Bay 12 Games Forum

Please login or register.

Login with username, password and session length
Advanced search  
Pages: 1 ... 25 26 [27] 28

Author Topic: Weapon research  (Read 149572 times)

zagibu

  • Bay Watcher
    • View Profile
Re: Weapon research
« Reply #390 on: December 25, 2010, 06:52:40 pm »
I've tried to run your tests on my machine, but there were some problems with the item strings (you have to be careful there, it hasn't got ANY error tolerance). The best thing would be to first manually set up the test and exactly write down the filters you employ letter by letter.
This corrected version worked.

Code: [Select]
Sword - High Master x Dabbling/10
1/0/0/79/0:13,3:13,17:13,18:13,19:13/steel short sword:1,steel mail:1,steel breast:1,steel helm:1,steel gaunt:2,steel high:2,steel greav:1,steel shield:1
2/0/0/79/0:0/steel short sword:1,steel mail:1,steel breast:1,steel helm:1,steel gaunt:2,steel high:2,steel greav:1,steel shield:1

It created a 5MB file, because the fights were pretty long. The reason is that the unskilled dwarf learns so fast, he is usually on legendary skills before having received any serious wounds, and from there on, the fight is pretty even. I think you won't see a significant advantage for the skilled dwarf. You could try modding the dwarf creature, so that it can no longer learn skills. It's what I have done for my tests, but I can't currently get to my long-term storage to retrieve the correct definitions.

I'm also going to upload a new version of the simulator that works with the newest DF in a minute. It also contains a readme.txt with some tips.

Note that if you have something else besides Windows XP 32bit, and find that even the corrected version still doesn't work, you're on your own. Unfortunately, I don't have Vista or 7, and I don't even know whether AutoIt works there.

[Edit] The new link is up in the original post.
« Last Edit: December 25, 2010, 07:07:15 pm by zagibu »
Logged
99 barrels of beer in the pile
99 barrels of beer!
If some dwarves know the way to the pile
0 barrels of beer in the pile!

win32anon

  • Bay Watcher
    • View Profile
Re: Weapon research
« Reply #391 on: December 26, 2010, 01:14:40 am »
Thanks, now is working properly. But where is the data stored? In the txt file is just a copy of the gamelog.

Exemple:
Spoiler (click to show/hide)

Isn't it converted into raw numbers?
Logged

Shinziril

  • Bay Watcher
  • !!SCIENCE!!
    • View Profile
Re: Weapon research
« Reply #392 on: December 26, 2010, 01:35:09 am »
Quote from: win32anon on December 26, 2010, 01:14:40 am
Thanks, now is working properly. But where is the data stored? In the txt file is just a copy of the gamelog.

Exemple:
Spoiler (click to show/hide)

Isn't it converted into raw numbers?
I assume the second program (the "evaluator") is used to convert the text into numerical data. 

Zagibu, has the evaluator been updated for 31.18 as well?  Attacks with biting/kicking/punching are fairly common now if the opportunity comes up, and should obviously be excluded from the armor testing data for individual weapons, as well as any weapon attacks that were parried or missed (which I assume were already excluded). 
Logged
Quote from: lolghurt
Quote from: Urist McTaverish
why is Dwarven science always on fire?
Because normal science is boring

veok

  • Bay Watcher
    • View Profile
Re: Weapon research
« Reply #393 on: December 26, 2010, 02:06:46 am »
Quote from: Orkel on December 15, 2010, 09:46:51 am
Isn't bronze better than iron IRL? Why did 31.12 change this to be otherwise?

Quote
Though bronze is generally harder than wrought iron, with Vickers hardness of 60–258 vs. 30–80, the Bronze Age gave way to the Iron Age; this happened because iron was easier to find

From what I recall, they were both at modern values, and then one got lowered down to "period" values (i.e., with more limited metallurgy) but not the other, causing the discrepancy.
Logged

win32anon

  • Bay Watcher
    • View Profile
Re: Weapon research
« Reply #394 on: December 26, 2010, 02:55:36 am »
And how do I use the evaluator?
« Last Edit: December 27, 2010, 06:03:10 am by win32anon »
Logged

Max White

  • Bay Watcher
  • Still not hollowed!
    • View Profile
Re: Weapon research
« Reply #395 on: December 26, 2010, 03:24:35 am »
*Reads first page*

Remind me, why did I ever bother with adam axes? Steel seems to do better...

EDIT: Wait, just noticed how it does against steel armour, so thats why...
Logged

win32anon

  • Bay Watcher
    • View Profile
Re: Weapon research
« Reply #396 on: December 27, 2010, 06:06:00 am »
So, I just finished running the default test I got 180 reports and now i want to convert it to a spreedsheet, how do I use the evaluator? Is the .pl file that I have to run? But I have no clue on how to do that.
Logged

zagibu

  • Bay Watcher
    • View Profile
Re: Weapon research
« Reply #397 on: December 28, 2010, 10:53:21 pm »
It's a perl script. I know, this is not perfect, but for me, it was the easiest solution, because perl has a fast regexp implementation and was supported on one of my OS (Linux). I guess you can run it with cygwin on Windows. Or there might be another Perl interpreter. The perl script generates numerical data that you can paste into the provided spreadsheet.

The evaluator has not been updated since the beginning, but, IIRC, it never counted biting, etc. You can take a look at the perl script and find this information in the patterns that I use to extract the data.
« Last Edit: December 28, 2010, 10:55:38 pm by zagibu »
Logged
99 barrels of beer in the pile
99 barrels of beer!
If some dwarves know the way to the pile
0 barrels of beer in the pile!

Shinziril

  • Bay Watcher
  • !!SCIENCE!!
    • View Profile
Re: Weapon research
« Reply #398 on: December 29, 2010, 01:12:59 am »
Looking at the script, it seems to look for attacks with the keyword bashes/hacks/slaps/slashes/stabs/strikes, which should exclude all natural attacks.  It does mean that the hit percentage of, say, axes against armor will be affected by their flat slap attack, but I suppose that is reasonable for actual ingame combat.  While it might be undesirable for pure scientific experimentation, that can be easily solved by modding the weapons used for the simulations to only have the attack of interest. 

I don't have any experience in Perl, but upon casually examining the script, it seems that attacks that have been parried (in addition to attacks that have been dodged) will simply be ignored in the calculations.  This seems reasonable, as attacks that do not contact the enemy are of no interest to us. 

How do you define a file as a "testreport" file?  I've managed to get a Perl interpreter working (ActivePerl), but the script seems to do nothing at this point.  I suspect this is because it is not actually receiving any files to analyze. 
Logged
Quote from: lolghurt
Quote from: Urist McTaverish
why is Dwarven science always on fire?
Because normal science is boring

zagibu

  • Bay Watcher
    • View Profile
Re: Weapon research
« Reply #399 on: December 29, 2010, 08:54:38 am »
The script either processes all files given on the command line as parameters, or, if none given, it looks for files in the current directory with names beginning with "testreport". It creates an output file for every processed input file.
Logged
99 barrels of beer in the pile
99 barrels of beer!
If some dwarves know the way to the pile
0 barrels of beer in the pile!

Shinziril

  • Bay Watcher
  • !!SCIENCE!!
    • View Profile
Re: Weapon research
« Reply #400 on: December 29, 2010, 01:01:40 pm »
This would be SO much easier if the test logs generated by the simulator started with "testreport" (rather than "df_arena_report").

Okay, I renamed all of the files.  Running the Perl script produced one text file for each test report, as stated.  Each contained exactly the same text:
Spoiler (click to show/hide)
and absolutely no numerical data at all.
Logged
Quote from: lolghurt
Quote from: Urist McTaverish
why is Dwarven science always on fire?
Because normal science is boring

zagibu

  • Bay Watcher
    • View Profile
Re: Weapon research
« Reply #401 on: December 31, 2010, 11:55:32 am »
How did you run the perl script? I'm gonna try exactly your way and see what I get on my system. Also, what kind of OS are you using? Can you post some lines of a testreport?
Logged
99 barrels of beer in the pile
99 barrels of beer!
If some dwarves know the way to the pile
0 barrels of beer in the pile!

Shinziril

  • Bay Watcher
  • !!SCIENCE!!
    • View Profile
Re: Weapon research
« Reply #402 on: December 31, 2010, 12:07:32 pm »
Spoiler: Example testreport (click to show/hide)

Using 64 bit Windows 7 here.  I ran the Perl script by double-clicking it (after some shenanigans to get the .pl extension properly associated with the Perl interpreter).  I had all the testreports and the script in their own little folder. 
Logged
Quote from: lolghurt
Quote from: Urist McTaverish
why is Dwarven science always on fire?
Because normal science is boring

zagibu

  • Bay Watcher
    • View Profile
Re: Weapon research
« Reply #403 on: January 01, 2011, 07:48:54 am »
Ok, I've reupped the evaluator stuff. There was a problem with recognizing rounds. Another thing is, that calculate.pl has to be adapted, if you test other creatures than dwarves. Currently, it only counts attacks of "Dwarf 1". If you test other creatures, or want to test attacks of the second dwarf, you have to change the pattern called $pattern_start.

With the fixed evaluator, I got the proper statistics out of your test file.
Logged
99 barrels of beer in the pile
99 barrels of beer!
If some dwarves know the way to the pile
0 barrels of beer in the pile!

Greep

  • Bay Watcher
    • View Profile
Re: Weapon research
« Reply #404 on: February 23, 2011, 06:45:05 pm »
So out of curiosity, how would this apply to what an ideal fighting force vs FBs and clowns would be?  It seems to suggest having a front line of adamantine axedwarves with a reserve of silver hammerdwarves for the really dense creatures.  Or would one be just plain better off with axedwarves for the occasional severing?
Logged
Pages: 1 ... 25 26 [27] 28
« previous next »