I believe that fleets are destroyed first. Though I do know there is a bug that we're working on fixing related to this I don't think it would apply here.

for convenience: i think that this is the game id.

gameID=147463

Really interesting case, I went digging into the source to figure out what was happening. Here's the actual code that determines the destroy order:
https://github.com/kestasjk/webDiplomacy/blob/6a7b05d6871d85fa057a232e3a92742e88fe9047/variants/install.php

The first consideration is the distance from home supply centers, which is 1 for all three of Livonia, Ukraine and the Black Sea. The second consideration is whether they are armies or fleets, and the third is the alphabetical order of the territory. Creating the string this way gives us

0001 Army Ukraine
0001 Fleet Black Sea
0001 Fleet Livonia

which explains the destroy orders.

Hmm, I wonder if the home center has to be owned by the country in order for it to be counted as a home center. I'll have to look into the code and see what's happening there.

http://goo.gl/kmCHQj , previous URL ran out of the page., Line 270.

Not sure how the code considers "home centers" but it might be different in games like Modern Diplomacy.

Yes, as santosh points out, there's currently a bug that causes armies to be destroyed before fleets - when it should be the other way round. The comments in that code indicate that it's more of a case of a mistake reading the rules rather than broken logic.

The fix is simple - but it would also be good to implement the DATC build tests too - since the site doesn't currently run them.

In the webdip implementation, the "home centres" are defined as the original centres owned by that power. They don't change if you don't own them. I believe this is what the DATC tests assume too, but I can't remember if it's tested or not.

^ TC is on the case, on the money. This has been a pet peeve of mine to solve. I can't stand the fact that we're not conforming to the actual rules.

From what I remember, it doesn't sound too hard to implement. Currently, the DestroyIndex table is initialized at the start, and left unchanged. It should be very few lines of code to make it update when you do the supply center recounting.

acornist: Next time this happens, demand a pause from your fellow players and the mods until they fix the disband.

This is really annoying, actually. Just I alone have brought this up a million times:
threadID=542473
threadID=528518 (pages 42–44)
threadID=662706
threadID=829390
threadID=977415
threadID=1061455
threadID=1108136

TC made the code to solve this. However, we'd want to test run it properly before it's implemented and that takes time. But it's closer to being solved than it ever was.

@JECE, it's not something I can fix, even with database access. Messing with the table that holds the game matrix is extremely complicated and not worth potentially messing up 500 games. My hope is we'll get a code fix ready soon, but the mods don't have the ability to switch the position of a unit.

By the way, I don't see anything about convoys. They need to be taken into account in the computation of the shortest distance:
"If a country in civil disorder has to remove units, the units farthest
from home (most distant from the nearest home supply centre as computed by the shortest available route, including convoys) is removed first. If units are equally distant, then remove Fleets before Armies and then in alphabetical order by the provinces in which they’re located."
* http://www.stabbeurfou.org/docs/regles/en/diplomacy_rulebook.pdf (pg. 19)
* http://www.stabbeurfou.org/docs/regles/en/Compl_Hasbro_en.html

jmo1121109: I swear that I've seen a unit moved before. Perhaps it was an exceptionally egregious error.

We can turn back the phase of a turn in the case of a site error, and it's possible Kestas might have a while ago, but unless he decided too it couldn't be done. Hmm, as I'm typing this I think I came up with something that might work. So if it comes up again we'll see what we can do, but hopefully we'll have the code fixed before that.

TheMinisterOfWar and jmo1121109: Sorry if I sounded rude earlier, by the way. I meant to say that I was annoyed because over four and a half years had passed since the glitch first cropped up and in much of that time its status was unknown. Most recently it was just assumed by posters that the issue had been resolved. I'm glad to hear that you are all aware of what is going on now.

As far as testing goes, I think the DATC can be quite useful as I conjectured here: threadID=829390.

JECE, are you offering to implement the missing DATC tests?

Any updates?

A_Tin_Can: How would I go about doing that?

In fact, I could just use Little Dipper (http://www.getjar.mobi/mobile/388543/) and identify where the DATC tests fail, but I'm sure that the webDiplomacy code has changed since SunZi wrote the app. Moreover, wouldn't it would make more sense to run the tests once the outstanding problems were fixed?