MikeB wrote: ↑Wed Jun 19, 2019 3:01 pm will match the start position ICCF will display to you on the board
I think the is the confusion. I didn't know even what ICCF is or that there was a standard organization when i asked the question. I have opened both files now and checked the fens and they do match the first 20 or so i looked at. Hopefully this is good not just for me but everybody who may have had some confusion on what the standard is here.
But if you could clarify further on what is ICCF and how this standard is used there.
Edit i just want to add i reread the question i asked and first reply and it was
Does this work for you Mike?
Nothing about that tells me i'm working with any standards beyond what the data source decided to use. And not that it cant be the best attempt at a standard. I sort of assumed you gave it your best. But i was unclear on that.
The confusion maybe comes also from the fact that I gave a link to another file without saying clearly what was the difference or the similitude with the file proposed by MikeB. Sorry for that.
If you are interested, you can see here the function that I used to number positions. See the function PositionNumber. To generate the positions, I used the method invented by Hans L. Bodlaender and explained in this article. After that I sorted my file using the PositionNumber function.
But it is also possible to generate the position directly from the "Scharnagl number". It's explained here.
I like the Bodlaender algorithm because you can use it to generate a position with a dice.
Roland Chastain wrote: ↑Thu Jun 20, 2019 10:57 am
The confusion maybe comes also from the fact that I gave a link to another file without saying clearly what was the difference or the similitude with the file proposed by MikeB. Sorry for that.
I think i always found chess 960 confusing. Did it originally in Pulsar back in maybe 2007 as i thought what the hell it's just chess with a set board which you already support and an added castle rule. Should be a breeze i told myself First version had bugs took time get them out. Then I added it to Lantern Chess to play and observe and i told everyone confidently i had done it. Some guy said something about a problem in observed games. Figured he was nuts. But i did notice our make castle function had some bugs in observed games in time.
I told my main tester on Pulsar who is also a partner on sorts on my apps, has contributed code to many, that I didn't really want to do fischer random in code. Would probably be bugs and the whole 960 programming get's unpleasant But I did it because it was hte next logical step for pulsar and i knew it
Now as to using these two data files, i could use either one, they are about the same minus one has full fen and other first row, not sure on a use right now. The original idea came from the main tester i mentioned. Letting in new game people select the FR position number they want to play. I'm just not sure anymore on the usefulness of that UI over simply offering chess 960 at the moment. So while I asked the question while i was exploring the option, as i do sometimes, exploratory questions, ultimately at least now not using it But as my comments indicated i remained interested in the idea. How standard these are etc and i see the ICCF or something uses them now.
Roland Chastain wrote: ↑Thu Jun 20, 2019 10:57 am
The confusion maybe comes also from the fact that I gave a link to another file without saying clearly what was the difference or the similitude with the file proposed by MikeB. Sorry for that.
I think i always found chess 960 confusing. Did it originally in Pulsar back in maybe 2007 as i thought what the hell it's just chess with a set board which you already support and an added castle rule. Should be a breeze i told myself First version had bugs took time get them out. Then I added it to Lantern Chess to play and observe and i told everyone confidently i had done it. Some guy said something about a problem in observed games. Figured he was nuts. But i did notice our make castle function had some bugs in observed games in time.
I told my main tester on Pulsar who is also a partner on sorts on my apps, has contributed code to many, that I didn't really want to do fischer random in code. Would probably be bugs and the whole 960 programming get's unpleasant But I did it because it was hte next logical step for pulsar and i knew it
Now as to using these two data files, i could use either one, they are about the same minus one has full fen and other first row, not sure on a use right now. The original idea came from the main tester i mentioned. Letting in new game people select the FR position number they want to play. I'm just not sure anymore on the usefulness of that UI over simply offering chess 960 at the moment. So while I asked the question while i was exploring the option, as i do sometimes, exploratory questions, ultimately at least now not using it But as my comments indicated i remained interested in the idea. How standard these are etc and i see the ICCF or something uses them now.
What we posted is very standard and many GUIs let you pick the 960 opening , using the line number as reference in the files submitted. e.g, position 519 will give you the standard chess opening - anybody who plays 960 semi seriously knows this. Also many GUIs will give you an option to let the GUI pick a position for you at random as well. Castling is very easy to figure out and if you don't include FRC with castling , you are probably wasting your time. To castle with the GUI, you just move the K to the R square your are castling with. Stockfish handles it correctly, you can probably use their logic.
MikeB wrote: ↑Thu Jun 20, 2019 7:21 pm
Castling is very easy to figure out and if you don't include FRC with castling , you are probably wasting your time. To castle with the GUI, you just move the K to the R square your are castling with. Stockfish handles it correctly, you can probably use their logic.
Pulsar with FR is in release a week or two now depending if i mean iOS or android. Anyone interested search pulsar chess in app store or play store. And we do use king takes rook as the castle move. In fact the rook the king can castle with is indicated as a legal square to move if possible and the user taps the King. We just as you also mention generate a random FR board for every game.
I saw in another post from 2017 when looking at archived posts on pulsar that you had used pulsar ios, mikeb, so if you still have a copy, try it out if you like.
adams161 wrote: ↑Thu Jun 20, 2019 9:26 pmPulsar with FR is in release a week or two now depending if i mean iOS or android. Anyone interested search pulsar chess in app store or play store.
adams161 wrote: ↑Thu Jun 20, 2019 9:26 pmPulsar with FR is in release a week or two now depending if i mean iOS or android. Anyone interested search pulsar chess in app store or play store.
So no release for Windows?
Well FR was in the existing Pulsar windows binary that is used on mobile as well. It was just never possible to play on mobile as I did not make a UI for it. I can be played on windows with pulsar since 2009. The mobile version does not play every variant even now possible on windows as i've not created UI for them all only six variants so far. For example windows can play shatranj but theres no UI to select that now on mobile. It has been considered since the first mobile pulsar on iOS in 2014 that the UI could be expanded to play more games that already exist in the pulsar engine itself but i had not made UI for.
If i get enough bugs i find in the engine itself to make it worth doing a new windows release i'll consider it but currently the engine code mostly works and most of my bug fixing is in the UI or the adapter to the engine code I'm not inclined to sit down right now and add more chessic search algorithms to pulsar like advance pruning techniques or whatever is currently trendy to make a stronger engine as i'm not interested necessarily in doing pure engine programming right now as much as make apps. Another aspect is the only appeal to adding more technology to search is to get more rating points and i'm not sure making pulsar with a ton of work 200-300 points stronger does much for my average mobile user who is probably under 2000 and almost always under 2200. The engine already feels like a better fit for most everyone who doesn't want to play a super gm strength engine and by being only say a 2600 ICC blitz engine its easier in ways to handicap for people than if i had a 3000 engine.
Edit: Another common reason to make engines stronger is for analysis but i've elected to import other engines for chess analysis. For example on iOS they can analyze with crafty their chess games and on android they can analyze with stockfish their pulsar chess games. So each pulsar app on mobile now comes with two engines. Pulsar more suited for play and a stronger engine for chess analysis.
One of the main reasons i've made the mobile apps for pulsar is because of questions like this, one i found in search going back to 2010 before the mobile pulsar:
As I love to play chess variants, I searched for some variant playing software. I found a few engines like Sjeng and Pulsar (do you know some others, please?). But I'm struggling a week now to get them work, I can't find any software that can play variants and accept those engines, and the Winboard seems to be only GUI that can run those engines and play those variants. I tried following engine installing guide to WB but everything failed. I'm not a programmer or so, so it's too difficult to import the engine to me. GUI's like Arena, ChessMaster, Shredder, ChessBase, ChessAssistant and so on only accepts the regular chess and regular chess playing engines.
It seemed to me the vast majority of my pulsar users were not finding it easy to use in winboard. Now he got a reply finally a month later that pulsar comes with winboard and he didn't need to install it. but even then that version of pulsar with winboard has no concept of levels and you'd need to tinker further with how to run it with my program arguments to get a non full strength version. Frankly let's admit it other than a few engine enthusiasts most were locked out of all pulsar could do when the only UI was winboard. That is the reason for the mobile apps now.
MikeB wrote: ↑Thu Jun 20, 2019 7:21 pm
Castling is very easy to figure out and if you don't include FRC with castling , you are probably wasting your time. To castle with the GUI, you just move the K to the R square your are castling with. Stockfish handles it correctly, you can probably use their logic.
Pulsar with FR is in release a week or two now depending if i mean iOS or android. Anyone interested search pulsar chess in app store or play store. And we do use king takes rook as the castle move. In fact the rook the king can castle with is indicated as a legal square to move if possible and the user taps the King. We just as you also mention generate a random FR board for every game.
I saw in another post from 2017 when looking at archived posts on pulsar that you had used pulsar ios, mikeb, so if you still have a copy, try it out if you like.
oh yea. I have to say its one of my favorite apps - playing crazyhouse is a great way to relax if you're killing sometime waiting for someone - just started playing 960 in the last 6 months.
I know i complained earlier i find FR programming confusing and painsaking at times but it was heavily tested and developed over 4-5 weeks and i'm actually confident that in play and pgn playback i think we got all bugs out. Could not do it myself, my tester was sharp too As I told him particularly in play it had to absolutely work with the castle moves bug free or my users wont find it credible to play if there are any bugs My tester also insisted PGN playback was important to him and that should always work bug free so we fixed bugs there I think i left a bug on castle moves in free play examine mode finally saying lets get this out. But its enourmouse work now to add variants to pulsar We need to support play on both android and ios and pgn logging and play back and then finally free play examine mode when in pgn mode But a ton of work went into making it work in a way that i felt users could find it credible as a playing program as FR I'm a professional i figure. I might find elements of FR programming tedious but work is force over distance. You just keep working and testing and debugging till done. Though i got lucky some of my existing FR code from other apps could be imported in so most FR bugs were in the adapter to the engine i wound up fixing .
But I did it because it was hte next logical step for pulsar and i knew it 
