@Pickett Back in early September I played the default level a few times and got 325 I think. Then my dad got 331. Then I just worked on adding features since then when I had some time. Today the little AI I wrote got 335 on the default level.
After stepping through its move sequence though I doubt I'll require that level of proficiency to unlock the next level or people will get too frustrated.
Actually, the main reason I made a little AI for it is so that I can quickly generate a score distribution for a potential new level. That way I can quickly and consistently calibrate difficulties across different levels based on percentiles in the distribution as opposed asking 10 friends to play it 5 times and then try to guess good thresholds.
Thank you. I'm thinking the theme will be that you are starting a music label. Each level is a city where you go around and listen to different bands at the various venues each night to sign them to your label. When you go back to your store you sell their records and get some money. Then those bands need time to write more songs while you go listen to the other bands.
Is there a good way to know :How to understand the result of ListCorrelate[{{u, v}, {p, q}}, {{a, b, c}, {d, e, f}, {g, h, i}}, {3,1}] ,where $\{K_L,K_R\}={3,1}\$
There'd be a lot of those. I mean one problem in my experience with that is the internal problem appears to be different than the problem initially reported.
A customer might report "hey this integral isn't working" or something similar, but the internal bug fix ends up being something arcane and basically not understandable about some strange internal thing which fixes a number of not obviously connected issues.
Another issue is with things that get improved, but can't understandably ever get fixed. Something like "this function now uses complex branch cuts more intellegently"
It's possible this would only partially fix an issue someone might see as a bug for example.
In the past a group of people would find the major bugs which were straightforward to report and send emails back thanking people for reporting them.
Of course, there are other ways to resolve this: more continuous maintenance release cycle, shorter fix cycle, and active informing of customers on their cases.
I work on a piece of software that has couple hundred developers working on it, and has thousands of man-years put into it, and there are maintenance releases - basically just to fix bugs - every couple months. Information on all fixed - and unfixed - support cases gets back to customers, and basically all of them are either verified or actively reasoned out to be either fixed or not fixed.
Any customer bug actually affecting a sane use case should get fixed in two maintenance releases or so from the current one.
Yeah. When a bug is a result of a complicated fix that might affect other things, you end having to decide whether it's worth it or whether it should wait till that piece of functionality gets an overhaul of some kind too.
I see this alot for example with Import/Export issues. If there's something wrong in a library being used, and you're likely to swap that library out in a year, it might be insane to mess around with complicated exporter for an insane format
Not that I'm thinking of several different formats in particular right now
Sometimes I get the idea I'm talking to a support bot... maybe it's just lots of auto-templating? Occasionally grammar looks like Eliza backreferencing my sentences. :}
I'm amzed looking back that I got through college without learning some of the common gotch-yas of symbolic computing. There wasn't any class for it. Fortunately, I had plently of exposure to numerical computing and so I wasn't too shocked.
Most incoming emails aren't like that. They're written by people who don't work in tech and aren't cognizant of how to report an issue so that the other person will understand it
@Searke I must say I'm waiting for Edison version of... whatever it is. About to get some boards soon, and they will be looking for a use case. Even inventing that I just run command-line Mathematica on them suffices as one. ;)
I haven't gotten to try it yet. I tried doing some stuff with the rasberrypi about a year ago and ended up making some silly toys that didn't do anything interesting
Too bad those Sparkfun daughterboards for Edison are still no-show. Well, pretty much everything is at this moment, apart from the CPU board, which you can't effectively even power without a daughterboard.
@Searke :)
If good installation image and sensible integration to Edison peripherals will be provided by WRI, I could imagine the combination becoming actually interesting tinkering platform.