[ggo-discussion] Re: A few bugs

On Fri, Dec 06, 2002 at 02:28:53AM +0200, Jarkko Lindblad wrote:

> I was scoring a game that just ended - I though I won the game by 0.5 pts
>    as gGo told me so - however, I actually lost the game by 0.5 pts as
>    gGo's internal scoring function didn't count one single-point area of
>    my opponent (it did however mark it as territory on the board, but
>    internally did not count it). After my surprising loss I pressed the
>    'edit game' button and in that board it didn't even mark that one area
>    point as territory (even though it should have as it was a point for
>    black). What I recommend you to do is to remove that silly
>    false-eye-detector as it is to blame I guess. Also I see little use for
>    such feature - rather implement the primiview-style scoring mechanism
>    for the editable boards.

I disabled the false-eye-detector for IGS games again. The problem is, IGS
does score false eyes as points. Now without going into a sophisticated
rules discussion, I don't care if IGS does or not or if that is right or
not. However, your point was valid, gGo was now showing a score that
differes from IGS score, if there were false eyes. So gGo will now skip the
false-eye-detection (so same code as before) and only use it in the SGF editor.

Basically, it does not matter which score gGo calculates. Even if it says
+500, the server will score the game itself. In earlier versions I did not
show the score in IGS games at all, as I thought it might be better to leave
this completely to the server. But as several people requested this feature
to see the score and territory during the scoring process, I enabled it. Of
course, this runs the risk of gGo showing not the real IGS score, but I
think the scorer was doing fine before the false-eye detection, and we will
be back to that again. Nevertheless, keep in mind it is absolutely
irrelevant what score gGo is showing. The fact that it does calculate the
score while removing stones is merely cosmetical. Unlike on KGS, where the
scoring process is synchronized between server and client, this is not
possible on IGS.

However, in the SGF editor it makes some sense in my eyes, as I recall
cursing about some games I reviewed with gGo and CGoban2, while gGo was
showing a wrong score while CGoban2 got the false eyes properly. The game
records ended with false eyes being unfilled. I wanted to fix this, as it is
somewhat ugly in the SGF editor.

Primiview: I'd love to implement that estimator. Maybe around christmas, I
will have some free time then. Currently a too big project to do inbetween
other things.


> The stone-cursor does not appear always when it should: try following:
> move the mouse s that the stone-cursor is over a stone. Then slowly move
> it over a free intersection next to the stone and back a couple times.

Found it, will fix that later. Minor issue, easy fix.


> Oh, yes, the good old bug is back and more severe than ever:
> 
> At the moment it looks like I need to write 'refresh' in the terminal
> window (practically) every time I take white stones, otherwise the black's
> first move won't show. I do have rather slow connection to IGS, btw.

This is still bugging me seriously because I cannot find the reason for
this, and it occurs very randomly for me.
I played a couple of games with gGo 0.3 on Linux, all White (as usual
<toggle irony off>), and the first black move has shown up properly.
I made some testing and asked a friend to use a faster client (so his board
pops up seconds before my Java board appears), and instantly do his black
move. This way gGo always got the move during the process of opening the
board window. In all cases the black stone got displayed once the window was
ready. However, I tested this all on Linux where I hardly got that bug in
the past anyways, I will do the same on Windows later.

I am not sure if the connection speed matters. I doubt there is much speed
difference between an analog modem and DSL/cable for the IGS communication,
except you have some ISP-side lag.

 Peter

Other related posts: