RMG options different from homm3


I realized that the behavior of the “Human or Computer” and “Computer only” options are very different from the original game. In VCMI, “Human or Computer players” must be greater than “Computer only players”. However, as far as I know, “Human or Computer players” is bounded by “players” in RMG templates and “Computer only players” is bounded by “cpu” in RMG templates. In this case, some value in the RMG template make no sense, like “players” : “2”, “cpu” : “6” in 2sm0k.JSON

Is this intentional, or is this a bug?


I think my PR (https://github.com/vcmi/vcmi/pull/645) fix this issue, can anyone check that?


VCMI projects seems to be dead. All vcmi developers are long gone.


Alex merged my PR recently. And he approved this one so probably will merge it soon. But I agree, we lack active contributors and probably some goals. Some project roadmap where it goes and what features we want first.


not to argue but it was not said that vcmi was the future of heroes 3?


Btw it might be worth to replace

can anyone check that


I tested it with various combinations of settings and it works the same as original game

So @dlshcbmuipmam I suggest you to test it yourself to be confident in your changes. Compare VCMI and original game, check UI behaves the same, check boundaries, check generated maps how much players/cpu was generated.


Thanks for the suggestion, actually I mentioned “I can generate 2 human players as a team and 6 cpu players as 3 teams with the “Clash of Dragons” template” in the PR. However, since I am new to VCMI and new to Hero3 (I just want to find an open source game to learn c++ and play it with my friends :grinning:), I don’t really know what else I can check.

By the way, I found a weird pointer-related bug that happens in Windows but not in Linux, I can kind of fix that by preventing destruction of resources, but I believe it can lead to memory leak. Should I discuss such issue here in the forum or in the bug tracker?


Probably you can create a pull request and discuss it with Alex on github. It will be easier to look on code I think.
Regarding to random map generation it will be great to test corner cases like 0 teams and maximum teams in various combinations to ensure it does not crash. I do not know why it is not merged. Maybe Alex is too busy. I just suggest to be more confident in our code because we do not have test team here. Some extra dev testing will not bring any harm.