Ahh, that'd explain it.

Is this a relatively recent change then? I'm sure I've used a unit to support hold or support move (or even move) when I'm pretty sure it'll be repulsed, and support held it from elsewhere.

at least it was so when I joined 2 weeks ago - so no recent change

Has been this way always I think.
But they're working on a new code, so you can support everything that doesn't 'move'.

Its always been like this. Do you think it should be made to all non-moves are support held then? If so, I think i could work that out quite soon for you, as an update for this version - kestas is currently working on the new version.

Well, we should stick to the official rules, I don't know how they'd rule on this.
But I know that if I'm support holding a unit that hasn't moved (because its own order failed), I'd expect it to accept that support.

Well, from what i know i believe its that if the unit moves it cannot have hold support, but the other 3 move types can.
comments?

In that case, I should still have Munich.
that unit was supporting a move. It didn't move itself. So it should have been supportable.

don't mix up the current implementation of the rules with the official rules

in the current implementation, only an army that was ordered "hold" can be given "support hold"

I know what happened and why in this instance, we're now discussing what should be made to happen- which involves making it official rules compatible.

It's not true that I'm working on the new version. :( I had time to devote to 0.75 over the hols but I won't have another large chunk of time like that for a while.

I expect 0.8 will be a piece by piece upgrade so I can fit it in with everything else. Fixes to the adjudicator will probably be in the order of fixed coast support, support holds can be applied to any non-moving unit, improved order handling, and finally convoys.

Thanks for your input and so on, you're very thorough beta testers :)

O, sorry then!
I have some hols v soon, so I should be able to get ahead of my work and get some more stuffs onto here during then. So, more upgrades coming soon...

Okay. I've lost some units this way too. I'm pretty sure you can support s'long as it's not going anywhere. ;)

OK, now I'm getting really confused.
A unit that has been attacked cannot attack itself, right?
So how come when I attack two provinces (Munich and Marseilles) that could attack my A Burgundy, this breaking their chance of attacking, the support move of my A Burgundy is broken?

Spring 1907, Diplomacy: Your fleet at Spain attempted to attack the army at Marseilles, but failed.
Spring 1907, Diplomacy: Your army at Burgundy was attacked by the army at Marseilles, but sucessfully defended.
Spring 1907, Diplomacy: Your army at Ruhr prevented the army at Munich from giving a support move.
Spring 1907, Diplomacy: Your army at Burgundy was attacked by a army at Marseilles, and had to defend itself, so couldnt support a move elsewhere.

So my A Ruhr stopped the A Munich from supporting a move, but my F Spain attacking Marseilles didn't stop them attacking and breaking my suport move.

Now, I know what you're going to say is that it has to be a successful attack to stop the defendant carrying out its order. But it's not- a support hold or move can be broken with 1 unit, and I swear I've broken direct attacks with 1 unit before.

>So how come when I attack two provinces (Munich and Marseilles)
>that could attack my A Burgundy, this breaking their chance of
>attacking, the support move of my A Burgundy is broken?
Attacking a unit doesn't stop that unit from attacking. Attacking a unit stops it from supporting or convoying, but not attacking.

Munich support move into Burgundy from Marseilles
Burgundy support move into ?? from ??
[Spain]->[Marseilles]->[Burgundy]
[Ruhr]->[Munich]

- Ruhr is moving into Munich, and Marseilles is moving into Burgundy. Unless Munich is supporting a move against Ruhr or Burgundy is supporting a move against Marseilles both support moves get canceled out right away.

- You're left with just [Spain]->[Marseilles]->[Burgundy]
[Ruhr]->[Munich], which all cancels out and leaves no-one moving.

The point is that the order given to Burgundy was prevented (a support move into Paris from Picardy).
This order should have carried through, because there was no unit capable of attacking Burgundy, as both of those adjacent (Marseilles and Munich) were being attacked- and if they are being attacked, surely they must defend themselves and are therefore incapable of attacking and thus breaking a suport move.

Hmm I looked it up just to be sure, see page 11 diagram 18 of the rulebook; "even though the Germany Army in Munich is dislodged by a Russian attack, it is still able to cut the support of the Russian Army in Silesia"

Not a problem with the adjudicator, for once ;)

No. An army can break another unit's support even if it itself is attacked.
Germany: A Munich to Burgandy, A Marseilles supp Munich to Burgandy, F Western Med to Spain
France: A Spain to Marseilles, A Paris to Burgandy.
This will result in all units bouncing out of burgandy. Spain's attack on Marseilles breaks Marseilles ability to support but Spain cannot move into marseilles because there is a unit there. Western Med's attack on Spain does nothing as the army currently in Spain does not move out and thus it is a 1 v 1 battle. This leaves Paris and Munich both attempting to move into Burgandy. Since they are equally matched, they bounce.
The adjudicator is 100% right.

figlesquidge sent in a patch that fixes this, so now you can support hold any unit which isn't moving. Good job :)

Attacks are applied before everything else, therefore no matter what happens, attacks cannot be cut. Supports, however, can be cut by attacks. I believe that's how the official rules work. =)

There are paradoxes and so on, trying to eliminate all of each type of move in one go just doesn't work in an adjudicator, you have to take a decision based approach; you ask "can I resolve this move?" if it depends on another move "can I resolve this move?" and so on. Then if you've resolved all moves that don't depend on other moves, but there are still moves left, then you have a paradox. Paradoxes involve multiple moves and are detected and resolved in one go, and then you continue selecting moves at random and asking "can I resolve this? what does it depend on? can I resolve this?"

This is how an adjudicator should be written, all attempts to do it other ways (including the way the my current one works) end up going nowhere fast, so somewhere in the future there'll have to be big changes made, if not a rewrite

"figlesquidge sent in a patch that fixes this, so now you can support hold any unit which isn't moving. Good job :) "

it works! thanks for that, it's a great achievement!

Is that patch going on Sourceforge? My school runs its own phpdiplomacy server and the faulty support system is a pain in my ass.

DougOfDoom; you can grab the code revision phpdiplomacy.net uses from CVS on sourceforge

(but it may be dangerous; database changes are often needed between revisions)