Not logged inRybka Chess Community Forum
Up Topic Rybka Support & Discussion / Rybka Support / Rybka 4.1 bug thread
- - By Dragon Mist (****) Date 2011-03-07 17:53
Rybka 4.1 in multi PV mode does not display anything on sampled search window (except the move under consideration, and maybe a countermove, but that's it). F11 GUI.
Parent - By Vasik Rajlich (Silver) Date 2011-03-07 18:09
I didn't finish fixing all of the sampled search bugs. This entire feature will be revamped, hopefully with the next Rybka release.

Vas
Parent - - By Maxiator (***) Date 2011-03-08 18:17
Confirmed with Fritz12 GUI and Aquarium 2010. Only one move, no sampled search
Let´s try new visualizer?
regards max
Parent - - By Dragon Mist (****) Date 2011-03-08 21:16
This won't help this issue at all.
Parent - By Maxiator (***) Date 2011-03-10 22:03
you are right, Muilti-PV is not working with Visualizer.
Parent - - By teepee (*) Date 2011-03-09 13:54
I'm confused...sometimes that doesn't take much.
If with Rybka 4.0 in multi PV mode the sampled search window displays an extended line of moves, why does this not do the same in Rybka 4.1?
Did fixing some of the multi PV bugs result in the sampled search display not displaying more than the move under consideration, and maybe a countermove?
Parent - - By Vempele (Silver) Date 2011-03-09 14:07

> Did fixing some of the multi PV bugs result in the sampled search display not displaying more than the move under consideration, and maybe a countermove?


No. This was a new bug in Rybka 4.1 beta 1 (before any bugs were fixed).
Parent - - By teepee (*) Date 2011-03-09 14:44
And we waited 9+ months for new bugs!!

Sloppy, at best.
Parent - - By Dragon Mist (****) Date 2011-03-09 15:34
Good things come to those who wait.
Parent - - By sockmonkey (***) Date 2011-03-09 15:37
To pay for Rybka 5? I'm not holding my breath for a 2nd customer service "miracle" in the Rybka 4 cycle.

Jeremy
Parent - By Dragon Mist (****) Date 2011-03-09 16:20
It's a free world. Mostly.
- - By Captain Jack Sparrow (*) Date 2011-03-09 07:24
Draw by threefold repetition bug.
Rybka shows 0.00 if a position occurs just twice in analysis mode
Example: 1. e4 a6 2. Nf3 Nf6 3. Ng1 Ng8 4. Nf3
Parent - - By Vempele (Silver) Date 2011-03-09 07:29
That's the opposite of a bug.
Parent - - By Captain Jack Sparrow (*) Date 2011-03-09 07:40 Edited 2011-03-09 07:44
i don't understand ! wrong "Draw by threefold repetition" implementation is a feature ?
Parent - - By Vempele (Silver) Date 2011-03-09 08:42
It's not an implementation of the rule (it's up to the GUI to implement that), it's just a feature that considers repeated positions drawn.
Parent - By Captain Jack Sparrow (*) Date 2011-03-09 10:55
ok i gave it a little thought and it turned out to be a smart way of exploiting the threefold repetition rule ...

                 ----------------------------[main node]----------------------------------------------------------------------
                     /                                                             |                                \                                    \
                    /                                                              |                                  \                                 .....
                   /                                                               |                                    \
            [N11: this will cause a repetition]                      [N12]                             [N13]
                                                                                 /         \                           /      \
                                                                                /           \                         /        ....
                                                                               /             \                      ....
                                                                              [N21]     [N22]

Now we don't evaluate N11 because we've done it before and by pruning N11 we gain a great performance boost.
and we give N11 a zero score because :
        1. if the side to move is losing (meaning N12, N13 ... score worse than 0.00) then playing N11 is our only hope for a draw
        2. if the side to move is winning (meaning N12, N13 ... score better than 0.00) then N11 simply is not the best move and won't be played.

this is very intelligent in Playing Mode, however, there can be an engine option to allow the correct behavior in the Analysis Mode, where correct evaluation results are sometimes more important than performance.
Nevertheless, as i already said, this incorrect evaluation almost never causes a misinterpretation (at least by humans) and therefore it's fine to leave it as is.
Parent - - By Captain Jack Sparrow (*) Date 2011-03-09 07:59
However this doesn't have a major effect on analysis of games since in almost all cases it is obvious that Rybka is making a mistake and easy to ignore it ... but sometimes it might be annoying to see the score jump to 0.00 when kibitzing live games ....
Parent - - By Permanent Brain (*****) Date 2011-03-14 03:11 Edited 2011-03-14 03:20

> sometimes it might be annoying


It is! Also, there may be much less obvious situations. It is an old problem, not just related to Rybka but to many engines. There are some which treat the first repetition differently, and/or won't go into a "bad" repetition if it is not forced. Unfortunately, one of the engine I thought to be most reliable here, fails too in the following example. Other engines do not fail here.

It is of course important that you enter the complete PGN, which contains "illogical" repetitions:

[Event "?"]
[Site "?"]
[Date "2011.03.14"]
[Round "?"]
[White "Repetition tricks"]
[Black "?"]
[Result "0-1"]
[PlyCount "10"]

1. g4 e6 2. f4 Nc6 3. Kf2 Nb8 4. Ke1 Nc6 5. Kf2 Nb8 0-1

rnbqkbnr/pppp1ppp/4p3/8/5PP1/8/PPPPPK1P/RNBQ1BNR w kq - 0 6


Engines to be tested in the position after 5...Nb8. Many will happily go for 6.Ke1?? 0.00 followed by the humiliating 6...Qh4#. I've seen one assigning 0.00 even with the pv 6.Ke1 Qh4#.

What does Rybka 4.1 do?

(Note that 4.Ke1 is not a repetition yet, because of the subtleness that White has lost his castling rights compared to the position after 2.f4.)

Added: I think this problem (not that particular example) has been discussed here once or twice, already. I don't remember exactly, but I think it was considered to add the more correct handling as a feature of the analysis mode (only), but keep the "effective" 0.00 assignement as it is for engine playing mode.
Parent - By Captain Jack Sparrow (*) Date 2011-03-15 07:50

> What does Rybka 4.1 do?
>


Same as Rybka 4: plays Ke1 and hopes for a draw . . .
Parent - - By bob (Gold) Date 2011-03-18 20:44
This problem is bogus.  If you score 2-fold repetition as a draw, you would not have fallen for this because the first time the position was reached you would see the mate.  So a contrived position that overlooks a mate the first time around doesn't mean this is a problem that is either fixed nor caused by calling 2 repeats (or 3 repeats) a draw.  You can construct a similar position where the mate was overlooked twice, and either approach will see the repetition the 3rd time around and end the game properly.

Certainly it is not unreasonable for a program to provide analysis based on the idea that the person making the moves up to this point was not a complete idiot???  There is a fix for this particular example, namely only calling the 3rd repetition a draw, close to the root.  But the down-side is worse, in that now you will be unable to see an important 2-fold repetition and take a winning position and let your opponent escape with a draw.  Do you design for idiots for for good chess players???
Parent - - By Permanent Brain (*****) Date 2011-03-30 21:01 Edited 2011-03-30 21:03

> Do you design for idiots for for good chess players???


Well... I also analyze my own games, sometimes. I welcome fool proof engines! :lol:

It is clear that a playing engine would never have such an absurd problem as shown in my example above. That position was just meant to point out what's it about, in a simple and clear way.

It is common that there are (logically) similar situations in real games of very strong players, too. Not about such extreme choices with losing blunders, but e.g. going into an unforced repetition vs. continuing with an advantage, even a small or unclear one. In analysis, it is no help if an engine shows 0.00 just because a player intended to repeat once, but not twice, to gain time on the clock. This can easily be misinterpreted if the realistic eval is close to zero, especially.

I tried to construct a less unrealistic example (but somewhat simple too; some repetitions may not be so "direct"):

5rk1/ppqn3p/2pb2p1/3p4/3P4/2NB4/PPQ2PPP/2K1R3 w - - 0 37


[Event "?"]
[Site "?"]
[Date "2011.03.30"]
[Round "?"]
[White "Repetition to gain time"]
[Black "?"]
[Result "*"]
[Annotator "User"]
[SetUp "1"]
[FEN "5rk1/ppqn3p/2pb2p1/3p4/3P4/2NB4/PPQ2PPP/2K1R3 w - - 0 37"]
[PlyCount "9"]

37. Bxg6 hxg6 38. Qxg6+ Kh8 {Let's imagine White was in time trouble here, and
the time control was after 40 moves. So he thinks, let's delay the decision if
to go for Re6 or not...} 39. Qh6+ Kg8 40. Qg6+ {Now we have enough time.} Kh8
41. Re6 $1 *

Analysing back and forth through this sequence of moves, 0.00 is wrong on every move. I tried some strong engines, and I found two which never evaluate 0.00 here. I think I should not mention their names. :lol:

I understand that "radically" 0.00 for the first repetition is probably useful and effective for the search, when playing. But for analysis mode, I always thought it should be treated more sensible, in the sense of what I tried to show above. Some programmers seem to agree, although sometimes it doesn't seem to work always as intended. Maybe it is difficult to have an engine handle this differently, depending on the mode, IOW if it is playing, or analysing?
Parent - By Akbarfan (***) Date 2011-03-30 21:38

> I understand that "radically" 0.00 for the first repetition is probably useful and effective for the search, when playing. But for analysis mode, I always thought it should be treated more sensible


fully agree with you!
Parent - - By bob (Gold) Date 2011-03-30 22:20
I suppose you could take the 0.0 as a "comment" from the computer...  "Hey, dummy, you repeated a prior position here..  You are either wasting time or you didn't notice."  :)

I understand the problem.  And perhaps in analysis mode one could do something to disable the 2nd rep = draw idea and only count the 3rd.  That would not be very hard to do and would not make a lot of difference in terms of performance...
Parent - - By Uly (Gold) Date 2011-03-31 07:26

> And perhaps in analysis mode one could do something to disable the 2nd rep = draw idea and only count the 3rd.


No, please, that may be best for analysis of postmortem games, but it's horrible and time losing when one uses the engine for analysis of ongoing correspondence games, where one has to find the best move and play it on the board, and the engine runs in circles for no good reason.

I maintain that the best solution is an engine option where the user can decide whether draw is 2fold or comes until the third repetition.
Parent - - By Captain Jack Sparrow (*) Date 2011-03-31 08:43

> and the engine runs in circles for no good reason.


The engine won't run in circles as long as it prunes repetitions within the search tree/line.
Parent - - By Uly (Gold) Date 2011-03-31 12:52
If the main move is scored 0.35 and the repetition is scored 0.35, why would it prune it?

Try self play with engines like Zappa Mexico II, Critter 0.90 or even the engine that starts with H, that is stronger than Rybka and we can't name, and you'll see they'll run in circles (both allowing a repetition from the losing side instead of playing a winning move, and playing a losing move instead of repeating a position). They'll also do it for analysis. All of them don't score 2fold repetition as draw.
Parent - - By Captain Jack Sparrow (*) Date 2011-03-31 15:28
OK let's use some examples.
Example A)
suppose that on the starting position we start the engine in analysis mode and _PLAY_       1.e4 e5 ...  (Let's call it position A)
now the engine starts calculating different lines deeper and deeper, one of them being this:         2. Nf3 Nf6 3.Ng1
at this point if the engine calculates     3. ... Ng8    then it is recalculating the same position it's currently calculating! (position A). so instead the engine should ignore this line 3. ... Ng8 like it doesn't exist in the list of possible moves for black !!!

Example B)
in another scenario suppose that we start the engine in the analysis mode on the starting position and play this:
1. e4 e5 (Position A)
2. Nf3 Nf6
3.Ng1 Ng8 (Position A)

now the engine again calculates different lines like before and one of them is this: 4. Nf3 Nf6 5.Ng1
here engine should consider 5. ... Ng8 for black and give a score of 0.00 to it and prune the rest (of the tree resulting form Ng8), since this move ends the game.

----------------------------------------------------------------------------------------
In the playing mode however in example A, instead of ignoring Ng8 the engine gives it a score of 0.00 and prunes the rest of the tree resulting from Ng8. this will force the engine to play Ng8 if it's losing, in hope of attaining a draw if the opponent is a fool !! (but this deosn't mean that the score of the position is 0.00 or the engine is not losing !!) And if the engine is not losing (meaning that there is a move that scores lower than 0.00) then Ng8 won't be PLAYED by the engine.

So the correct behavior in the analysis mode definitely is to score the second repetition as a draw owing to the fact that when you are analyzing a game you assume both sides to be perfect players. A score of 0.35 means if both players play a perfect game from this position, they would draw. the tricks used in engine play to fool the opponent (or in this case to optimistically hope that the opponent is a fool) shouldn't tamper with correct analysis results ... and since in both cases you are pruning the repetition, scoring the second repetition is as simple as checking a flag and shouldn't incur a noticeable performance cost.

I'm not an engine author, but if these assumptions are right, i don't see any reason the engine should score the first repetition as a draw in analysis mode.
Parent - - By Uly (Gold) Date 2011-04-02 16:54

> Example B)
> in another scenario suppose that we start the engine in the analysis mode on the starting position and play this:
> 1. e4 e5 (Position A)
> 2. Nf3 Nf6
> 3.Ng1 Ng8 (Position A)


No, the point here is, why would you do that? You never do that in analysis, if you notice a repetition, you go back to the position that originated it, in this case, 1...e5, when analyzing, you never have Ng8 in there as you know it leads to a repetition, you just tag Ng8 as "draw" in your analysis file, and prune it.

As for games, reaching that position will never happen, black has no reason to play Ng8 at all because there are plenty of other moves that give black a slight edge, so it doesn't have a reason to go for a repetition. And, if for the sake of discussion, all other moves are worse than draw, then Ng8 should be played, because, why did white play Ng1 in the first place? The reason is probably going to be there again the next time around, and if black has nothing better than a draw, then going for the repetition is the best choice.

In conclusion, scoring two-fold repetition as draw is the best behavior:

For analysis: It saves time since the engine will not suggest repeating moves from the winning side when there are better alternatives. It will also suggest repeating positions from the losing side, which is good because, if the other side played a repetition, it's probable the opponent has nothing better and will go for the draw. This assumes the user is smart and wont analyze looping positions for no good reason.

For gameplay: Same as above (but the engine plays the moves instead of suggesting them).

For postmortem analysis: The game is already done, so this doesn't affect any result, yet, two-fold repetition as draw is useful. Example: Suppose you lost a game as black, in where you reached a closed middle game with lots of piece shuffling, and a couple of move repetitions would have been possible. At some point, you could have made a move that keep on repeating, but you choose a new line instead that led to losing. The two-fold repetition engine will point it out to you, the move that repeated the position would have been better than the one that lost the game. The three-fold repetition engine will point nothing and won't notice the loop, this is known because they themselves will suggest such losing moves or play them on a game because they don't see a difference between going for a loop or not. Information is withhold from the user.
Parent - - By Captain Jack Sparrow (*) Date 2011-04-02 18:59
OK i'm going to make a very clear example why scoring the first repetition as draw in analysis mode can sometimes be WRONG !!

suppose that two unknown players are playing a game and you are analyzing their game. they reach this position:

2k4r/2p3b1/7p/8/7N/1K6/3B2PP/2R5 w - - 0 3


this position is winning for white according to crafty:

3.Bf4 Rf8 4.Rxc7+ Kd8 5.Bg3 Bf6 6.Nf3 Re8 7.Rh7 Re3+ 8.Kc4 Bg5 9.Nxg5 hxg5 10.Rf7
  +-  (5.95)   Depth: 13/23   00:00:00  2267kN

now suppose white plays the correct move Bf4 and black plays Bf6 now white makes a mistake and plays Bd2??

2k4r/2p5/5b1p/8/7N/1K6/3B2PP/2R5 b - - 0 4


this is a mistake because black can now play Bxh4 and draw an otherwise lost game !! but instead of playing Bxh4 and drawing, Crafty suggests that black should play Bg7?? and lose the game ...
this happens because worse than 0.00 doesn't always mean losing !! a score of 0.40 in the endgame is also a draw ... an engine with twofold repetition does not understand this !!

> if you notice a repetition, you go back to the position that originated it


in the above example you can never find Bxh4 unless you restart the analysis after Bd2 which means when kibitzing live games you have to reload the engine !!!

> if for the sake of discussion, all other moves are worse than draw, then Ng8 should be played


i just showed that all other moves are NOT worse than draw but Bg7 was played (incorrectly) ... !

> It saves time since the engine will not suggest repeating moves from the winning side when there are better alternatives.


with correct implementation the engine never suggests repeating moves for the winning side ! can you give an example ?

> At some point, you could have made a move that keep on repeating, but you choose a new line instead that led to losing.


anywhere in a game you can repeat your moves but the problem is that your opponent won't :grin: if there is a situation where a forced repetition can save a losing side, an engine with correct implementation will also detect it ... here is an example:

k4r2/pRP5/p1R5/8/6n1/2Q1b3/6PN/2R4K b - - 0 1


black can force a threefold repetition with Nf2 and draw ... Fritz 12 finds this easily ...

> This assumes the user is smart and wont analyze looping positions for no good reason.


we are examining obvious made up examples here ... there might be situations where the correct evaluation of a position isn't that clear to a normal user ...

Overall, we are talking about the correct behavior here ... the correct behavior in analysis is to always show the best move no matter how strong the opponent is ! assuming one side to be a fool (because they made a foolish mistake) and suggest foolish moves accordingly (as shown in this example) is not the correct bahavior .... assuming the user is smart enought to evaluate positions himself and detect blunders is not the correct behavior ...
Parent - By Uly (Gold) Date 2011-04-10 04:31 Edited 2011-04-10 04:45

> now suppose white plays the correct move Bf4 and black plays Bf6 now white makes a mistake and plays Bd2??


Why did white do that? And what stops it from doing it again? Apparently you are analyzing a game by weak players that will blunder like that, so I see these likely scenarios:

1.- Black plays Bxh4, and goes to lose. After all, he's losing against a player that plays blunders like Bd2??, so he's even weaker.

2.- Black plays Bg7, and White, being the weak player that he is, plays Bd2 over and over, ending the game in a draw (this is better for black than 1.)

3.- The players are so weak that the game result is random. White didn't play Bf4 because it knew it was best and knew the plan to win the pawn, he played it randomly just like he played Bd2??, otherwise he would have played 2.Rxc7 for sure. With this, the players are likely to switch back and forth between lost game to drawn to win, and lost again, which move is played doesn't really matter.

4.- An engine playing Bg7 could beat the white player. After stating the weakness of White, then one finds likely that an engine playing Bg7 would draw by repetition or would turn the tables if White finally played Rxc7.

Now, it's unlikely that you are analyzing this game at all unless you were one of the players, so now the possibilities are, that if you were white, you learned that Rxc7 instead of Bd2 could have won the game. If you were black and won then [the result was random], if you lost, then you learned Bg7 was your best shot at attempting drawing by repetition, and if you drew (with Bxh4, most likely human move) then you learned that Bg7 could have repeated a position (this is extra information given only by the 2-fold repetition engine that you would otherwise have not noticed).

>we are examining obvious made up examples here


I think your example is implausible, you don't really see blunders like Bd2?? in real games, not after the player had made Bf4 before, with the idea of capturing the pawn in mind. Missing a mate in one is more likely to happen.

>with correct implementation the engine never suggests repeating moves for the winning side ! can you give an example ?


Happens all the time, just try telling an engine that scored 3fold-repetition as draw to play against itself and you'll notice looping, there'll be acceptable ones where the winning engine repeats a position and then plays something else and goes to win, but there'll be unacceptable ones where the winning side lets the losing side repeat a position, and then plays something else and goes to win. So, you have a won game, but the losing side could have kept on repeating, so you tag the game as drawn, and don't know if the winning side could have played something else to really win, the game was a waste of time (it's the same for analysis).

But these loops are a correct behavior by the rules of chess (after all,the winning engine does win in the end), so the problem is not the implementation, but the concept of "a position that is a repetition of a previous position is just like any other position" is the problem.
Parent - - By Banned for Life (Gold) Date 2011-03-09 08:05
If I were you, I would start drinking heavily at this point...
Parent - - By Captain Jack Sparrow (*) Date 2011-03-09 08:38
I'm sorry. am i making a mistake somewhere ? could you please explain ?
Parent - - By Banned for Life (Gold) Date 2011-03-09 08:41
There are advantages and disadvantages to an engine scoring the first repetition as a draw. In most cases, the advantages of doing it outweigh the disadvantages, so almost all engines score the draw on the first repetition.
Parent - By Captain Jack Sparrow (*) Date 2011-03-09 10:56
thanks.
Parent - - By Uly (Gold) Date 2011-03-10 02:04 Edited 2011-03-10 02:06
Yes, I actually hate that engines like Critter or Zappa Mexico don't score 2fold repetitions as draws. Now, they have a score of 0.90 for a position and one wants to have a self-play to see their winning plan. The result? A winning line that first loops, the winning side just lets the other repeat a position, the losing side plays a losing move instead of keep on repeating, the winning plan is meaningless, one has to repeat the process without the loop, and then a new loop might appear, ARGH!

I've never faced a disadvantage of "first repetition is draw", it just saves time.
Parent - - By Richard Vida (**) Date 2011-03-10 21:41

> Yes, I actually hate that engines like Critter or Zappa Mexico don't score 2fold repetitions as draws.


I didn't know that this causes such annoyance, I always thought it is more 'correct' this way. I will fix it in Critter 1.0.

Richard
Parent - By Uly (Gold) Date 2011-03-10 21:47
Thank you very much Richard!
Parent - - By vroger007 (**) Date 2011-03-30 18:45
in that case, where can I get critter 0.9.... before you "fix" this? :lol:
Parent - - By Uly (Gold) Date 2011-03-30 18:59

> in that case, where can I get critter 0.9.... before you "fix" this? :lol:


http://www.mediafire.com/?syg0g65cj56d7g3

I also recommend Critter 0.70:

http://www.mediafire.com/?zjmnggirnzw

That despite being weaker has interesting playing style and very different evaluation from Critter 1.0 (the case is similar to Hiarcs Paderborn 2007 and Hiarcs 12 in style, thought the strength difference wasn't that big).
Parent - By vroger007 (**) Date 2011-03-31 20:34
thanks! I'm running a blundercheck already :-) And next thing I will check is whether it has BU or not ;-)
Parent - - By Dragon Mist (****) Date 2011-03-09 10:07
lol
Parent - - By vroger007 (**) Date 2011-03-29 22:06
This is indeed very old and for analysing games, it IS annoying.
I know it is not a bug but intentional behaviour, but a rule is a rule:
a 3-fold repetition is a draw, a 2-fold repetition ISN'T.
This behaviour has caused many discussions, and I did receive feedback
from several engine/GUI developers many years ago, especially after
publishing some results on old engines where some of them didn't
show this behaviour, see
http://users.skynet.be/fa041770/ToChessBase/PruningBug.htm
I personally know dozens of chessplayers who would immediately buy a
new engine if it would not score a 2-fold repetition as a draw!
And that's because they care about correct analysis,
they don't care about an engine gaining or losing 25 elo's or needing
6 minutes per ply for an overnight analysis.
Parent - - By Uly (Gold) Date 2011-03-30 01:57
What? That's backwards, 2-fold repetition is the way to go, otherwise, the engine that is winning delays the game by repeating positions, and the engine that is losing, and has a chance to repeat a position (where it's likely that the opponent would keep on repeating position that would lead to a draw) may not do it, and go to lose instead.

If you are winning, and you get a move that gets you closer to the goal, and another that repeats a position, for which one do you go? An engine that doesn't score 2fold repetition as draw doesn't see the difference.

If you are losing, and you get a move that keeps losing, and one that repeats a previous position, why do you deviate from repetition? An engine that doesn't score 2fold repetition as draw doesn't see the difference.

So, only scoring 3fold repetition as draw makes engines more blind.

We are in disagreement about what behavior is better, but we can agree on that it could be an engine parameter, that was all players can get the behavior they want.
Parent - - By vroger007 (**) Date 2011-03-30 18:40
I would be (very) satisfied with a parameter, yes.
BTW: I was/am talking about engines doing human-human OTB game analysis,
not about engines playing a game.
and BTW2: for the same reason I am also waiting for a next rybka version with BU :-)
Parent - - By keoki010 (Silver) Date 2011-03-30 20:04
Don't hold your breath! :cry: It just won't happen!:smile:
Parent - By vroger007 (**) Date 2011-04-09 11:22
don't worry, I won't ;-)
I took a chance on the engine that we cannot name...
it has BU (so how difficult can it be? :-) )
and since it is told (AND appears!) to be stronger than rybka, I assume its analysis will be at least as good
or better than rybka's, and anyway more "complete":
positions where BU wins and rook/queen only draw DO appear in human-human OTB games,
even if only once every 15 years. So I'm checking out whether the unname-able (how do you spell that?)
engine has maybe also less bugs,
or whether a bug-report gets fixed faster than "hopefully at the end of March" of a not-mentioned year :-)
hmm... sorry but I think I will only come back when Vas again descends from being "in the cloud"...
no offense though!
Up Topic Rybka Support & Discussion / Rybka Support / Rybka 4.1 bug thread

Powered by mwForum 2.27.4 © 1999-2012 Markus Wichitill