19.08.2014, 14:54
The "box" which was mentioned is more specifically the AMX virtual space that the PAWN script operations are limited within. (Please correct me should the info be incorrect)
If by complexity you mean the actions generated in p-code (.amx) versus the instructions generated when compiling with a C++ compiler, see this post by ****** (much other stuff worth checking out there as well
).
I have been writing a huge gamemode in C++ for 2 years now. When I started, I did some timing tests when working with MySQL queries as my own method had eliminated the need for interacting with the AMX layer (calling plugin natives, pipelining the arguments, calling the AMX callback, etc) and provided a much better-looking system for working, the results showed significant improvement. Under strain testing, I'm quite sure everything would be a lot faster, but it most likely doesn't matter, because for me, the main gains from moving to C++ were better code structure, better access to dynamic memory management, better containers and OO-programming. To some extent, this all can be done in PAWN, but that has its own downfalls and C++ is really more convenient.
There are, though, some pitfalls that I took. I had been testing alone or with a few pals for months and then launched the server. It instantly got 30-40 players, who were all eagerly waiting, but since I was much more of a C++ novice back then, I was not aware of some common, some less common mistakes. You have to look out for more issues such as accidental invalid memory access or anything of the similar kind.
If by complexity you mean the actions generated in p-code (.amx) versus the instructions generated when compiling with a C++ compiler, see this post by ****** (much other stuff worth checking out there as well
).I have been writing a huge gamemode in C++ for 2 years now. When I started, I did some timing tests when working with MySQL queries as my own method had eliminated the need for interacting with the AMX layer (calling plugin natives, pipelining the arguments, calling the AMX callback, etc) and provided a much better-looking system for working, the results showed significant improvement. Under strain testing, I'm quite sure everything would be a lot faster, but it most likely doesn't matter, because for me, the main gains from moving to C++ were better code structure, better access to dynamic memory management, better containers and OO-programming. To some extent, this all can be done in PAWN, but that has its own downfalls and C++ is really more convenient.
There are, though, some pitfalls that I took. I had been testing alone or with a few pals for months and then launched the server. It instantly got 30-40 players, who were all eagerly waiting, but since I was much more of a C++ novice back then, I was not aware of some common, some less common mistakes. You have to look out for more issues such as accidental invalid memory access or anything of the similar kind.