TNoodle status

[Jeremy Fleischman]

Here's a list of open issues for 2014:

https://github.com/cubing/tnoodle/issues/156 - Chen Shuang's proposed fix is very simple, but I'm not convinced it actually works. I honestly don't understand this stuff all that well, so I've asked Clément to look into this. I've also hijacked this issue to figure out why our automated tests didn't catch this.

https://github.com/cubing/tnoodle/issues/148 - I've been working on this when I can for the last few days. It's almost done, I hope to polish it off tonight.

Sébastien mentioned that 4x4x4 initialization wasn't working for him on 0.8.0-india, which is news to me. It has been working for me.

https://github.com/cubing/tnoodle/issues/153 and https://github.com/cubing/tnoodle/issues/151 are also something I would like to fix for 2014, but I'm willing to let this slide if the WCA is.

ETA-wise, we have a fix for 156, and I'll be done with 148 tonight. Square-1 filtering (153 and 151) are something I need to investigate before can give an ETA.

-Jeremy

[Shelley Chang]

I have a competition that needs skewb scrambles tomorrow. Can someone give me a link to something I can use?

[Jeremy Fleischman]

Sorry for the late reply, Shelley. Please use https://github.com/cubing/tnoodle/releases/tag/v0.8.0-india for skewb. It's also fine to use this version for all other events if you haven't generated scrambles for them yet.

[Sébastien Auroux]

Shouldn't we just send this link around on the delegate list or is the new version really only a few hours away?

[Jeremy Fleischman]

That sounds like a good idea, yeah. The new version is still more than a few hours away, unfortunately. I'll email the delegates list right now.

[Jeremy Fleischman]

Updates:

156 -  Does not block 0.8.0

148 - I finished this last night. It is pretty well tested.

151 & 153 - I have a fix for these that I want to implement tonight. Lucas, do we want to change the sq1 notation/filtering in the regulations to be consistent with the sq1 +2 rule? Even if I change this, I believe there are still issues with Chen Shuang 's optimal sq1 solver that will keep us from being able to filter as aggressively as we want (<20 moves).

I fixed the 444 initialization issues people were seeing. Unfortunately, our tests do not run against the finalized jar,  which is how this bug sneaked out. As of right now, I don't believe it's worth inventing infrastructure to be able to test the finalized product, as I don't anticipate this breaking often, and any breakage should be obvious when testing manually. If I'm wrong about either of these two things, we can look into running automated tests against the jar file.

I believe I've seen some pretty noticeable performance regressions. This is to be expected with sq1, as our new filtering takes around 10 seconds per scramble. Hopefully we can alleviate this by hooking into Chen Shuang 's solver, but I do not want to make that block this release. I also am worried that my changes for BLD reorientation (148) have introduced some performance issues which I'd like to look into.

[Lucas Garron]

The Regulations do not specify a Square-1 metric, so that's not a problem until we define it.

The Regulations currently have a scramble limit of 11, so that's what it will be until there's a discussion about moving it to 20 properly.

With 156, what do you mean with "does not block"? The scramble state must be random (with some filtering), and that includes some 11-move scrambles.

»Lucas Garron

--
You received this message because you are subscribed to the Google Groups "WCA Scrambler Team" group.
To unsubscribe from this group and stop receiving emails from it, send an email to wca-scramble...@googlegroups.com.
For more options, visit https://groups.google.com/groups/opt_out.

[Lucas Garron]

Just checked out jfly/master. Looks good for 333bf/444bf/555bf/333mbf.

(5BLD has width 4 instead of 3, but that's fine.)

How ready is this? I know we prefer Monday, but Tuesday/early Wednesday is still okay.

If a Delegate doesn't have time to generate new scrambles, that can be forgiven, but I'd like to have new competitions running on this as soon as possible.

»Lucas Garron

[ILKYOO CHOI]

If this is thoroughly checked to be ready, then I prefer have an exception and release it today or tomorrow so that competitions this weekend can use this scrambler. 

It seems like there will be some exceptions regarding some release dates for WCA committees.

-- ILKYOO

[Jeremy Fleischman]

Ok,  I thought the use of the word "move" in https://www.worldcubeassociation.org/regulations/#12c implied a metric. That's why I created https://github.com/cubing/wca-documents/issues/130 in the first place. At the time I created it, Lucas confirmed that I was interpreting the definition of "move" correctly. Either way, their are 2 issues here. One is what metric we want tnoodle and the WCA regs to agree on. I want to change tnoodle's metric to agree with Chen Shuang scrambler. The other question is what degree of filtering to do with do with the chosen metric.

I fixed the bug 156 was originally created for (https://github.com/cubing/tnoodle/issues/156#ref-commit-809b362). 156 is still open is because I want to know why our tests didn't catch it (I suspect that our test would have caught it, but they didn't get enough runs), but this does not block releasing 0.8.0.

Right now, I'm pretty happy with everything, but it is barely tested. I'm going to submit a massive pull request tonight, and poke people to look at the relevant pieces of it.

Should 5bld scrambles end with turns like 3Rw instead of 4Rw? I promise that that's very easy to change if we want to change it.

[Sébastien Auroux]

Hi Jeremy and others,

I prefer to release on Monday and to use the intermediate time for testing. I'm very surpised that the BLD orientation causes performance regressions, but improving this is not on top of the list right now I guess.

Yes, please change the 5BLD scrambles, if that's easy to do. "4Rw" would be very unfamiliar for most people I think.

2014/1/8 Jeremy Fleischman <jeremyfl...@gmail.com>

[Jeremy Fleischman]

Thinner reorientation is done here https://github.com/jfly/tnoodle/commit/963efa7dd54da2a2606ec3a735c4969fbb426e70.

[Jeremy Fleischman]

For 0.8.0, all I need is for Lucas to tell me what needs to be done for 444 bld.

I've also been in discussion with James about possibly changing TNoodle's json format to make it easier to upload scrambles to the wca website. If I decide this is necessary, I would like for those changes to go into 0.8.0. I won't be able to look into that until tomorrow (Saturday) morning.

[Jeremy Fleischman]

I've finished up everything I want to do for 0.8.0. It's lightly tested. Lucas, could you take a quick look and accept https://github.com/cubing/tnoodle/pull/161 if you like it? Feel free to bump the version number to 0.8.0 and do a release if you can. Otherwise, I'll do it when I wake up tomorrow, if you're ok with the changes.

[Lucas Garron]

I'd rather not do the release myself.

0.8.0 has been added to the allowable list in the API, though: https://www.worldcubeassociation.org/api/v0/scramble-program

»Lucas Garron