G_Alloc()

[razorace]

As part of the my CoOp work, I've come across a problem with G_Alloc running out of space. Well, I looked at the actual function and it seems a bit inefficent. It only allocates in 32 char chunks and never allows for deallocation.

 

Does anyone know of a more efficent way to do G_Alloc()?

[ensiform]

I would try using trap_TrueMalloc and trap_TrueFree, best methods that I know of. And can be used from all of the modules.

 

trap_TrueMalloc example:

 

g_sess is clientSession * and trueMalloc wants the first arg to be a void **, second arg is the size you wish to alloc.

 

trap_TrueMalloc( (void **)&g_sess, g_maxclients.integer * sizeof( clientSession_t ) );

 

trap_TrueFree example:

 

argument is the pointer you are freeing from truemalloc and again wants **

 

trap_TrueFree( (void **)&g_sess );

 

you really do not need (void **) but gcc likes to be picky about those kinds of things.

[razorace]

I suppose I could do that, but then I'd have to worry about properly deallociating the data. I suspect that Raven did it this way to simplify the process (not requiring dealloc) and to statically allociate the "dynamic" memory.

[ensiform]

And what is wrong with having to dealloc the mem? Btw, that isn't what raven did that is old q3 code (G_Alloc)

[dumbledore]

well if you wanted G_Alloc() to be able to dealloc mem you'd have to do the same thing anyway :S

[razorace]

Well, crap. Poke holes in my logic then.

 

Actually I was interested to see if it was really nessicary to only allociate in the temp buffer in 32 char chunks. It seems silly to me since it's just a big array buffer anyway.