First: Thanks again for the data, I added it and appended a little .zip-file to the first post containing the updated .ini-file. I won't update the main-upload for small changes like this since uploading things is some work for me.
Well I've had to do some investigating and make a few more filters.
That was intended, I couldn't possibly come up with all possible filters since the system is extensible.
Found out Chevy is not considered a tank but a "StdWheeled" so had to make a filter to limit them.
"category" is "Tracked", "type" is "StdWheeled" - I am just using the terms the game uses internally. Yes, you kind of have to look some things up in the assets.ini - I was thinking about changing that (e.g. just give a dropdown-list) but that might limit some things. Also, I was looking at things from an economic point of view in the end, and with very few users and not even a guarantee that the program would ever be used, I decided against implementing additional comfort-options.
I don't use mass average, dont use at least one assault or light to give it some wiggle room.
"Mass average" should theoretically make it so that a certain mass is preferred and others are chosen less often (using a gaussian distribution) or never (depending on the softness). It's hard to predict if that setting makes the generator too slow, but I added it when Vivicector mentioned that my old generator tended to add too many Solitaires to the drop-lists.
It seems to get angry when there is too many possibilities. I was making 65ton 6 man droplists just fine, but when I bumped it up to 70ton 6 man it was coming up with barely anything. I increased the min/max to 92% and 99% from 94% to free it up a bit more.
The problem about some choices is that other criteria will conflict: Imagine "6 men", "65 tons each", "150k CBills each" - that will look for lists but it'll certainly not find any; other, less insane combinations might have theoretical solutions which are really hard to find (since things are looked for randomly). So if you include BV, CBills and tonnage, all the settings have to make sense and you have to think a step ahead. As I said: The generator lacks sanity-checks or some kind of user-notification if finding a list is really improbable (I guess the time needed to find a list could mathematically be predicted and shown...but that would be a lot of work which I would have done back in 2011 when things looked a bit better for the game ;( ).
Enabled Hovercraft yet see few Harassers, it seems to heavily prefer Solitaires for some reason when it needs at 25 ton asset.
That's a really good point. Might be BV- or CBill-restrictions, most of the criteria first judge all available assets and assign a "badness"-value, so even "Solitaire A" and "B" might have completely different probabilities to show up for certain sets of criteria. I guess a random generator isn't build to be predictable.
Instead of narrowing down limitations and waiting for them to pop up(sometimes they dont at all) I prefer relaxing and letting it pick bunches then manually pairing off the similar lists.
Well, that's a sign of bad design on my part. Balancing out the criteria is some work, but once done in a reasonable way the generation is still pretty fast.
Also manually changed Partisan tonnage because it has never been considered at full 80 tons in any organized event, like ever.
Have it set to 50 tons because apparently that's what Star decided it was going to be but I prefer 60 tons myself.
I heard that before.^^ I don't like tinkering with the values since you can include like five dozen criteria that are supposed to be balanced as a set, no particular need to tweak certain criteria manually. But since the program is open source, I guess I have to accept seeing people "butcher" it.^^
Is there anyway to give certain assets probability? Seems the more players I have the more it loves to add solitaires. :X
Sounds like you should use the "average mass/price/BV"-thingy I mentioned before. Yes, that can be used on any numeric property, not only mass. And during my tests I never saw an abundance of super-light assets.
Same thing with tanks, it loves to have at least one tank most of the time.
Brute forcing enough lists can provide options but sometimes its annoying how it INSISTS on certain things.
The whole "gaussian distribution" makes the generator make "good choices" on a regular basis, even though the user might not consider them "good". Add an "average"-criterion (or several...which might have interesting effects). I recommend something like "65 tons w/ softness of 0.7" (which would completely block assets below 0.3*65=19.5tons).
To sum things up: Sorry for the inconveniences, but many of the sanity-checks to keep problems from cropping up might not be easy to implement. And since I never heard back from the results of the testing done in conjunction with the "Chaos March" I will assume that the time needed to improve the program would be in no reasonable relation to the use it might have.