Print

Please login or register.
1 Hour 1 Day 1 Week 1 Month Forever
Login with username, password and session length

[Upright Path]

Has anyone do any tests to see whether the product chance is actually a random chance on each product, or whether it keeps track of the total 'chance' for each object type, and when 100 is reached one of those objects is spawned?

IE- Will it always take 2 reactions with this product to result in one stone? Or is there a 50% chance each time that it will result in a stone? Meaning you could use the reaction twice and get 0 stone, 1 stone, or 2 stone, depending on how much a RNG likes you?

[PRODUCT:50:1:STONE:NONE:NONE:NONE]

[Warlord255]

I had assumed that it was chance, with the example of Galena's 50% silver chance, but upon further inspection, it turns out that the random chance factor is now kept in the stone file for Galena;

[METAL_ORE:LEAD:100]
[METAL_ORE:SILVER:50]

So I'm not as sure anymore.

[Upright Path]

It's always been there. Or at least, that's where I always remember it being. But this is more of a question on the actual Reactions, on still on my NETBOOK and it'd take me 20 minutes to even test what I suggested before, heh.

If anyone could get me the results of a couple of tests, like...

Make three reactions, each with a different one of the following, and run each reaction X times, then post how many Iron Bars for which reaction and the number of times we should be able to tell whether it's still the <QUANTITY> * <CHANCE> / 100 for creating items thing from the last release.
[PRODUCT:20:5:BAR:NO_SUBTYPE:METAL:IRON]
[PRODUCT:10:10:BAR:NO_SUBTYPE:METAL:IRON]
[PRODUCT:1:100:BAR:NO_SUBTYPE:METAL:IRON]

[NecroRebel]

In the previous version, there was some debate over this, and with testing, we found that the probability model was true in all cases, both in terms of ore reactions and custom smelter reactions. slink also asked Toady in the Future of the Fortress thread, and it was confirmed:

How do the reactions work at the smelter?  Are they percentage probability or percentage completion?  Popular opinion had it that they are percentage completion but testing by a couple of us in the modding forum indicated that it was percentage probability with some weighting from unknown sources.  It was definitely not percentage completion weighted by same operator/same smelter.  We decided that only Tarn Adams could answer the question.  This is relevant in this thread because the new custom shops will undoubtedly be using the same sorts of reactions.  This is a hugely important feature in the new version, for some of us.

The way it works currently, which is open to additions of course, it to go through each product, and if it clears the chance (a single roll), it gives you all of the number.  It doesn't roll the chance for each individual one of a given product, and no invisible partial products are stored.  If you have odd numbers of reagents coming in (like lots of extra), you can get extra product, but the single chance must be cleared first.  So pretty much the clunkiest, most simple interpretation.

The probability model is almost certainly still the true one, rather than the %-completion model. That is, each reaction have a %chance of getting the whole product, and these chances do not add up. As Toady says, no invisible partial products are stored.

[soul4hdwn]

http://www.bay12games.com/forum/index.php?topic=53159.0

Spoiler (click to show/hide)

Products-
[PRODUCT:CHANCE:NUM:ITEM_TYPE:ITEM_SUBTYPE:MATERIAL:MATERIAL_SUBTYPE]
CHANCE - The 'Chance' that an object will appear when the product is completed. Each building keeps an internal count for each product with the matching ITEM_TYPE:ITEM_SUBTYPE:MATERIAL:MATERIAL_SUBTYPE and when the total chance for an object is 100, one is created at that building.
NUM - Number of objects to be created.
ITEM_TYPE - The type of the Item, see the ITEM_TYPE above.
ITEM_SUBTYPE - The raw specific subtype of the item, see above.
MATERIAL - Allows you to specify the Material used for the object. GET_MATERIAL_FROM_REACTION:X, returns the MATERIAL of the item in question. Used so that you don't have to create a reaction for EVERY stone type, for stone reactions, or to get information about a piece of leather so that the armor made by it is <animal type> leather armor.

[Upright Path]

Thank you kindly! Just updated the Wiki with that information.

[sunshaker]

http://www.bay12games.com/forum/index.php?topic=53159.0

Spoiler (click to show/hide)

Products-
[PRODUCT:CHANCE:NUM:ITEM_TYPE:ITEM_SUBTYPE:MATERIAL:MATERIAL_SUBTYPE]
CHANCE - The 'Chance' that an object will appear when the product is completed. Each building keeps an internal count for each product with the matching ITEM_TYPE:ITEM_SUBTYPE:MATERIAL:MATERIAL_SUBTYPE and when the total chance for an object is 100, one is created at that building.
NUM - Number of objects to be created.
ITEM_TYPE - The type of the Item, see the ITEM_TYPE above.
ITEM_SUBTYPE - The raw specific subtype of the item, see above.
MATERIAL - Allows you to specify the Material used for the object. GET_MATERIAL_FROM_REACTION:X, returns the MATERIAL of the item in question. Used so that you don't have to create a reaction for EVERY stone type, for stone reactions, or to get information about a piece of leather so that the armor made by it is <animal type> leather armor.

We had this discussion here about the last version and the Chance thing was discovered to be pure chance, no partial products, see http://www.bay12games.com/forum/index.php?topic=47700.15 about half way down. Adding stuff to the wiki without testing it first (which is really easy to do) leads to arguments later on and the spreading of incorrect information. Please before editing the wiki do some testing on this, its pretty simple to do.

Spoiler (click to show/hide)

By "sunshaker"
maybe a multiple product reaction, no reagent, clean install

1 bar of type a at 75
1 bar of type b at 75
1 bar of type c at 50
1 bar of type d at 34
1 bar of type e at 25

run the reaction 10 times if it is progressive you will have predictable results
a will have 7 bars
b will have 7 bars
c will have 5 bars
d will have 3 bars
e will have 2 bars
run it through the manager so you will get it 10 times without error.
destroy smelter, build new smelter in a new location, repeat test compare results.

By "Slink"
Okay, I tried it.  Not a clean install, though.  I used an established fortress so I could count on everything being in working order.  The reactions made stones that were not present on the map.

[PRODUCT:75:1:STONE:NO_SUBTYPE:STONE:BASALT]
[PRODUCT:75:1:STONE:NO_SUBTYPE:STONE:TALC]
[PRODUCT:50:1:STONE:NO_SUBTYPE:STONE:OBSIDIAN]
[PRODUCT:34:1:STONE:NO_SUBTYPE:STONE:PEAT_CHUNKS]
[PRODUCT:25:1:STONE:NO_SUBTYPE:STONE:MUDSTONE]

I queued 10 of them through the smelter and watched as they were completed all in one session by the Dungeon Master.

basalt = 8
talc = 8
obsidian = 5
peat chunks = 1
mudstone = 4

I had the smelter dismantled, and rebuilt.  Once again, the Dungeon Master completed the set of 10 all in one session.

basalt = 16
talc = 16
obsidian = 12
peat chunks = 3
mudstone = 5

Adjusted from the first set, that becomes as follows.

basalt = 8
talc = 8
obsidian = 7
peat chunks = 2
mudstone = 1

Summary:

75:8,8
75:8,8
50:5,7
34:1,2
25:4,1

Edit: done with 40d

By "NecroRebel"
Huh. Interesting. I guess that means that my test was a fluke. Which is unfortunate, because several times I have told people that the percentage-completion model was the correct one, which means that I've been unwittingly spreading misinformation, which is always bad 

This is also unfortunate because I had wanted to change this mod produce one-and-a-fraction beds for many of the creatures sometime this weekend, but this is impossible to do in the probability model. I don't think making a full bed and having a chance of creating another full bed is quite what I had in mind; I wanted to model using scraps to make a second bed, but having second beds randomly appear doesn't really fit that.

[Upright Path]

Nah, I just said that it was pure chance. Not to test it.

[sunshaker]

Nah, I just said that it was pure chance. Not to test it.

My point was the information quoted by soul4hdwn indicates that it is not pure chance and as you said you added it to the wiki (or at least that is the way the thread reads), I'm saying that soul4hdwn contradicts what we have tested for a previous version and that if information is going to be added to the wiki which is disputed either it needs to be tested to resolve the dispute or it needs to be marked as disputed.

[Upright Path]

Ah.

Hah, hah!

He quoted my own, earlier, post as proof. A lot of that was/is assumptions.