Page 85 of 95

Re: NTai XE9.8 experimental for 0.76b1

Posted: 17 Nov 2008, 16:45
by DJ
modinfo.tdf

Re: NTai XE9.8 experimental for 0.76b1

Posted: 17 Nov 2008, 16:53
by hoijui
what about something like the errorhandler?
..the game should go on, not halt, right?

Re: NTai XE9.8 experimental for 0.76b1

Posted: 17 Nov 2008, 17:01
by AF
hoijui wrote:what about something like the errorhandler?
..the game should go on, not halt, right?
If you start a game with 10 NTais and there are no config files, obviously the game will not carry on in a meaningful manner when the enemy commanders sit idle?

Nor would it work well in a comp stomp where one of the AIs has a missing config file and sits idle, wildly tipping the balance in one teams favor

Re: NTai XE9.8 experimental for 0.76b1

Posted: 17 Nov 2008, 17:03
by AF
DJ, I believe that NTai parses that tdf file in order to read a tag, which in BA should have a value such as "BA". This value is then used as an identifier across multiple versions of BA so that a single config can service multiple versions.

This way the configuration files do not break every time a new BA comes out and I don't have to keep re-releasing NTai with no binary changes.

Re: NTai XE9.8 experimental for 0.76b1

Posted: 17 Nov 2008, 17:07
by DJ

Code: Select all

//
// Generated with TAS Modit v2.2 Beta
//

[MOD]
{
	name=Balanced Annihilation V4.7;
	description=Moooooo!;
	modtype=1;
	numdependencies=3;
	depend0=springcontent.sdz;
	depend1=otacontent.sdz;
	depend2=tatextures.sdz;
	[NTAI]
	{
		tdfpath=BA;
	}
}
well i don't know where its getting that config name from then! Just to be clear if i remove it, it creates another one (BA47.sd7.tdf), if i start ntai with the blank one it doesn't give an error. If i change the new file to point at the original config it runs.

Re: NTai XE9.8 experimental for 0.76b1

Posted: 17 Nov 2008, 17:13
by AF
So then it would appear to be an issue with the generation of that value. I can understand that logic since NTai determines the locations and files to use based on that main TDF file, and that files found using the value in question.

Re: NTai XE9.8 experimental for 0.76b1

Posted: 18 Nov 2008, 15:20
by KingRaptor
By any chance, would NTAI be capable of reading a config file set from inside a mod .sdz?

Re: NTai XE9.8 experimental for 0.76b1

Posted: 18 Nov 2008, 15:35
by AF
It reads from the VFS so yes

Re: NTai XE9.8 experimental for 0.76b1

Posted: 18 Nov 2008, 19:42
by AF
I committed a fix and a tiny refactor which should make the configtaskmanager class a little safer and prevent assertions stopping the AI from running, and instead handling the issue there and then.

with regards to the config file issue here is the code in question that controls the value:

Code: Select all

		CLOG("Loading modinfo.tdf");
		TdfParser sf(G);
		sf.LoadFile("modinfo.tdf");

		CLOG("Getting tdfpath value");
		info->tdfpath =  sf.SGetValueDef(string(cb->GetModName()), "MOD\\NTAI\\tdfpath");
As can be seen the call to the interface cb->GetModName() provides the default value to be used should there be no MOD``NTAI``tdfpath value in that tdfparser

I have modified the tdfparser class in a commit while typing this post, to return booleans signifying wether the file loaded or not (though for now i would use these as a guide, I have not added checks for bad parsing of the data such as junk etc). the code that loads the file above has now changed to:

Code: Select all

CLOG("Loading modinfo.tdf");
		TdfParser sf(G);
		if(sf.LoadFile("modinfo.tdf")){
			L.print("modinfo.tdf loaded into parser");
		} else {
			L.eprint("error modinfo.tdf retrieval, it may or may not have been loaded");
		}

		CLOG("Getting tdfpath value");
		info->tdfpath =  sf.SGetValueDef(string(cb->GetModName()), "MOD\\NTAI\\tdfpath");

Re: NTai XE9.8 experimental for 0.76b1

Posted: 06 Jan 2009, 12:40
by 1v0ry_k1ng
do you have a stable version of this or what?
when i configed ntai it was pretty much the best performing AI spring ever had besides the peak of krogothe's KAI- and now whats happened? I can probably convert my configs to work again but is there any AI to put them into?
and if the answer is no, just say no :(

Re: NTai XE9.8 experimental for 0.76b1

Posted: 06 Jan 2009, 12:43
by hoijui
viewtopic.php?f=15&t=16503

Re: NTai XE9.8 experimental for 0.76b1

Posted: 07 Jan 2009, 00:56
by 1v0ry_k1ng
I did not test them, so no guarantees!
well,

Re: NTai XE9.8 experimental for 0.76b1

Posted: 07 Jan 2009, 02:09
by AF
I do not surf these forums anywhere near as meticulously as I used to, and your lucky I saw this on the same day as it was posted. If people need my attention your better emailing me or using msn as those messages get routed all the way to my phone so the turn around for a reply is much faster.

Re: NTai XE9.8 experimental for 0.76b1

Posted: 08 Jan 2009, 21:57
by 1v0ry_k1ng
you didnt answer the question :?

Re: NTai XE9.8 experimental for 0.76b1

Posted: 08 Jan 2009, 22:19
by hoijui
AF usually doesnt have a version of NTai compiled for current spring (he does not use MinGW). besides.. you did not ask anything that i did nto answer, and if i remember right, others were reporting the NTai from my package to work. If that is not the case, you shoudl say that clearer, and possibly give some hints (error message, logfile, ...)

Re: NTai XE9.8 experimental for 0.76b1

Posted: 08 Jan 2009, 22:52
by 1v0ry_k1ng
im asking because I wrote the best profile for NTAI of all time and then it stopped working next version, gg all my work

Re: NTai XE9.8 experimental for 0.76b1

Posted: 08 Jan 2009, 23:14
by hoijui
... you are asking what?
does the NTai from my package work for you or not?
is it the Config Editor youd need?
what is your problem?

Re: NTai XE9.8 experimental for 0.76b1

Posted: 11 Jan 2009, 07:55
by AF
Toolkit should still work although Id rather rebuild it in java and change the config format than bother adding extra features to it. I am still interested in bug reports though, especially stacktraces.

I dont maintain a build of spring and the necessary build environment to test spring on my machines because the overhead of setting it up and keeping it up to date for mingw32 is not worth the effort. As a result I do not release builds of NTai or map exporter AI or any of my other AIs here anymore.

Hopefully this would change with the new C interface, though I am saddened that it has yet to make it into trunk despite the urgency with which it is required.

NTai + XTA + you config should still work save for changes in XTA itself meddling with the config. If things are borked in horrible ways that have no explanation such as the inability of the commander to ever do anything, or crashes, then let me know, but don't use this thread to do so, post it here and make me aware of it by also saying it in an email. Emails are all funneled into my personal mail and I can access and reply via my phone which I always have ( unless Im on the train and the signal cuts out in a tunnel ) so email me if you need my attention and explain everything in the emails.

Re: NTai XE9.8 experimental for 0.76b1

Posted: 11 Jan 2009, 09:15
by Hoi
So you stopped releasing your spring ai's on the spring forum?

Re: NTai XE9.8 experimental for 0.76b1

Posted: 11 Jan 2009, 19:49
by AF
I have source code in the svn although with the switch to git I dont know what Ill be doing since I don't know any clients for windows that just 'work' and if I had the will to install and maintain stuff like msys and cygwin then I would be building mingw32 builds of things but Im not.

The only time I build NTai or any of those binaries is for my portfolio and personal debugging, usually under visual studio 2008. Since I haven't rebuilt my portfolio zip file in a while I haven't released NTai in a while.

Until the C API is out in the official installer build there will be no releases of any of my AIs or groupAIs coming from me

The overhead of releasing and building is simply to great compared to the value obtained from release, the cost gains ratio is appalling.