Guild icon
Teeworlds
discord.gg/teeworlds / development
For discussions around the development of the official Teeworlds
Between 2020-08-02 00:00:00Z and 2020-08-03 00:00:00Z
12:25
01b9de8 Fix rand init error message - ChillerDragon 208d8c7 Merge pull request #2675 from ChillerDragon/pr_... - oy
Avatar
101b7b3 fixed line breaks - oy 293209e do a proper bounds check for types when unpacki... - oy
16:31
d03c1f2 mitigate weak hook - nheir 241f62c ignore dragging force when firing ninja - nheir 25e9ce4 move drag velocity to CCharacter::TickDiffered - nheir 6cefadb comment drag velocity - nheir 1894f14 fix: use drag velocity on client side - nheir
Avatar
Is there any guidelines for modding or any overview of internal TW architecture or docs exists?
Avatar
Learath2 had this
19:46
19:46
or someone from ddnet party xd
Avatar
That's hardly useful it's just the relationships between the classes
19:54
There are no guidelines no docs. We all just wing it 😛
19:55
For vanilla there is "Don't use STL"
Avatar
hm, why 🥴
Avatar
Well in the past the reason was the way matricks managed the memory
20:44
But since mem_alloc is now just malloc, it's there for performance concerns I guess
20:44
Though I'm sure a std::vector<T> implementation from any respectable vendor performs better than our array<>
Avatar
Even array<> isn't used that often. There are several places with linked lists, sometimes allocated from CHeap. Dynamic allocations are just avoided in general
Avatar
Linked lists can perform comparably the way we use them (inside preallocated chunks and intrusively) but most of the time array<> and vector<> will handily outperform them because of cache hits and prefetch
Avatar
The hashmap implementation we have for the serverbrowser is not even close to optimal either
Avatar
Most of these list were introduced before array<> I think
Exported 24 message(s)