It's here! XE8 is here!
*Gives Alantai a cookie*
...Not much for five months' work...
*Gives Alantai a large pack of cookies*
That's better.
Can't wait to try this out...
NTai XE10.1b
Moderators: hoijui, Moderators
Ok big thign i forgot to mention
type .cheat followed by .aicheat.
That will make NTai start cheating at which point any stall problems dissapear completely, NTai gets full LOS, and 100 handicap, and unlimited resources and disables a few algorithms that are no longer required, making it much tougher...
Also lindir, I added in antistall to every single routine in the unviersal build but as a result NTai's performance took a huge hit, so i removed that. There are tags for the XE8 antistall algorithm though. I've simply removed the antistall bit from FACTORY_CHEAP, B_MEX and another I've forgotten, if I add them again then NTai will stop building mexes under EE and gets horrific stalls.
I'll be uplaoding source code in the next 20 mins along with the source for the groupAI's. I've included the nanoblobz mod and dll's/AI Data inside the archive too for those of you who dont like installers.
type .cheat followed by .aicheat.
That will make NTai start cheating at which point any stall problems dissapear completely, NTai gets full LOS, and 100 handicap, and unlimited resources and disables a few algorithms that are no longer required, making it much tougher...
Also lindir, I added in antistall to every single routine in the unviersal build but as a result NTai's performance took a huge hit, so i removed that. There are tags for the XE8 antistall algorithm though. I've simply removed the antistall bit from FACTORY_CHEAP, B_MEX and another I've forgotten, if I add them again then NTai will stop building mexes under EE and gets horrific stalls.
I'll be uplaoding source code in the next 20 mins along with the source for the groupAI's. I've included the nanoblobz mod and dll's/AI Data inside the archive too for those of you who dont like installers.
-
Lindir The Green
- Posts: 815
- Joined: 04 May 2005, 15:09
http://www.darkstars.co.uk
XE8 Vanilla (no installer) + source code for XE8 and the group AI's + nanoblobz 0.4
XE8 Vanilla (no installer) + source code for XE8 and the group AI's + nanoblobz 0.4
-
Lindir The Green
- Posts: 815
- Joined: 04 May 2005, 15:09
Compiling the vanilla release
As I run Spring on Ubuntu AMD64, I've been trying to get this new release to compile up. I've fixed a few minor issues (like the new SunParse.cpp using _finddata_t *slap*) so far but I've hit a problem compiling MetalMap.cpp - it looks like AIClasses is being used and I can't see a definition anywhere for it.
It's dying (*BOOM*) trying to compile this:
Am I just going blind? 
I've tried my usual find anything anywhere commands:
and no definitions are visible. Any clues? Is there an AIClasses.{h,cpp} missing from the Vanilla source release?
Cheers,
Toby Haynes
It's dying (*BOOM*) trying to compile this:
Code: Select all
class CMetalMap
{
public:
CMetalMap(AIClasses* ai); <-- *BOOM*
virtual ~CMetalMap();
void Init();
int NumSpotsFound;
int AverageMetal;
Code: Select all
-->find . -type f -exec grep -nH AIClasses "{}" \;
./AI/Global/NTAI/Helpers/MetalMap.cpp:3:CMetalMap::CMetalMap(AIClasses* ai)
./AI/Global/NTAI/Helpers/MetalMap.h:10: CMetalMap(AIClasses* ai);
./AI/Global/NTAI/Helpers/MetalMap.h:43: AIClasses *ai;
Cheers,
Toby Haynes
-
unpossible
- Posts: 871
- Joined: 10 May 2005, 19:24
I like how the dominators and merls keep at a standoff distance! impressive to see them actively backing off and still attacking :)
how come NTAI doesn't build mmm's and is a bit frightened of making too many units/attacking?
It seems to like sacrificing it's comm too :)
is this all down to the nanostalling crippling it?
how come NTAI doesn't build mmm's and is a bit frightened of making too many units/attacking?
It seems to like sacrificing it's comm too :)
is this all down to the nanostalling crippling it?
all units do ranged attacks ..
Spring SP 0.5 should work with AA, but I don't make it, ask android_x
Also the metal class you're trying to compile is the wrong file.
That is krogs class v3, but it's currently using v2, which is in one of the subfolders of that, possibly the terrain folder.
Also I'd like to see what changes you made to csunparser
Spring SP 0.5 should work with AA, but I don't make it, ask android_x
Also the metal class you're trying to compile is the wrong file.
That is krogs class v3, but it's currently using v2, which is in one of the subfolders of that, possibly the terrain folder.
Also I'd like to see what changes you made to csunparser
Also, I'd like ti if I had some help, if someone could brign the EE buildtre eupto speed, ti works but it could be a heck of a lot better.
That and soem people to help lindir get the XTA buildtree where he wanted. He said ti wasnt meant for public consumption when compelte and people have noticed a few problems from it as a result.
That and Acidd has bene workign on learnign what eh can fromt eh buildtree definitions to start work on an AA buldtree. I'll release some documentation shortly to help witht hata dn I'd appreciate ti if you helepd him aswell.
I'll eb working on predictive dgunning and a few tweaks such as krogs metal class v4 upgrade from v2
That and soem people to help lindir get the XTA buildtree where he wanted. He said ti wasnt meant for public consumption when compelte and people have noticed a few problems from it as a result.
That and Acidd has bene workign on learnign what eh can fromt eh buildtree definitions to start work on an AA buldtree. I'll release some documentation shortly to help witht hata dn I'd appreciate ti if you helepd him aswell.
I'll eb working on predictive dgunning and a few tweaks such as krogs metal class v4 upgrade from v2
-
Targ Collective
- Posts: 202
- Joined: 12 Nov 2005, 14:16
Some people seem to have a problem with NTAI while some people don't. Maybe it's an issue user-end rather than developer-end. That said, in my testing process NTAI behaved poorly. (Sorry Alantai, I appreciate just how much work you put into this...)
Tested in XTA, although the universal build support showed a performance in AA that was only slightly worse.
On Mars the constant low-level metal content everywhere seems to confuse NTAI, causing it to place mexes in silly spots. (Tested on Cathaldra (base-level zero, specific spots), no such things there, so I presume NTAI has a faulty ability to spot contrast in the metal map, or a lack of prioritisation).
On Cathaldera, where I gave NTAI a chance to go at it, I .gave myself a pair of Kroggies to register as a threat. I then .spectated, leaving NTAI to its own devices. NTAI built far too many resource buildings, didn't tech up anywhere near fast enough, and sent in groups of level one units never greater than seven to ten or so. As for berthas and nuke silos, no sign at all.
On the plus side NTAI was very very good at seizing ground, so your expansion routines were not at fault. Eventually I got bored of waiting for it to assemble an attack force of sufficient power to be worth watching though, and .gave myself Krogoths while spectating until NTAI was wiped out (half in hopes that it might pull itself together, half in frustration).
I don't think NTAI was supposed to behave like that - it didn't take advantage of its massive energy surplus by building enough MMMs - so I'm a little puzzled. Does NTAI scale its attack to the opponent's own skill? That could explain its relative passivity, but it should really attack, attack, attack until the enemy is dead.
Hope this helps!
Oh, and have a cookie for listening.
*Hands Alantai a cookie*
Tested in XTA, although the universal build support showed a performance in AA that was only slightly worse.
On Mars the constant low-level metal content everywhere seems to confuse NTAI, causing it to place mexes in silly spots. (Tested on Cathaldra (base-level zero, specific spots), no such things there, so I presume NTAI has a faulty ability to spot contrast in the metal map, or a lack of prioritisation).
On Cathaldera, where I gave NTAI a chance to go at it, I .gave myself a pair of Kroggies to register as a threat. I then .spectated, leaving NTAI to its own devices. NTAI built far too many resource buildings, didn't tech up anywhere near fast enough, and sent in groups of level one units never greater than seven to ten or so. As for berthas and nuke silos, no sign at all.
On the plus side NTAI was very very good at seizing ground, so your expansion routines were not at fault. Eventually I got bored of waiting for it to assemble an attack force of sufficient power to be worth watching though, and .gave myself Krogoths while spectating until NTAI was wiped out (half in hopes that it might pull itself together, half in frustration).
I don't think NTAI was supposed to behave like that - it didn't take advantage of its massive energy surplus by building enough MMMs - so I'm a little puzzled. Does NTAI scale its attack to the opponent's own skill? That could explain its relative passivity, but it should really attack, attack, attack until the enemy is dead.
Hope this helps!
Oh, and have a cookie for listening.
*Hands Alantai a cookie*
-
Lindir The Green
- Posts: 815
- Joined: 04 May 2005, 15:09
Lindir, my point is that the XTA buildtree is at fault here, and I dont mind you actively working on it but instead other people working with you correctign mistakes and telling them the howto part...
atm XE8 uses v2 of krogothes class which isn't too good for spring style maps though support did improve after my last bought of going on at it. Also it seems I'll have to fiddle around tweaking the resource building rules and make NTai have more B_RULE_EXTREME's in the buildtree. That and v2 of krogs class has an error in calculating ranges that's fixed in v3 which should ahve an impact on spring style maps
Also there are tags binded to the antistall routine..
I'll ahve to get the buildtrees updated, and twiddle the resource rules for an XE8.1 release. That and add altarics lobby client and predictive targetting and v4 of krogs class.
Aside from that I'd appreciate it if the XE8 source was uploaded to the SVN. The svn still has XE4 source and I think an XE4 binary is bundled with spring, which all makes for a very bad image. If you cant be trusted to update a buggy version that's several major versions out of date then I might aswell ask the NTAI source be taken down from the svn in its entirety.
Those poor linux people who think they're getting the stable XE8 when they compile the svn and get instead a weak XE4 with crashbugs.
Also about those who say that XE8 is crashign for them, I'd like to see some mroe data than simpyl sayign it's crashed. You're sayign you're getting messages saying that there's bene an exception in the AI? turn off AI exception handling, and see if it still occurs, and send logs, aswell as details on what mod/map/factions/other AI's.
As far as I am aware XE8 is 100% stable with no crashbugs, and the evidence seems to support that, I've had messages from people who've never been able to run AI's with spring who say that XE8 is the first AI to work for them.
As for crashes on metal maps, I've ran XE8 on metal maps and nothing crashed as have several other people who've been testing XE8 the last month or so, so simply saying it crashes will not make me any the wiser, i need more information..
And a thankyou for the kind words from people =)
atm XE8 uses v2 of krogothes class which isn't too good for spring style maps though support did improve after my last bought of going on at it. Also it seems I'll have to fiddle around tweaking the resource building rules and make NTai have more B_RULE_EXTREME's in the buildtree. That and v2 of krogs class has an error in calculating ranges that's fixed in v3 which should ahve an impact on spring style maps
Also there are tags binded to the antistall routine..
I'll ahve to get the buildtrees updated, and twiddle the resource rules for an XE8.1 release. That and add altarics lobby client and predictive targetting and v4 of krogs class.
Aside from that I'd appreciate it if the XE8 source was uploaded to the SVN. The svn still has XE4 source and I think an XE4 binary is bundled with spring, which all makes for a very bad image. If you cant be trusted to update a buggy version that's several major versions out of date then I might aswell ask the NTAI source be taken down from the svn in its entirety.
Those poor linux people who think they're getting the stable XE8 when they compile the svn and get instead a weak XE4 with crashbugs.
Also about those who say that XE8 is crashign for them, I'd like to see some mroe data than simpyl sayign it's crashed. You're sayign you're getting messages saying that there's bene an exception in the AI? turn off AI exception handling, and see if it still occurs, and send logs, aswell as details on what mod/map/factions/other AI's.
As far as I am aware XE8 is 100% stable with no crashbugs, and the evidence seems to support that, I've had messages from people who've never been able to run AI's with spring who say that XE8 is the first AI to work for them.
As for crashes on metal maps, I've ran XE8 on metal maps and nothing crashed as have several other people who've been testing XE8 the last month or so, so simply saying it crashes will not make me any the wiser, i need more information..
And a thankyou for the kind words from people =)