Print

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

[Sutremaine]

1. Is skill rust displayed by default? I have dwarves that are rusty in various minor skills and one who's definitely lost a level to rust, but I don't see any orange highlights anywhere.

2. Where's the manual? I've forgotten how to do some things.

[splinterz]

1. Is skill rust displayed by default? I have dwarves that are rusty in various minor skills and one who's definitely lost a level to rust, but I don't see any orange highlights anywhere.

2. Where's the manual? I've forgotten how to do some things.

1. unfortunately i've found no reliable way of determining rust vs very rust and as far as i can tell there are bugs with rusting skills that make this even more difficult. the only thing i'm able to detect is if a skill has already begun losing xp. those skills will be orange in the tooltip and in the dwarf detail pane.

2. i know, i know it's sorely needed with the new features, i just haven't had the time. after this next patch goes out i'll try to get a little webpage setup or something, as well as moving these discussions about the branched version to a new thread. i haven't heard anything from DwarfEngineer for over 6 months, so it's looking less and less like a merge will ever happen.

[zwei]

1. Is skill rust displayed by default? I have dwarves that are rusty in various minor skills and one who's definitely lost a level to rust, but I don't see any orange highlights anywhere.

2. Where's the manual? I've forgotten how to do some things.

1. unfortunately i've found no reliable way of determining rust vs very rust and as far as i can tell there are bugs with rusting skills that make this even more difficult. the only thing i'm able to detect is if a skill has already begun losing xp. those skills will be orange in the tooltip and in the dwarf detail pane.

Actually, Quietust just figured it out:

rusty = skill.rating > 0 and skill.rating * 0.5 <= skill.rusty
very_rusty = skill.rating >= 4 and skill.rating * 0.75 <= skill.rusty

[splinterz]

1. Is skill rust displayed by default? I have dwarves that are rusty in various minor skills and one who's definitely lost a level to rust, but I don't see any orange highlights anywhere.

2. Where's the manual? I've forgotten how to do some things.

1. unfortunately i've found no reliable way of determining rust vs very rust and as far as i can tell there are bugs with rusting skills that make this even more difficult. the only thing i'm able to detect is if a skill has already begun losing xp. those skills will be orange in the tooltip and in the dwarf detail pane.

Actually, Quietust just figured it out:

rusty = skill.rating > 0 and skill.rating * 0.5 <= skill.rusty
very_rusty = skill.rating >= 4 and skill.rating * 0.75 <= skill.rusty

well i'll be damned, that's great news, thanks! i'd asked a while ago in the dfhack thread and on irc but nobody responded. ;P

[gn0rt0n]

Are there any versions of the current build available for OSX? All I am finding is old stuff. I thought I had a link for it in the past, but can't find it now. Maybe I'm just remembering incorrectly.

[thistleknot]

request:
option to delete labor optimization plan
derp

I have an idea for a great great great great feature

the thoughts of a dwarf

I.e. the z screen...

normally to find my angry dwarf's, I have to view him in DT, see who's angry, and then open up DF and find him via the Unit screen (after sorting the units)...

imagine... I could do this all from within DT... (I could read the negative thoughts from DT and then focus on the dwarf afterwards, rather than reading in DT, and trying to hunt him within DF)

it would even be nicer if DT could center on the dwarf.  Wow, what a great idea.

[splinterz]

yeah it's been asked a few times before to include the recent thoughts... we'll see..

[Maklak]

An option to cheat and change a Dwarf's skills, attributes and personality, like Runesmith did and JDwarf does would be useful for some kinds of science and testing.

In my opinion, displaying thoughts in the big tooltip over a Dwarf would be useful, but I would rather have a dfhack command to display thoughts and the number of dwarves that have them.

[Sutremaine]

1. unfortunately i've found no reliable way of determining rust vs very rust and as far as i can tell there are bugs with rusting skills that make this even more difficult. the only thing i'm able to detect is if a skill has already begun losing xp. those skills will be orange in the tooltip and in the dwarf detail pane.

I still can't see it anywhere... I don't know if I'm missing something obvious.

Spoiler (click to show/hide)

Skills in yellow are Rusty in-game, and skills in orange are Very Rusty. The skill with the red box around the XP total is the one that's lost a level. All of this is my highlighting.

I'm using branch 19.3 and DF 34.11.

[feelotraveller]

I would like to be able to see my Dwarves starting or maximum attributes in addition to what their current value is.  Wouldn't matter which and I'm guessing that starting values would be easier.

[thistleknot]

maklak:
attributes. i like feelots... idea of starting max tracking.
u think theres a way to combine (maybe some sort of weighted average?) this into attribute % by

1. using a value derived from placement in raw bins with (current method), possible weight .5
2. placement from initial starting value and max value (ie 200% of initial would be 100% for this, initial would be 50%).  What would need to be known here is what is the minimal value a dwarf can fall below initial.  This represents how much a dwarf can actually train up/down, possible weight .25
3. possibly a 0 to 5000 scale (50% would be based on bin median)
3. a value based on max possible for that attribute (i.e. 100% would be 200% of last raw value).  What would need to be known here is what is minimal value a dwarf can fall below lowest bin (i.e. below 1st value defined in raws for that attribute), this represents how much a dwarf is out of possible max of an entire population (i.e. 200% of maximum possible starting value is the max any dwarf can hope to attain, and is relative to all dwarfs). possible weight .25

Update:
I don't think 3 is needed.  It can be accounted for in 2.
Example:
Two dwarf's with the same current value would different in their attribute % by comparing their initial values to current possible max.  Almost like a learning rate.
A dwarf that was at it's current possible max, would have his current value / 1 (or something like that).  These values need to be normalized.
A dwarf that was not at his current possible max would have a lower divisor, and hence a higher end result #.

currently attributes are determined by where they fall within the starting bins. if they exceed or drop below possible starting bin values, the go into the <1% or >99% values.

For the answer to 2 and 3, I see there are tokens that define the "rust" of an attribute
MENT_ATT_RATES
Mental attribute gain/decay rates. Defaults are 500:4:5:4.

and

Phys_att_rates
Physical attribute gain/decay rates. Defaults for STRENGTH, AGILITY, TOUGHNESS, and ENDURANCE are 500:3:4:3, while RECUPERATION and DISEASE_RESISTANCE default to 500:NONE:NONE:NONE.

looks like default rust is 4.

Whatever that means, it's possible the decay formula is the same as Q's as pointed out earlier.

If it is (testing I guess?), maybe we can come up with a more comprehensive formula for identifying attribute %s.

[splinterz]

i think in 19.3 the only place the 'lost xp' is indicated is on the dwarf panel, not in the tooltips. since the conditions of rust were figured out recently, i've already incorporated them into the next build, and they'll also highlight the tooltip too.

all the decay variables do is set limits on a set of counters. each counter increments to it's max and then the next counter is incremented, etc.. all this is for the purpose of adding layers of rust. the number of rust layers is what's used to determine if a skill is rusty or very rusty. now that this can be determined, it should probably be considered when finding ratings for skills (specific to role calculations)

the maximum attribute values for a caste shouldn't be too difficult to display, but i'm not sure of the best way to roll them into role calculations, or if it's necessary.

[thistleknot]

on all that stuff about attributes.

splinterz and i after a lot of discussion, decided to extend the last frequency bin to double the last value. (a dwarf can attain twice his initial starting value in attributes, so it goes to show the max a dwarf can achieve is 200% max embark possible value).

see
PHYS_ATT_CAP_PERC
http://dwarffortresswiki.org/index.php/DF2012:Creature_token

so, strength, that max embark value is 2250, then a new binary frequency will account for 2250 to 4500.

so now instead of being
<1% 0 to 1st bin (below lowest embark possible value)
1 to 99% being the 6 bins (old 98/6% per bin)
>99% past 6th bin (i.e. past 2250)

will now be
<1% below lowest embark possible value)
1 to 99% being 7 bins (so 98/7) 7th bin being 2250-4500 for strength
>99% beyond last bin

it's important to have a 99%+ for multiple castes (like mwmod) because the castes bin's values are factored by the caste frequency (which is kind of confusing, but anyways).

Just posting this for anyone who cares.

[Maklak]

Well, as long as I get enough different displayed values for attributes and I can see who is better at a glance, I can manage.

Regarding your previous post, I really see no point in using different methods for this and taking an arithmetic weight for them. I didn't answer it, because it was quite difficult for me to figure out what you wanted me to do. The way I see it, you're over-thinking a problem with a simple solution. You may want something complex for advice on what to assign to a Dwarf, but the part that just picks what values to display for an attribute need not be that way.

Yes, you can compute the median, assume it's 50 and actually do a lot of calculations on bins, get the cumulative distribution function for an attribute that takes into account the caste frequencies, then use it to map that attribute values to their cumulative distribution function, so that for each attribute the displayed value is it's probability of being equal to this or less for a starting dwarf. But:
1) You still have a problem with high physical attributes from military training and 1% above starting maximum is a very conservative estimate, so you'll still end up with some "99+" after a few years, which is against the point of displaying so many values. 
2) Some other methods would be simpler and more intuitive, take less effort, coding time, debugging, etc, while being effective too.

Consider my proposal from some time ago:
int disp = (attr + 25) / 50;    // Use "00" for >99
It has a resolution of 50, which to me is sufficient.
You can guestimate the attribute at a glance, unlike the bin method. 20 is obviously around a thousand.
The density is the same everywhere, so 30 is 1.5 +- 0.06 as much as 20, while with the bin method you only know that bigger is better, but not by how much.
No "Dead zone" between 39 and 63.
Yes, it will almost never use the full spectrum of numbers, but it doesn't need to. For a normal population you should get values in <500, 2000>, perhaps more, which maps to <10, 40>. 30 values with resolution 50 would suffice for me here.
Yes, you loose the information about the median (or average), but it could be displayed in an extra row, like those that are above dwarves in labour assignment grid.
You have no irregularities for weird attribute distributions, such as [1000:1000:1000:1000:1000:1000:1000] or [1000:2000:3000:4000:5000:5000:5000] or whatever else the modders come up with.

So to me, the simple method is preferable here, but I can manage with other methods, as long as with what you use:
1) Higher value always means higher attribute - v 20 beta satisfies this.
2) A lot of different values are used and they are easy to read at a glance.

[thistleknot]

99% is double starting (which is based on PHYS_ATT_CAP_PERC token) maximum possible embark
example
strength
450:950:1150:1250:1350:1550:2250
double 2250 is 4500

0-450 is <1%
450-4500 is 1-99%
4500-5000 is 99%+

the importance of bins is to allow us to compare place markers between diff attributes. those place markers r defined in the raws and represent a populations breakdown of probability and offer meaningful tracking of progression.

place markers means
take agility
150:600:800:900:1000:1100:1500
1500 agility is the same % wise as 2250 strength
which matches probability of distribution (at least concerning embark, and what can be expected each migrant wave)

what i was trying to do w the starting value (the post u didn't comment on) was to somehow include a dwarfs possible max attribute he can train up to. ie max potential (which is 200% starting).  so two dwarfs at 2500 strength would be rated differently by comparing their max attainable strength.

This is where I was hoping you could help Maklak.
comparing their starting values shows what their max value could be.  Two dwarf's with the same current value would be compared by: A dwarf with a higher max potential value would be rated higher than a dwarf with a lower max value.

Kind of similar to how skill rating was adjusted by comparing their rate.  Of course the rate formula was a bit complex, I was hoping you could help offer an idea for this as well.