mimo

Once again a useless report ! That's understandable from a new comer, but you are not anymore. Please attach the commands.txt file.

Strange. Which version of the game do you use ? svn or A18 ? and does the bot build only a few building or none at all ?

should be fixed in r16745. Thanks for the report, but as usual, that would be better with the commands.txt file, if only to check that the fix works.

Concerning your specific question on the preference given to some units compared to other possible ones, there is a function in petra/entity-extend.js which try to compute an "absolute" strength for each units by combining the different attack and armour strengths with appropriate weights, and this is what is used by petra to choose which unit to train. But this is an old function which has not been updated since years while all attack and armour stats have changed a lot. There is no pressing need to update it now as the game is still evolving and balance is far from finished, but feel free to modify this function if you need it.

None of these files have the problem, and they seems to be too short for that. I think you are old enough on this forum and should be able to test yourself your commands.txt before uploading them just go in the directory where you have the commands.txt file and run YourPathToPyrogenesis/pyrogenesis -mod=public -replay=commands.txt (or something analogus if you are on windows)

This can't be the right commands.txt, it has only 400 turns.

As usual, your bugs reports are not helpful. Please provide a way to reproduce it, the commands.txt file for example.

You still have the sandbox level, in which the AI will not attack, so you have plenty of time to practise and attack it. Otherwise, can you be more explicit: when you are attacked by the AI in very easy, how many units does it have and how many do you have ? I've never played at this level, but as the gathering rate of the AI is reduced by a factor 2 compared to yours and in addition it has to keep some defense units at its cc, you should have more defensive units than the AI has attacking ones. In the starting phase, do not build too much structures and do not invest too much resources in technologies, and find the right combination between training units and building structures.

Was it a new game ? and reproducible ? If you played on a saved game from a version before 16548, that warning is expected.

Concerning the AI problems, there were cases (but should be quite rare) where petra was not able to find a proper placement for its dock. Or it may happen that, because of the poor pathFinding in water, its ships are blocked by some obstacle and do not move. The dock placement is fixed in svn, and the pathFinding is being worked on. For the AI levels, are you aware that you already have several available (click on the gears between the player name and the civ name in the game selection window)? In the default (Medium) one, the AI has the same gathering speed as the player, so it is not "Too fast to be realistic" , but you can change the level towards easier ones in which AI have a reduced gathering speed. The gathering speeds are 0.5, 0.66, 1, 1.33 and 1.66 for very-easy, easy, medium, hard and very-hard respectively.

You may also find one map in scenarios: "Sicilia Nomap", in which you start in a ship

I don't mind if it is supposed to be playable or not, and that was not the matter. You wanted feedback, and that were mine. Every new user will first try the specific maps of the mod, and if disappointed, will look elsewhere

Hi, I've just tried this mod. Nicely done here are a few points I noticed: JS error with China-Guilin map on A18 because not finding aegis: you should not have any AI on your maps, so using the default one. China-Guilin map has some strange squares (seems to be water, but weird) Having the possibility to select bridges and seeing them decaying is strange as they are here only for eyecandy purpose. You should either make them non-selectable as the other bridges in the game, or not decaying. But I could not really play it, my FPS was too poor

Looking at your commands.txt, the first line start with start {"settings":{"PlayerData":[{"Colour":{"b":200,"g":46,"r":46},"Name":"Player 1","AI":"","AIDiff":3,"Civ":"maur","Team":0},{"Colour":{"b":20,"g":20,"r":150},"Name":"律西马古","AI":"petra","AIDiff":3,"Civ":"mace","Team":0}],"Seed":56131,"AISeed":10873,"CheatsEnabled" Seems to be a problem with the second player name. Did you use a mod ?

Thanks for the report. Fixed in http://trac.wildfiregames.com/changeset/16159

hi, that's not implemented. The difficulty with walls is in the positioning and checking of obstructions. But if you want to look at it, you can take inspiration from the function findDefensiveLocation (also in headquarter.js) which find the best location for towers and fortresses. It is called from the constructionPlan (in queueplan-building.js) when asked to build a tower.

yep improving the petra doc is in my plans, but I never find the time to do it

niektb: i think the problem is in your sentence "to develop a new AI" which is certainly a several years long program, I've seen on this forum a lot of people wanting to create their own ai from scratch, without being aware of what it means, and which have disappeared. But there are still other ways as to help and improve the existing ai, or even to fork it, which is much more manageable and I think more rewarding as you can see its effects in game quite rapidly. teiresias: I would be happy with a pure documentation tool, but this one has to interpret the code which gives compatibility issues, and this is for me a real concern.

I've not understood what is the compatibility problem that you mention in your original post ? could you give more details, I thought it was just a documentation tool ? If it has JS compatibility issues which prevent us from using some JS features that we would want to use, that's a very strong point against its use.

There are much bigger issues for the ai to build a palisade (placement in particular). But all these points only support my comment #9:to extend mod possibilities in SP, we need more people working on the ai. Some of the needed tasks are simple, they "just" require time,

The ai already uses classes and try to be as dynamic as possible, but the needed changes are not fully completed. And the choice of structures to be built is one of the missing pieces. When the ai wants to build a field, a cart civ will always try to build a cart_field, whatever is in its units templates.

Can you be more specific on what you want the AI to learn ?

Concerning your first problem, Petra does not (yet) use the building list from the template, but assumes thoses from its civ. The solution is very easy: we need a function which take the list of existing entities for a given player, from it get the list of available buildings, and from it check if one will fit with the kind of building we want to build. The only difficulty is that it should be fast enough (must be done dynamically). A perfect task for any modder who want to contribute to the AI

Can you be more explicit of your mod and what you have deleted ? there is no reason for the brits not to behave correctly, so that must come from your changes. Futhermore which version are you using A17 or svn ? and do you have a link to this mod ? edit: from your description, one possibility is that you try to start the game without cc ? that can't work for the time being and need http://trac.wildfiregames.com/ticket/2877

What do you mean by different AIs ? levels ? is yes, you can have a start by looking at http://www.wildfiregames.com/forum/index.php?showtopic=18905