DaniAngione wrote:Can we PLEASE know what the bug was, when you solve it, just for the sake of curiosity and love for your work?
you could be 99% sure that it was NullPointerException
DaniAngione wrote:Can we PLEASE know what the bug was, when you solve it, just for the sake of curiosity and love for your work?
Granger wrote:Could we get a slightly longer (like 15 seconds) warning before shutting down the server so we have a chance to put down lifted stuff?
Granger wrote:Could we get a slightly longer (like 15 seconds) warning before shutting down the server so we have a chance to put down lifted stuff?
loftar wrote:Sorry, that last restart was apparently unnecessary. I had just failed to entirely recompile my dev server when testing the bugfix.
DaniAngione wrote:Can we PLEASE know what the bug was, when you solve it, just for the sake of curiosity and love for your work? :D
DaniAngione wrote:Can you please go and sleep some after it's fixed?
loftar wrote:DaniAngione wrote:Can we PLEASE know what the bug was, when you solve it, just for the sake of curiosity and love for your work?
Basically, I had named the wrong callback function on wound timers after sleeping, which caused the timer recorded in the wound not to get nulled out when expiring, in turn causing that timer to be doubly freed later on when the wound was freed. In the case of the original crash, this took the shape of freeing a completely unrelated data structure (an indexing structure for claims) that had simply happened to get allocated at the same memory address that the timer had previously resided at, causing its memory to be reused for other things (judging by the patterns of the data having overwritten it, an obstacle cache) while still being referenced.DaniAngione wrote:Can you please go and sleep some after it's fixed?
Will consider. ^^
Users browsing this forum: Claude [Bot] and 58 guests