MArcomage

Please help me understand this card?
I need to manually discard a mage before playing unicorn?
Or it discards itself?

It discards a Mage itself. If there is no Mage in hand - then you won't get 30 gems.
By the way, can this card select lowest rarity or at least non-rare Mage? It's kinda sad when you have Shadow Unicorn and rare Mage and don't trigger Legend keyword because this rare Mage was randomly discarded.

NG_Beholder wrote:
It's kinda hilarious when you have Shadow Unicorn and rare Mage and don't trigger Legend keyword because this rare Mage was randomly discarded.

Fixed that for you!

I think this is already a very strong card, and doesn't need a buff. The possibility you describe makes the card more interesting and strategic, in my opinion.

I've had that happen to me and had a thread discussing the possibility back before we could create threads attached to the cards themselves. I don't recall exactly, but I believe I was counting on the facility gain to defend against my opponent's next card. Things like this that should work a certain way and then don't, not because of flawed logic but because of the order of operations, detract from the game as a whole.

We discussed hard-coding a check into the card itself to see if the discard effect would interfere with the Legend effect, but Mojko (rather justifiably imo) insisted that we keep the code more clean and not have to go back and edit individual cards should keyword effects be changed at a later time.

Actually, I do agree with you there. I think it's reasonable that this card can discard a rare mage, but it's confusing if this prevents the legend keyword from taking effect. I would propose the following change to how card effects are evaluated (my changes are in bold, the rest is how it's written in the FAQ):

0) Take a snapshot of the player's current hand.
1) Card effect
2) Keyword effect based on the snapshot obtained in step 0.
3) Normal drawing - new card is added in place of the played one (can be disabled or effected by card effect or keyword effect)

I think this is just more intuitive for the player, and makes things a bit more predictable. We fix the problem of legend not procing even though you had a rare in your hand when you played it. You also fix other annoyances, such as playing Corpse Behemoth with 70 undead tokens and 8 undead cards in hand, but not having the undead keyword proc.

I haven't fully thought this through, so it may be that this introduces its own set of quirky behaviour, but it seems pretty reasonable to me.

Actually, almost all game data is cached for the changes computation purposes. However, what you propose is to run keyword effect on the same data as keyword effect (player's hand). Right now they are executed in sequence which results in deterministic behavior.

I'm not sure if the snapshot you proposed also contained game attributes, but even if we consider to run keyword effect on cached player's hand, it leads to inconsistencies. For example if both card effect and keyword effect are discarding cards from hand. If both effects effect common position in hand, then we have an inconsistency.

I don't think this solution is a good idea, however we may consider to reverse the order of card effect and keyword effect execution, so that keyword effect would happen first. This would be consistent, but would also require changes on some keyword effects (Charge for example).

Alternative decision: what if Shadow unicorn discarded a rare Mage and there is no more rares in hand, player would receive some compensation? 10 gems or additional +1 Magic, something like this?

Mojko wrote:
I'm not sure if the snapshot you proposed also contained game attributes, but even if we consider to run keyword effect on cached player's hand, it leads to inconsistencies. For example if both card effect and keyword effect are discarding cards from hand. If both effects effect common position in hand, then we have an inconsistency

I was thinking of copying only the player's hand, nothing else. Could you give an example of cards that could cause the problem you describe? I don't know of any keywords that can discard cards from the player's own hand, except for flare attack, and I don't see how Phoenix or Wildfire gem could cause the problem.

If you're just worried about it being unsound in case new keywords or cards come along, then can't you just make it so that keyword effects override card effects where there's a conflict? (Or vice versa).

By the way, I think evaluating keyword effects before card effects will still have confusing effects. For example, the player might be confused by playing Dwarven Dome at 49 tower and having his wall increased.

NG_Beholder wrote:
Alternative decision: what if Shadow unicorn discarded a rare Mage and there is no more rares in hand, player would receive some compensation? 10 gems or additional +1 Magic, something like this?

I'm not sure there's enough room in the card text to add that extra conditional.

dindon wrote:
I was thinking of copying only the player's hand, nothing else. Could you give an example of cards that could cause the problem you describe? I don't know of any keywords that can discard cards from the player's own hand, except for flare attack, and I don't see how Phoenix or Wildfire gem could cause the problem.

I don't think there is a card with mentioned conflict yet. Still, I don't like this inconsistent solution.

Another experience I had recently that dindon's suggestion would help against concerned Corpses behemoth. I was timing its use such that it would trigger the Undead keyword effect, refunding enough resources that I could afford to play Undead Cavalry (selected to remain on hand, of course) on the next turn. Unfortunately, the Undead tokens only counted those Undead on hand after the mass discard effect.

Since token gain counts the played card, it didn't occur to me that this is calculated after card effect. The mouseover predicted the token gain based on what was on hand, of course. It threw off my plan in such a way that I feel it should either be corrected or occur with a warning of sorts.

All right, I'll add this issue to my TODO list ;-)