[bug report] Game incorrectly classified as draw.

@reddish
I don't know about documentation but if you had done a search on the forum, you would have found hundreds of posts about it

Hi reddish,

I agree with you on all counts and I found the Lichess rules:
http://en.lichess.org/forum/lichess-feedback/no-mating-material-shouldnt-win?page=7#63

and it would be good if these rules were fleshed out a bit and explained somewhere.

Also, I doubt these or similar corner cases were considered! Thankfully, these cases are unlikely to occur with an opponent (deliberately or otherwise) timing out:
lichess.org/editor/5k1K/8/7P/6n1/8/8/8/8_w_-_-
lichess.org/editor/7K/8/5k1P/4b3/8/8/8/8_w_-_-

@jimj12 It took me 5 minutes and two search engines to find the Lichess official rules. Your comment is not constructive.

Hi Toadofsky,

Thanks for finding the forum posts where this issue was discussed. Clearly there are differences of opinion on this issue, and the currently solution is not optimal in all cases. Most of all, it's a shame that it seems the case that the currently implemented heuristics are only documented in a forum post.

I like the two example cases you provide; those can actually occur in real games, and I would be pretty miffed if it happened that my opponent gained half a point simply by letting his clock run out on purpose.

I'm mostly just happy that cases like these are pretty rare!

It seems my comment got misunderstood.
I said it's a bit silly to count it as a win but I nontheless AGREE that it should be a win. It could theoretically still end as win so that's reason enough.

Hmph... I need to figure out how to test a fix not just for standard chess but (regression test) for all variants:
github.com/ddugovic/scalachess/commits/insufficient_mating_material

That's some readable code. I don't even know scala and can still easily read it. :)
This could still be "tricked" by positions where the material allows for mate, but it's impossible to reach from the current position. But that might be hard really to implement.
(e.g. here neither side can win even though the material would allow it lichess.org/editor/7k/6pP/5bPp/7P/8/1K6/8/8_w_-_- )
EDIT: But in atomic that won't work, right? I never played atomic, but can't one explode the enemy king by capturing the enemy queen, even though in normal chess that material doesn't allow for mate?

#1 What you are talking about is not a bug but a difference in rules between lichess and FIDE regarding rules for drawing.

FIDE rule: Game is draw when your time run out and there is no legal sequences for opponent to mate your king.

In online play without arbiter, FIDE Online Arene is using this rule solely to all extent.

http://support.premiumchess.net/how-to/draw-cases

The disadvantage is that many positions where you have to play very very stupid and actively help your opponent to mate you will be judged as lost, making this rule be a bit silly if applied to that extent.

The lichess rule has the disadvantage that it judges certain positions where there is a forced mate as a draw.

A compromise could be letting a comp do a fast eval and judge draw if there is no forced mate and win if there is a forced mate.

#2#3

You are wrong. What do you say if black underpromote g-pawn to knight and helps white to play until this position?

sv.lichess.org/analysis/7k/5K1n/6N1/8/8/8/8/3b4_w_-_-

This is how FIDE Online Arena works. I agree it can be silly but it's true.

@blackzombie Actually @Toadofsky changed the code towards being the same as in the FIDE rules, right?
Also, what do you need the underpromotion for, you can just mate against the B alone.
Also, yes it's silly, but if you play blitz or bullet time SHOULD be an important factor, that's the whole point of it. So losing on time is simply an alternate loss condition as opposed to checkmate. And as mentioned further above otherwise you could just time out in positions that really are lost.