You can't cut the support of an army that is giving support against you. Check out the last rule at http://phpdiplomacy.net/intro.php: "a unit which the support is being used against cannot force the supporting unit to defend itself".

Or am I wrong about this? (I'm quite new to diplomacy, sorry.)

I've never found this problem before, and looking in the code It doesnt look like this is what happens. Are you sure this is what occured? Could you and the other player[s] involved post up the gamemaster responses please?

It's a bug because you can't cut the support of a unit which is supporting a move against you. Though I thought that the adjudicator worked fine here, could you tell me the game ID and/or post a screenshot?

>Why not use an existing adjudicator?
First off I'm a computer science guy, not a web developer. I like dealing with security, the adjudicator, and SQL optimization+locking, not HTML and CSS, so I wouldn't want phpDiplomacy to become a shell for an existing adjudicator for that reason.

But mostly phpDiplomacy is about being user friendly. This means letting users get info into and out of the adjudicator easily. This has caused the order validation and processing code to become separate from the adjudication code, as it has to be if you want it to not let users enter bad orders, and especially when you want things like entering orders with drag and drop. If I had started off with an adjudicator I'd have to worry about whether or not it can validate orders without processing them, etc. With a native adjudicator you can just reach in and change it as you want.
The goal is an adjudicator which is both DATC compliant (Kruijswijk bugs me about this whenever I mention phpDiplomacy on the mailing lists ;-) ), and meets the needs of a web 2.0 Diplomacy implementation, and I don't think you find that in any adjudicator around.

You're right that the FAQ should give more warning though, thanks for pointing this out.

About the support problem: This just happened to me as well, but it turned out the adjucation was correct. In addtion to the rule that a unit cannot cut support against its self, the rules say that a dislodged unit's support is ALWAYS cut, even if the unit that dislodged it was the one it was supporting against.

This may not be what happened in your game, but it seems likely.

Thank you for the answer, I find it reassuring. Still I'm not sure if you could not go the way of having a custom made validator between the user and an out-of-house developed adjucator. I praise the user friendlyness of phpDiplomacy, and validation is a big part of it, but I'm not sure it needs to be tangled with adjucation.

But if you enjoy working on the adjucator then go ahead, I'm sure there's nothing wrong with it that a few complete rewrites can't solve :). Seriously if you have the enthusiasm you can certainly fix the bugs -- I'm sure all of the other adjucators had tons of them too. Too bad it's in PHP. Were it written in Python I would consider giving a hand. (Of course if it were written in Python it would not have any bugs, hehe :).)

Oh, right, about the possible support cutting bug!

It happened in this game: http://phpdiplomacy.net/board.php?gid=377

Here's a piece of screenshot:
http://csomalin.csoma.elte.hu/~cyhawk/supportcut.png

I'm Austria. My orders were Gal->Rum supported by Bud and Ser (I had a unit in Galicia of course :)). Russia's orders were Ukr->Gal supported by War and/or Rum and a support hold for Rum from Sev. Turkey also supported hold in Rum from the Black Sea (the bastards are conspiring against me :)).

So I think the results are correct -- Rumania was defended and Galicia was overwhelmed (I've retreated to Vienna as seen in the screenshot). It was just this strange message that caught my eye in the GameMaster box.

It looks like I'm consistently mispelling adjudicator :). Sorry about that, I'll try to stop.

From your above statement the adjudicator was correct. No bugs here! :)

But it is not the case that Metternich471 described. Rumania's support was broken -- while it was Galicia that was dislodged. Please explain how Galicia could break a support against itself.

Hmm, good point. I know the adjudicator usually processes that set of orders corrects (A->B supported by C, B->C results in A moving in to B). Maybe there's just something funky with the gamemaster message?

Taking a look at the source it appears that the "can't cut support against self" rule is only taken into account when the support cutting attack (Gal->Rum in my case) is itself unsupported.

This makes it work for a large number of cases, but I don't think for all of them. In my case it seems that Russia must have supported the attack against Galicia from both Warsaw and Rumania otherwise his attack would have failed and incorrectly so!

I could of course be reading the source the wrong way.

I agree with your logic. However, looking at the source, I cant find any reason why this didnt resolve correctly.
What do you suggest

I'm not sure, and I'm in a hurry now, but my guess would be that in phase_diplomacy.php the parts with comments "Stand all support moves which are being attacked with >1 support" and "Stand all units which are attacking a support move with 1 support" should be changed to one piece of code with the comment "Stand all support moves which are being attacked from anywhere but the province against which they are giving support". I don't think the attack moves against supporting units have to be handled in any special way.

That would be my suggestion, but it's not worth much, because I don't understand the code that much. I have no idea why it is currently implemented the way it is. Why handle attacks with support differently from attacks without support?