Recording trouble, again

For things that have to do with those crazy test servers... and yeah. By request of z-man, and, of course, you gotta obey...

Moderator: Z-Man

Post Reply
User avatar
Z-Man
God & Project Admin
Posts: 11624
Joined: Sun Jan 23, 2005 6:01 pm
Location: Cologne
Contact:

Recording trouble, again

Post by Z-Man »

Again, most debug recordings done on Bugfarms don't play back long enough to get to the interesting points. I'm investigating, and that means for you that Fortress will see some more restarts than usual. Sorry for the inconvenience.

It also means that most bug cries of last week were useless :(
User avatar
Z-Man
God & Project Admin
Posts: 11624
Joined: Sun Jan 23, 2005 6:01 pm
Location: Cologne
Contact:

Post by Z-Man »

It also means that the server crashes occasionally, apparently. Umm, right when the first person joins, in fact, right now.
Edit: ah, that was wrtl's fault :) I'll forward all dead cats I find in my mail to him :)
User avatar
Z-Man
God & Project Admin
Posts: 11624
Joined: Sun Jan 23, 2005 6:01 pm
Location: Cologne
Contact:

Post by Z-Man »

It may be linked to the spontaneous combustion of player walls. At least, I've tracked it down to a point where during the recording, a player wall is destroyed for an unknown reason, and when playing back, it is still there. Perhaps valgrind can find something.
User avatar
Z-Man
God & Project Admin
Posts: 11624
Joined: Sun Jan 23, 2005 6:01 pm
Location: Cologne
Contact:

Post by Z-Man »

I was comparing pointers to no longer existing objects. While no crash or anything valgrind could have detected was caused by this, the result of the comparison was undefined nevertheless and caused the nonreproducible behavior. It should be fixed now.
Post Reply