1. I like to solve acm-like problems.

2. I like timus but IMHO sgu is better.
Just one example... (I have more but they all alike)
I wrote solution to problem 1005. Stone pile.
Timus said that it's ACCEPTED. Then I generated
about 20 random tests. My "AC" program FAILED
on 50% of these tests! It's not good at all!
BTW, I don't know about such examples on SGU.

3. Good luck to all who tries to make timus better!
No need to be upset 'cause UVA is worse than TIMUS

:)

Why are you so sure you program failed them?
It's rather difficult to check large random test without the correct answer (I mean, finding it yourself).
In case it's TL, maybe your computer has less performance
or you are using a different check system that counts astronomical time (if you have a lot of other processes it might have considerable effect).

Who said my random tests were large?

I can check them manually! So correct answer is
pretty obvious.

Problem 1005 is very old. Problemsetters of that time couldn't imagine that their problems will be solved in 2006 by thousands of people.

I can't remake all problems in Volume 1 :)
Please, appeal to weak problems of nowadays only.

In fact Romanovsky describes method that looks like Hopcraft-Karp, but in is O(V^3). (He thinks that adding a chain to matching is O(1))

there are so many discussions. That is why I have written to increase the number of dialogues. :)