Slow Event Polling (Linux x64)

[Senzaki]

Actually, it is not detected as a joystick in the end, because the open call fails. But if SFML tries to open it, it is because it is detected as a joystick at first. I've added printf when open is called, and it is called once per pollEvent call.

[Laurent]

Ah, I see. I should add a flag so that joysticks that failed to open don't retry every time pollEvent is called. Sounds like a lot more sane

[Senzaki]

In this case, would it be possible to add a function that would unset this flag ? It may be useful in some cases.

[Laurent]

In which cases? If I fail to open an input node, I think it's safe to assume that I'll fail to open it during the lifetime of the application.

[Senzaki]

You're right, if anything changes users can just restart the application.
But about the high-res mouse problem, I must admit I don't really understand why it would be inconveniant to offer the option of having mouse motion synced with the display() call, since it's an option offered by several famous commercial games (like Starcraft II).

[Laurent]

It might be the only solution, but I will investigate more before implementing it this way, because :

- it's ugly (as I said, it's a workaround, not a feature)
- it's very inconvenient to implement -- but I don't expect you to understand that, unless you've read the corresponding source code

[Laurent]

The accelerometer problem should be solved with the latest commit.