MantisBT - VCMI
View Issue Details
IDProjectCategoryView StatusDate SubmittedLast Update
0001774VCMIMechanics - Objectspublic2014-04-18 17:342014-05-10 16:33
Reporterbeegee 
Assigned Tobeegee 
PrioritynormalSeveritytweakReproducibilityN/A
StatusclosedResolutionno change required 
PlatformOSOS Version
Product Version 
Target VersionFixed in Version 
Summary0001774: Check random distribution for various game aspects (primary skills, secondary skills, etc...)
DescriptionI don't know if this is an issue, but I think someone (preferably a tester, ideally with up-to-date VCMI build) should check if the distribution of random events is balanced. A random event would be the distribution of primary skills, secondary skills, etc... I recognized this when I was refactoring random number generation, for 90%+ gameplay relevant random events the same generator is used.

Equal distribution can only archieved when for every game aspect a seperate generator will be used. I don't know if this is possible when generators are shared for different features.
Steps To ReproduceE.g. You may test randomness of primary skills with newly implemented auto-leveling for heroes. Set a high exp for a hero and play around with it a few times and look at the primary skills stats. IMO I archieved it that a stronghold hero had much higher defense skill. (Warning: This has to be verified!!).
Additional InformationPerhaps someone can list all gameplay relevant random events, at least the most important, that would be nice.
TagsNo tags attached.
Relationships
Attached Files
Issue History
Date ModifiedUsernameFieldChange
2014-04-18 17:34beegeeNew Issue
2014-04-18 17:34beegeeStatusnew => assigned
2014-04-18 17:34beegeeAssigned To => Warmonger
2014-04-18 17:35beegeeDescription Updatedbug_revision_view_page.php?rev_id=2555#r2555
2014-04-18 17:39beegeeNote Added: 0004634
2014-05-04 14:47WarmongerNote Added: 0004646
2014-05-04 14:48WarmongerNote Edited: 0004646bug_revision_view_page.php?bugnote_id=4646#r2561
2014-05-10 16:05beegeeNote Added: 0004651
2014-05-10 16:06beegeeAssigned ToWarmonger =>
2014-05-10 16:06beegeeAssigned To => beegee
2014-05-10 16:06beegeeStatusassigned => feedback
2014-05-10 16:33WarmongerNote Added: 0004652
2014-05-10 16:33WarmongerStatusfeedback => closed
2014-05-10 16:33WarmongerResolutionopen => no change required

Notes
(0004634)
beegee   
2014-04-18 17:39   
Just as a info, for secondary skills are separate random number generator will be used, so that shouldn't be an issue.
(0004646)
Warmonger   
2014-05-04 14:47   
(edited on: 2014-05-04 14:48)
http://dilbert.com/dyn/str_strip/000000000/00000000/0000000/000000/00000/2000/300/2318/2318.strip.gif [^]

Human testers are not suitable for testing randomness of any kind. Also, I see no reason why using same RNG with for different area would affect distribution. random number generator is random and subsequent rolls should be random and independent as well.
(0004651)
beegee   
2014-05-10 16:05   
Cool image! OK, perhaps we could write unit tests to check distribution.

So it isn't an issue to use the same RNG instance for different areas? I'm not sure, that's why I created this mantis point to check if it's really a problem.

Someone had implemented secondary skill selection by using a separate RNG. It would be nice to know why a separate RNG was chosen.

Warmonger, if you think this isn't an issue, feel free to close this mantis point.
(0004652)
Warmonger   
2014-05-10 16:33   
Because secondary skills are not really random, they follow certain algorithm.

Closing them.