Farthest from home center, other than that I don't know.

There's a known issue with how the auto-disband priority is calculated.

South Africa has orders entered so the answer is there will be no auto disband.

Furthest from nearest Home SC, fleets before armies, in alphabetical order. There are some issues with this site, but I don't know if they've been fixed/what exactly they are.

From here: http://www.wizards.com/avalonhill/rules/diplomacy.pdf

"If a country in civil disorder has to remove units, the units farthest from the country are removed first. If units are equally distant, then remove Fleets before Armies and then in alphabetical order by the provinces in which they are located."

webdip has a table that is built one time for each variant that sorts the territories from furthest to closest including tiebreakers for each country. At one point that table was messed up for Classic and possibly the other variants. I do not know if that was ever fixed.

Should be distance from SC, and I think it's just alphabetical order from there, with the fleet/army designation skipped.

*HSC

Rules say from home SC

hmmm. I shall compare this to actual results in the World games and see if it is predictive. Many thanks, all!

The rule held until I examined this game: gameID=72817
Autumn 2009, Froz had two SC's and three pieces: one on the ocean, one in HSC Len, and one in SC South Pole. The auto-disband destroyed the piece in his home SC.

Alas. Any comments?

You have too much time on your hand?
And how do you stand playing world games? They are wayyyy too long for me. I've only successfully completed one, and I got really bored.

as far as i can tell the code just says 'ORDER BY i.destroyIndex ASC LIMIT 1'

and my mysql just isn't good enough to understand... (never really paid any attention to INNER JOIN commands)
and the comment just says
'Remove units at random for any destory orders that weren't successful' - which i don't think this code does...

@Indy - How do you know it was an auto disband in FA in gameID=72817?

@Ora - Why *don't* you pay attention to INNER JOINS? I know there is a table that contains the disbands for each SC relative to each nation providing a weight (destroyIndex), but if that table has incorrect entries, then the incorrect unit could get disbanded. The only way to fix it would be to review all the data and that could betime consuming for the world map.

yep, i believe there is a table containing the disbands, but clearly i can't tell from a look at the code because i lack the mysql skill...

infact, i think i've always used php where i could, and simplified my sql queries (letting me do any join like operation outside of the database... pprobablly far less efficient that a well setup database)

Yes, and the destroy index is the order in which they should be destroyed for that nation. So by ordering by the index in ascending order, you get the first one to be destroyed.

Except that doesn't account for the fleet before army tie breaker. But as PE says, perhaps that part is being skipped, not done correctly.

I haven't looked at the table, but there could be two entries with a fleet or army flag for coastal territories and the inner join restricts to only the appropriate entry for the current unit type.

So lets say you have territories A-G

A is 3 units away from F and only handles armies
B is 3 and is sea zone so only fleets
C is 3 and is coastal so handles both
D is 2 away and only does armies
E is 2 away and only does fleets
F is 2 away and only does armies

Deletion Indexes:
A would have 3
B would have 1
C with an fleet would have 2
C with an army would have 4
D would have 6
E would have 5
F would have 7

End result in order would be:

BF - 1
CF - 2
AA - 3
CA - 4
EF - 5
DA - 6
FA - 7

I haven't had a chance to look at the auto-disband table code yet but have been wanting to. Your design seems ideal.

1) The code is on github (https://github.com/kestasjk/webDiplomacy). You don't even need to download it to take a look, as github does a nice job of providing the ability to browse and view code (even with syntax coloring) right in the browser.

1a) (side note) Github is really cool. For a trivial change in the timestamp code, I was able to fork, edit, commit, and pull request all within the browser without ever even downloading the code to my machine.

2) There are known issues with how auto-disbands are handled for CDed powers. I documented an issue in a classic game where DATC test cases 6.J.4 and 6.J.9 (see http://web.inter.nl.net/users/L.B.Kruijswijk/) seem to be violated and forwarded that onto Kestas. He confirmed that it appeared to be a bug and that it was on his todo list.

2a) There also probably issues in the world games as well, as there also seem to be some known issues in how the geography is handled (e.g., invalid retreat paths, etc.). I personally have not taken a deep look into those bug reports so I am not familiar with those apparent problems.

2b) Sleepcap, the maintainer of vdiplomacy, recently submitted a pull request on github with several fixes and updates to the variants. It does not appear that those changes have been rolled in yet, and I'm not sure how many of these specific issues that code has addressed, but it's worth taking a look at that to make sure you're not solving any problems that have already been fixed.

3) Another issue is that the entire sections 6.H, 6.I and 6.J of the DATC test pages seem to be omitted from the DATC tests run by WebDip (see http://www.webdiplomacy.net/datc.php). So, another important item for the todo list for coders that want to contribute would be to code up the test cases of those sections.

3a) The DATC test cases seem to be imbedded into an SQL install script (https://github.com/kestasjk/webDiplomacy/blob/dev/install/install.sql). See lines 195-221, in particular lines 204 and 221, which are very long and where all of the action is contained. It seems that it would be quite painful and tedious to manually add new test cases to this file, but hopefully Kestas has a script to generate them from a more manageable format.

As this thread is starting to get more technical, it would probably be a good idea to take this discussion into the developer forum (http://forum.webdiplomacy.net) in order to have a more stable record of this discussion.