Bug in xboard 4.4.4; banned for ICC rated play!

[sje]

Hmm, let's hope not.

First, I mis-reported the game count; it is now 332 games.

The battle is still going with no sign of any bug.

The debug file:

Code: Select all

$ wc xboard.debug
  890154 5910433 32006019 xboard.debug

[bob]

Hmm, let's hope not.

First, I mis-reported the game count; it is now 332 games.

The battle is still going with no sign of any bug.

The debug file:

Code: Select all

$ wc xboard.debug
  890154 5910433 32006019 xboard.debug

Does it hang on FICS without -debug?

It might be ICC related, somehow...

[sje]

Hmm, let's hope not.

First, I mis-reported the game count; it is now 332 games.

The battle is still going with no sign of any bug.

The debug file:

Code: Select all

$ wc xboard.debug
  890154 5910433 32006019 xboard.debug

Does it hang on FICS without -debug?

It might be ICC related, somehow...

Yes sometimes with the unexpected "undo" command same as ICC.

[sje]

Hmm, let's hope not.

After 492 games, it's starting to look like activating the xboard debug output deactivates both the modify-after-free bug and the bad undo bug.

Code: Select all

$ wc xboard.debug
 1306493 8732196 47297515 xboard.debug

[mhull]

Hmm, let's hope not.

After 492 games, it's starting to look like activating the xboard debug output deactivates both the modify-after-free bug and the bad undo bug.

Code: Select all

$ wc xboard.debug
 1306493 8732196 47297515 xboard.debug

Bob (and others) had problems on FICS at CCT time controls. Maybe he could test his setup over there again with debug to see of the problem goes away.

[sje]

Hmm, let's hope not.

After 492 games, it's starting to look like activating the xboard debug output deactivates both the modify-after-free bug and the bad undo bug.

Code: Select all

$ wc xboard.debug
 1306493 8732196 47297515 xboard.debug

Bob (and others) had problems on FICS at CCT time controls. Maybe he could test his setup over there again with debug to see of the problem goes away.

I don't think that the time control has much to do with either bug, but I could be wrong. The bad-undo bug seems to be strongly linked to an out-of-turn opponent resignation; this may have a timing component, but not one related to the chess clock.

570 FICS games so far in a single session and no bugs yet with the debug flag activated. If you want to try your luck against my program at FICS, its handle is Symbolic and it's seeking blitz games.

[mhull]

Hmm, let's hope not.

After 492 games, it's starting to look like activating the xboard debug output deactivates both the modify-after-free bug and the bad undo bug.

Code: Select all

$ wc xboard.debug
 1306493 8732196 47297515 xboard.debug

Bob (and others) had problems on FICS at CCT time controls. Maybe he could test his setup over there again with debug to see of the problem goes away.

I don't think that the time control has much to do with either bug, but I could be wrong. The bad-undo bug seems to be strongly linked to an out-of-turn opponent resignation; this may have a timing component, but not one related to the chess clock.

570 FICS games so far in a single session and no bugs yet with the debug flag activated. If you want to try your luck against my program at FICS, its handle is Symbolic and it's seeking blitz games.

The problem I've had in the past (4.2.7) occurred mid-game where xboard thought the opponent was on move when it was in fact the player's move. This is the same thing that happened to Bob (and others) on FICS.

I lost my computer account on ICC because this would happen once every 600 games or so and my program would lose on time in a 45 5 game. The opponent was very irritated to wait a half hour or more with no moves to win the game on time. On FICS this became really chronic for two or three users. But other users had no problems at all with the same setups.

I can't figure out if your bug is related to this bug.

[sje]

After 740 games in a single FICS session, Symbolic has scored 704 wins, 29 losses, and 7 draws. And as of yet, no xboard bugs have been seen with the -debug option in use.

[sje]

I have restarted Symbolic on FICS to test some changes in Symbolic. The only change related to the xboard bugs is now Symbolic will ignore any undo command when connected to an ICS. Any undo will show up in Symbolic's log, however.

No serious xboard bugs were found with the -debug option in play.

[sje]

Hmm, let's hope not.

"Abandon all hope, ye who enter debug mode."

After ten days of near constant battle on FICS with the -debug flag activated, neither the surprise-undo bug nor the modify-after-free bug have appeared.

I will try some more FICS warfare on a different machine with all settings unchanged.