Wolfram Mathematica

Q: Dynamic's SynchronousUpdating->False broken in Graphics

9:47 AM

#dailyFun

When Dynamic content is bigger, the updating is broken: data = RandomReal[{-1, 1}, {10^4, 2}]; show = False; Checkbox[Dynamic@show] Dynamic[ If[ TrueQ@show, Length[data], "not shown"], SynchronousUpdating -> False ] Graphics[{ Dynamic[ If[TrueQ@show, Line[data], {}], Synchr...

3 hours later…

12:30 PM

@Searke and others. Is anyone aware if FrontEnd is going to be able perform simple operations on associations as it can on lists atm?

DynamicModule[{x = {0}}, Table[Slider[Dynamic[x[[1]]]], 500]]
DynamicModule[{x = <|1 -> 0|>}, Table[Slider[Dynamic[x[1]]], 500]]

in the second example it has to call Kernel which slows things down.

Q: Problem with CurrentValue["MouseOver"] when graphics are wrapped with Deploy

1:12 PM

I'm pissed off. I reported that bug over two years ago and it is still there. And it affects every f**** Dialog notebook and every other part of front end which is Deployed->True.

Bug reproduced in V9.0.1 and persisting through V11.0.0 Please take a look at following toy example: Graphics[{ Dynamic[{ If[CurrentValue["MouseOver"], Red, Blue], Thickness@.05, Line[{{0, 0}, {.5, 0}}]}], Dynamic[{ If[CurrentValue["MouseOver"], Red, Blue], Disk[{.5, 0}, .2]}] }, Plo...

graphics dynamic bugs

data1 = RandomChoice[DateRange[{2010, 1, 1}, {2010, 5, 31}, "Day"],
   2000];
data2 = RandomChoice[DateRange[{2010, 3, 1}, {2010, 8, 31}, "Day"],
   2000];
CreateDocument[
   DateHistogram[{data1, data2}],
   StyleDefinitions -> #,

   WindowSize -> All
   ] & /@ {"Dialog.nb", "Default.nb"}

In dialog notebook Tooltip is triggered even by rectangles which are behind...

Every Dynamic/FE tutorial should start with "reconsider javascript if you value your time".

Presure released, going back to work :)

1:26 PM

@Kuba Have you considered javascript as alternative? Just saying... :)

Am I the only one who did not know that this is possible?

@halirutan at least 18 folks know that: mathematica.stackexchange.com/a/121173/5478

@halirutan yes I did and I am but I invested to much to just go away :)

2 hours later…

3:18 PM

@halirutan I've wanted something like that for a while. Does it work with Module too? I'm at dentist so can't check for an hour or so.

Michael E2

@Kuba @halirutan Also here: mathematica.stackexchange.com/a/54874/4999

3:38 PM

@MichaelHale We already had posts that emulated this behavior.

@RunnyKine

@Sascha can one set number of neighbors to infinity?

3:53 PM

My implementation of the whole thing is rather quick and dirty because I needed it only for visualization purposes. I don't know how performance will be if you deal with either lots of points or neighbors. The number of neighbors in my version has to be uneven though (to avoid having a tie between both classes, since I only count individual number of neighbors in the respective classes )

@Sascha, I am not sure why the two clusters don't merge as one when you increase the threshold distance. Surely, a point on the blue cluster is closer to a point on the green cluster as you increase than threshold.

I probably didn't get the intend of your question in the first place. Do you want to seperate data into a (given) number of clusters or do you have some existing data clusters and now want to classify new data entries not in the original clusters?

@Sascha, I have a set of points generated from some Simulation. I want to put points in a cluster if that point is not more than a certain distance from one point in the cluster. So assume you start out with a random set of points, you pick one point to start with and then join other points to it using the condition, until no other point meets that condition, then that is a cluster. And you continue with another point not in the cluster you just found etc.

I think I understand now:) I will look into this and see if I can come up with an answer

@Sascha. Thanks

2 hours later…

5:44 PM

@halirutan Ah, it doesn't work with Module. I shouldn't start adding a bunch of utility functions to my init.m or such because then if I send a notebook to a teammate it might not work. I'd like something like Get["Michael Hale's utilities"] that accesses the latest version of a web file, with an optional date parameter, which takes the latest published version of the file prior to that date. Then I can add helper functions but still just send the plain notebook to people.

Maybe CloudGet

Edmund

5:56 PM

Anyone joining the New in 11 webinar?

@MichaelHale We use exactly this web-fetching for the SE Uploader, you know?

@halirutan I do not know. I guess if I get them to all add the loading helpers I want with a single CloudGet call in their init.m files then that is just a one time change they would have to make.

@MichaelHale It's really pretty easy. When I share code like this and you evaluate it:

Import["http://goo.gl/NaH6rM"]["http://i.stack.imgur.com/j841k.png"]

Then the first Import["http://goo.gl/NaH6rM"] does nothing more than loading the latest version of the github package for decoding. Just look up the URL and you find

raw.githubusercontent.com/halirutan/Mathematica-SE-Tools/master/…

This package is loaded with the Import statement and it returns as last expression a function for decoding. Let's call this last function decode. What we have then is nothing more then

decode["http://i.stack.imgur.com/j841k.png"]

which decodes the png image into an expression.

@halirutan Ah yes, I've seen you share snippets like that. Neither of those are easy to memorize though for me to quickly type at the top of a new file. I want something like your first Import permanently in mine and their init.m. But then I want it to enable loading with easy names and dates. So I can just say Get["Michael Hale's utilities"] at the top of a new file. Then before I send it if I'm concerned about future breaking changes I could just change it to

Get["Michael Hale's utilities",{2016,8,30}]

@MichaelHale Forcing people to install something locally is always hard because it's inconvenient.

(for them)

6:10 PM

@halirutan Well in this situation it's their job and it's only a one time thing, so it might work. I could probably memorize a one time short URL though for a system like yours.

@MichaelHale For the code-sharing here it is really easy, because no one has to remember any URL. You just share and they copy&paste it.

If someone would need to load my package, an URL no matter how short would be worse than installing a package locally.

@halirutan No I think your system is great for copying and pasting snippets here in chat. On my team others haven't written Wolfram code yet though. The biggest situation was I made a tool to help one expand a training data set and auto-retrain as they went through a bunch of data. Then when the quality seemed good enough they could automatically process the remainder with the model. So they used my notebook but didn't edit it.

But now I'm just back to generating non-program results or making CDFs for UI prototypes. Not clear what version of sharing workflow will become a more pressing issue in the future.

andre

6:38 PM

spherical buttocks ? anyone interested ? It's up to you !

JasonB

7:25 PM

@andre Well, um, yeah, but that seems kind of a personal question

Pickett

@MichaelHale Tylio's init file has an update command. If you had that, then they could install it once, and update it regularly by just typing updateInitFile[], or you can write it into your file.