| Age | Commit message (Collapse) | Author |
|---|
|
|
|
error, due to the stack to throw the error already being full
|
|
data via a looping set of references
|
|
a script
|
|
processing 'while' token of 'do'-'while' statement.
|
|
with negative size
|
|
wasn't caught properly, so they could cause crashes of OpenTTD instead of the AI
|
|
|
|
machines failed
|
|
others were still regular inline), but make sure inline is always a 'forced' inline (I am looking at you MSVC)
|
|
meta-function doesn't return an integer
|
|
it mostly solves the sort() issues
|
|
|
|
included file
|
|
|
|
fix stack overflow
|
|
|
|
|
|
the squirrel compiler to crash
|
|
difference was more than "MAX_UINT"/2
|
|
the AI debug window, use the relative path from /ai/ instead
|
|
|
|
using MSVC update openttd-useful, otherwise make sure you've got liblzo2 and it's development files installed.
|
|
-Revert (r17273): this fix is included in Squirrel 2.2.4
|
|
I've never heard of)
|
|
|
|
|
|
|
|
not be cleaned up properly during crash handling. Occasionally this causes asserts to be triggered or crashes.
|
|
warning about them
|
|
|
|
assert is enabled assert (catch possible overflow bugs in nightlies/RCs), otherwise just increase the stack's size (don't get into invalid reads/writes in releases)
|
|
tail recursion
|
|
|
|
|
|
with not enough parameters can cause a crash
|
|
3rdparty directory.
|
|
|
|
incorrectly but crash the AI instead
|
|
/ Sleep instead of just printing an error message in the AI Debug Window
|
|
|
|
|
|
|
|
belongs there
|
|
|
|
was halted while one or more generators were still in a 'running' state
|
|
the pointer to it is NULL
|
|
those via a call/acall/pcall function
|
|
called (indirectly) via call/acall/pcall OpenTTD crashed. Fix this by disallowing AIs to be suspended while called via call/acall/pcall.
IMPORTANT FOR AI WRITERS: AIs can no longer call any DoCommand functions (change anything, build vehicles, etc.) in a function called (indirectly) via call/acall/pcall. Where possible, please rewrite your code so it doesn't use call/acall/pcall
|
|
and reintroduced in r16241 again, this time without breaking the regression test.
|
