MArcomage

http://picasaweb.google.nl/lh/photo/sWYNPN75aUotlwoSeIB43Q?feat=directlink
http://picasaweb.google.nl/lh/photo/lpbVCkZiqaDJRCCPuzjM0A?feat=directlink
http://picasaweb.google.nl/lh/photo/kb_XaE622Jjcdhb5COKtaw?feat=directlink

As you can see I tried to discard the same position with the second militia as I did with the first. Looking at the end situation would almost make me believe I just had bad luck and received the same card as I tried to discard. Almost, since, if you look at the place where the discarded cards are displayed, you only see Elven archer, and now scrollbar for other discarded cards.

Odd...

OK, I see what you're saying. I hope you don't mind if I paraphrase.

He had two Militia cards on hand. He used the first to discard Elven archer from the fourth position and received Road to Valhalla in its place. He then used the second Militia to attempt to discard Road to Valhalla from the fourth position, but the card was not discarded. It did not appear in the Discarded area, so it's not just a matter of receiving the same card again due to odd luck. The rest of the Militia card effect (wall and stock) did occur each time.

Odd, and probably replicable.

Actually I think this happens with Tribute as well
When you discard two cards from your own hand and two cards from opponent's hand, when the same card is received at the same position, it does not appear in the 'discarded pile'

It's not a bug, you see. Rather a feature of the current system. The discarded cards are computed by comparing your hand before and after the card was played.

If the Tribute targets itself for discard, it does not discard itself, so that's why you see only one card discarded.

Now let me explain what happened when you played that Militia. The problem is that you discarded a card and you recieved the same card to that position. So when comparing the hands the system sees no difference. The newcard flag partially helps to fix this problem.

In the case that card was in you hand and it wasn't marked as "new" and you discard it and receive the same card, the system can tell, because the newcard flag is now set (and it wasn't set before).

But in the case the newcard flag was already set for this card, the system cannot see the difference.

Fortunately this case should be quite rare according to my probabilities study.

We keep this system, because it has very simple and elegant implementation. The hand difference computation is independant of card effect code.

This flaw cannot be fixed in this system, rather the system would have to be reworked.

That make sense. It wouldn't be worth writing additional checks for such a low probability.

Check ;-)