Page 2 of 2

Posted: 01 Aug 2006, 00:02
by malric
Still doesn't compile if you use "use_mmgr=True" (gcc (GCC) 3.4.6 (Gentoo 3.4.6-r1, ssp-3.4.5-1.0, pie-8.7.9))

I do not know if this is still used. If so I can make a patch so that it compiles if someone uses use_mmgr=True.

Posted: 01 Aug 2006, 00:21
by hollowsoul
@AF
Had quick test of NTAI

AI builds only lvl1 factory of each type (no water on map so no shipyard of course)
Aircraft lvl1 con built mex extractors, but AI suffer from building multiple in 1 spot. Also was building a vix plant lvl1 ontop of another vix that was helping aswell

AI didnt build any Energy Buildings

AI did build attack kbots though & a scout Vix :-)

edit:- Forgot to say was playing with 2 bots me = spec and running @ 3x speed. Was some lag after first factory build i.e delayed sync.
But assumin u hardcoded building lvl1 factorys = cause of it. Since first time i saw it

Posted: 01 Aug 2006, 00:33
by hollowsoul
@AF
Had quick look @ AI logs
Any chance u can change AI code so it doesnt assume script.txt = spring_datadir/script.txt

see if u can get the script.txt path from spring
ie if u user types
spring /home/deco/script.txt
bot will load up /home/deco/script.txt

edit:- any chance of tryin to loadup script from default position & if its not there. try and get location from spring Just an idea

Posted: 01 Aug 2006, 00:36
by AF
I can get the script from spring, but not in 0.72b1, for which rc22 is coded for, so it wont be in the code just yet

Posted: 01 Aug 2006, 02:11
by Tobi
malric wrote:Still doesn't compile if you use "use_mmgr=True" (gcc (GCC) 3.4.6 (Gentoo 3.4.6-r1, ssp-3.4.5-1.0, pie-8.7.9))

I do not know if this is still used. If so I can make a patch so that it compiles if someone uses use_mmgr=True.
Guess I should add a builder to buildbot to test that. It's supposed to not be broken, but not many devs use it continuously...

a patch would be welcome yes :P

Posted: 01 Aug 2006, 02:45
by AF
logs sent by pm to me by hollowsoul show a heck of a lot of errors pertaining to cb->GetUnitDef() returning an error value.

Has something changed in how cb->GetUnitDef(const char* c) works? Because it doesnt return error values this regularly in 0.72b1, and it would certainly explain the problems with NTai, and why some people can get AAI going, as in the past I've known AAI or OTAI to cache unitdefs....

Is this anything todo with case sensitivity?

I have an idea I'm going to try out.