RevolutionDCM for BTS

@Bob the T

That is an interesting scenario. Whatever is "fixed" in Revolutions is now fixed in RevDCM because they are one and the same since RevDCM 1.00.

Yup, I know that. I don't remember if it was ever fixed in Revolutions though, I just remember one of the mods that used Revolutions fixed it but it was not fixed (art the time) in Revolutions.
 
Ugh.

The assert is not the cause of the fatal hang. It's tied to a frontiersman, a unit I have added to the mod. But here is the thing, there is nothing wrong with the XML or the art. I think there is something going wrong with the AI here. It looks like the AI is confused about what to do with this unit.

Glider is it possible setting up a unit to be able to attack (but not capture cities) and having it's default UnitAI be Explorer could mess things up? That's the only thing I can think of that could be causing this. The unit is a str 6, upgrade for an explorer, which can attack but not capture. There are quite a few of these running around on the map (and must have been for quite a few turns), and have never noticed a problem before (or had this reported in other games or from the RevDCM 1.02 build, and this unit has not changed since then), but this one unit that's in a stack with a bunch of samuri is definatly causeing this fatal hang.
 
More information. Changing it's UnitAI type to Explore instead of Attack stopped the fatal error. However, I want the AI to be able to use these units to attack with. For now, I'll fix this issue by making it so no exporers/scouts/pathfinders can get UNITAI_ATTACK (put that in the noUnitAI field). I will test further and see if I can get this behavior to trip for explorers so you and jdog can fix whatever logic is causing this. Also, if I'm right here, this would also cause a fatal error with default gunships.

Edit: Nope I can't reproduce this, it's something going on in that particular save, with this stack. But as I said, changing the UnitAI to explore from attack saved it. Trying to replicate a similar stack in a fresh game doesn't get this to trip though. In this save the player this trips from is at war with all his neighbors and has one city, I don't know.... I can't figure out how to reproduce this.
 
@Phungus
Yeah definetly these types of peculiarities could upset the AI. Honestly, the 2009 AI technology is actually just a MASSIVE COMPLEX old fashioned analog clock with millions of millions of cogs interconnected. If any cog jams, the clock stops ticking.....

Unless it is reproduceable at least to some degree, it's not worth even attempting a fix. This is simply because if you fix any "problem" in the AI especially non-reproduceable rare errors in custom setups, it can well introduce a problem somewhere else of equal or worse magnitude in the standard setup.

The AI is paranoid-schitzo. It cannot adjust to change except in very narrow parameters and will simply go nuts and flip if you change any of it's world. Normally we like the AI to this when we want to go to war on it, but perhaps in this case it's possibly better to close of BTS, go read a book and let the AI do some self-reflection.

Cheers.
 
glider I've narrowed it down quite a bit. Please check out the last couple posts in the Revolutions Bug report thread, I think you might have enough to go on here to isolate it.

Edit:
And an unrelated request. Any chance you could post links in the OP to the mods that use RevDCM as a base mod? Mainly Legends of Revolution and Rise of Mankind, but there may be others.
 
Hi glider,

the isColor python exception is really bugging me as I try to troubleshoot some python additions.
Can you please help in correcting it?
It's a None type reference, but even if I detect it and don't let the code to call any functions on it,
it still causes problems when it returns in the UnitNaming component...
 
@Avain
Yeah the isColor is a bit of a pain for modders and mergers. There is a comment about it in the RevolutionDCM/docs/revolutiondcm_merge.txt back at version 1.68 of RevolutionDCM. I'll repost it here for you:
1.68 Unit Naming "isColor" error
- Any mod that introduces a new combat type needs to modify BUG at:
- Unit Naming.xml (necessary) for example:
<!--RevolutionDCM - add new combat types here or get python isColor error-->
<option id="Combat_SPY" key="CombatSPY"
type="string" default="DEFAULT"/>
- BugUnitNameOptionsTab.py (not necessary)
- in case modifying this, Bug xml text has to be added as well.
Basically what that is trying to say is that you need to declare any new unit combat types to the unit naming system or else get the non-fatal and completely non-critical but annoying isColor error.

@Phungus
You would like the links put where because I do not know what OP is. I think you mean the download link and documentation which I abbreviate to DLD. :)

Cheers.
 
@Avain
Yeah the isColor is a bit of a pain for modders and mergers. There is a comment about it in the RevolutionDCM/docs/revolutiondcm_merge.txt back at version 1.68 of RevolutionDCM. I'll repost it here for you:
Basically what that is trying to say is that you need to declare any new unit combat types to the unit naming system or else get the non-fatal and completely non-critical but annoying isColor error.

Great, thanks for the info! :goodjob:
 
@Phungus
You would like the links put where because I do not know what OP is. I think you mean the download link and documentation which I abbreviate to DLD. :)

Cheers.
OP = Opening Post, or the first post in a thread ;)
 
Hi!

First of all, thank you for such an incredible mod!

Here is a problem though: when I play "Official RevolutionDCM 2.0 Release" from svn I do not get any random events and I am sure that AI does not get any too. They are not disabled in custom settings screen so that they should occur. Any idea what is wrong (or what I did wrong :) )?

Thanks.
 
@Chem1
There are a good few people involved in this mod. Glad you enjoy it. Did you download it from here:

http://forums.civfanatics.com/downloads.php?do=file&id=8322

because the SVN sourceforge version of this mod is what is undergoing development and could be problematic. Suggest you try again from this link above. I'll be interested to hear how you go.

Cheers.
 
@glider1

I have downloaded version of the mod you suggested. I played some 250 turns and no random events occurred for me as well as for AI. I double-checked options again and they are not disabled. Any ideas what to look for, please?

:-)
 
@Chem
I'll keep an eye out for it. It could just be "randomness" doing it's magician trickery on our perception. A few weeks ago, I finished a game where there were heaps of random events on the AI but none on me. I don't think it is broken as of RevDCM 2.0, but always have to keep an eye out on it. Wonder what other peoples experiences are.
Cheers.
 
I've thought about it more, and I don't want to rip out the RevDCM tab entirely, it has it's uses. But I'd like to disable it unless cheat mode (chipotle) is active. Any ideas on how to do this?

Also any ideas on the critical bug since I've now tracked it to a specific cause (Unit with 2 moves in a stack, the 2 move unit has withdrawl chance and can't capture, stack ends turn next to city and 2 moe unit has UNITAI_ATTACK)? Or is this entirely in jdog's court? Also what about the other core things with RevDCM, mainly exposing RevIdxes (since I'll go ahead and work on getting a scrolling scoreboard in there)? Again is this jdog only?

Oh also it seems there are some text issues with RevDCM 2.00:

http://forums.civfanatics.com/showpost.php?p=8101709&postcount=300
I can confirm the first one shown in pure RevDCM, definatly not caused by any changes in Legends. Not sure about the second, but it's a safe bet the same.

As far as exposing RevIdexes, and all that, I totally agree with this poster. Please read it, it would be really nice if you guys prioritze the issues this poster brings up:
http://forums.civfanatics.com/showpost.php?p=8100787&postcount=297
 
An idea brought up in the BUG forum is adding a button maybe by the log for bringing up the options window. You could have one just for the DCM part of it.

Yes I have already been talking about it. I had thought more buttons on the interface than that even perhaps. But just a lot changes too fast is not good. WoC was a big jump. And I am sure glider and jdog need to get acquainted with it. But like I said to Emperor Fool about the interface will need to be hacked really if wants buttons to be just like standard buttons instead of DDS buttons. Just wanted to let you know it has been discussed.
 
glider, I want to do a little more testing to be sure it is not random, but I don't think I am experiencing any random events either for myself or for the AI. Using RevDCM 2.0 download, not from SVN. I'll test more and let you know what I find.

Cheers,
ripple01
 
Also any ideas on the critical bug since I've now tracked it to a specific cause (Unit with 2 moves in a stack, the 2 move unit has withdrawl chance and can't capture, stack ends turn next to city and 2 moe unit has UNITAI_ATTACK)? Or is this entirely in jdog's court? Also what about the other core things with RevDCM, mainly exposing RevIdxes (since I'll go ahead and work on getting a scrolling scoreboard in there)? Again is this jdog only?
I had quite a lot similar experiences after I had merged RevDCM 1.0 (or was it 1.02) to RoM 2.6. The game hang on AI players turns usually soon after some minor/rebel civ had turned into full civ. In Rise of Mankind I tracked it to be caused by units in stack which had collateral damage but didn't have any bombard or flanking abilities ie. just plain collateral damage (does the SDK handle this kind of situation?). Part of the problem was that Barbarian Civ and Revolution components changed unit AIs after it had spawned them so I ended up making two changes to my mod: removed collateral ability from those units and disabled all UnitAI changes from Barbarian civ / Revolution python modules so that unit AIs are set by SDK the normal way. In regular RevDCM I managed to get this hang only once after many tries while in my own modpack it happened constantly. Now with RevDCM 2.0 I haven't had such problems anymore though I could test it out by adding the collateral damage back for those units...
 
Exciting mod here, very nice. I hope to build up the WoC Project to be compatible with your schema....not the original idea to have unique ports of the WoC if it can be avoided.

Now that the modular loading of the WoC has been out for awhile....and modders like you folks using it....I think it is time to adjust the WoC Project with the feedback of the modders using it. It is really important to me to get your feedback :)
 
I'm just discovering the fact that most of the options of this MOD are configurable; I think I'll reconsider using it :p

That said, would it be possible to indicate in the OP which versions of the mods are used? I'm thinking mostly of BetterAI and BUG; is it BUG 3.6 ?
 
Back
Top Bottom