[robocup-small] New Referee Box Protocol

[Christopher Head]

-----BEGIN PGP SIGNED MESSAGE-----
Hash: RIPEMD160

By this, do you mean we should have one button/command for goal kick
and a different one for corner kick, instead of combining them both as
Direct Free the way we do now (with old protocol and with the proposed
new protocol)? Is this something that would be useful for AI? The ball
position is quite different between the two cases; I would think the AI
could figure it out very easily based on that (besides which the rules
are exactly the same for the two cases, so really, the AI doesn’t need
to know the difference). We can separate them if people have a definite
reason, but this is something I really hesitate to do because it means
we need to add more buttons and more keyboard shortcuts for the
operator—and right now, we’ve already pretty much run out of space on
the numpad!

As for time, I agree it isn’t something we need to spend a lot of…
time… talking about :) It’s one line of code for teams to convert
whatever we send into whatever else they want, by multiplying or
dividing by some appropriate power of ten. Personally the proposal I
like best so far is integer microseconds. Way more resolution than we
need (we’ll never achieve anywhere near microsecond resolution with
NTP), but since 2^64 µs = half a million years, *also* way more range
than we need. Milliseconds is an uncomfortably low resolution, and
nanoseconds is an uncomfortably low range (58 years), so microseconds
is a nice tradeoff IMO.

Chris

On Fri, 9 Nov 2012 22:52:59 -0200
"Angelo Gurzoni Jr" <jgurzoni at yahoo.com.br> wrote:

> Hello there
> I always get amazed to see how many bright people we have around. We
> need to discuss more, that's great !
> 
> I'd say the time format is not critical, so we may want to make sure
> we don't forget the rest of the protocol going too deep about the
> time format  discussion. For instance, how about having a command and
> button for corner/goal kick ?
> 
> About the time, though, I have the opinion that we don't need
> "much" (I know, this is very relative) precision, and the level we
> get from the SSL-Vision format versus its code complexity is a good
> trade-off. So why not to just use the same, to make it easy ? if it
> was to change, then I'd say to change both together. The NTP/ refbox
> having NTPD idea is great. Let's not forget, though, we most likely
> won't be able to get more than the millisecond order of sync with it,
> making anything more precise hard to attain
> 
> 
> 
> Cheers
-----BEGIN PGP SIGNATURE-----
Version: GnuPG v2.0.19 (GNU/Linux)

iEYEAREDAAYFAlCeHC0ACgkQXUF6hOTGP7d5fwCeLqX+xigULoaK/V+aBS9CPh1x
bXsAnj83fHy0UeuoPDzXWeHeWfdMmECT
=fiDY
-----END PGP SIGNATURE-----