world, and if someone engages in actions that that make life worse for everyone I would disapprove of those actions even if they came for free.

Something different: When will you add Chess960 support to Fairy-Max?