@R.M. @WReach Thanks very much:)

My previous trial is f = # /. x_?AtomQ :> {x, x} &. However, it is not right. f[{1, 1}]==>{List, List}[{1, 1}, {1, 1}]

Did someone know how to deal with bad arguments when a function's UpValues is a pure-function?

The built-in BSplineBasis[] indeed ingnored this problem.

E.g. in THREE.js the 'y' axis is the one that is the vertical one by default. What is the reasoning behind this convention?

@Pickett any idea?

@Kuba No, never encountered that before.

It violates the principle of least astonishment, I might note.

@Pickett stackoverflow.com/questions/8272297/…

@p

@Pickett and all those additional packages, like OrbitControls are assuming this so even if you change the 'z' to be the vertical axis then the mouse rotation goes wrong :D

@Kuba curious... now I too want to know why they did that :P

@Pickett I hope I will find a good explanation why it is so. I will be pissed of otherwise, I hate conventions that are useless but exist only because people are refusing to change :P

@halirutan Hi. You are a graphics guy, do you know anything about that?

Heya @Kuba

@JacobAkkerboom Hi Jacob :)

:) how are you? Seems like you are very active lately in little projects like this svg business :P

@JacobAkkerboom let's say I'm not happy with cloud front end performance and the fact I don;t know what works and when it will work :)

but except that, I'm fine ;)

@Kuba Well that last bit is good to hear :P.

StringCount["blog.wolfram.com/2016/01/19/announcing-wolfram-\
programming-lab Wolfram Blog Announcing Wolfram Programming Lab \
Stephen Wolfram 1453237926 I’m... - 12h ago by Wolfram Blog", \
"wolfram", IgnoreCase -> True]
(*6*)

Still think the WPL looks great though

@JacobAkkerboom the acronym or the product? :P

@Kuba Hehe it's always good to use acronyms to keep the StringCount low I suppose :P. But yeah the product looks good. I like the "explorations"

Gotta go now, cya!

@JacobAkkerboom Have a nice day. ps. maybe I like it too, I have to check, maybe I already know it but they like to rename things :)

@Kuba Do you actually use IGraphM for something or were you just playing with it?

@Szabolcs only here really, don't have many graph related problems at work lately :)

Has anyone run into the following problem before? (@rcollyer)

I am working with MatrixPlots. They are used to monitor an ongoing (LibraryLink) calculation, and good performance is important. By default MatrixPlot downsamples large data before plotting, which is great!

The only problem is that because of this downsampling now the coordinate system of the matrix plot doesn't match up with the standard Graphics coordinates. (And I'm not talking about the reversal of the axis directions, but a complete rescaling.)

I need to plot stuff on top of the MatrixPlot, and the coordinates don't line up ...

The simple solution is MaxPlotPoints -> Infinity but this also slows down things quite a bit :(

Should we allow Wolfram Cloud questions? Even WRI says it's beta, so its far before beta... I just tried to deploy a little bit more complex FormFunction and I already have material for at least 10 questions/bug reports.

@Kuba I think the rule of thumb should be that questions are only about Mathematica programming inside the Mathematica application. And about using the Mathematica application. Not about other Wolfram applications or Languages or products. Those other questions can be asked at community. my 2 cents on the subject.

@Nasser My point is, they shouldn't :) But they are, there was a question about the tagging convention on meta. I mean, what's the point if there are many many cases where one could ask e.g. why Grid's Spacings aren't working in deployed formfunctions etc etc.

@Kuba Is the WPL different from the Wolfram Development Platform (WDP) or the original Wolfram Programming cloud (WPC)? I'm getting confused with all of these acronyms from WRI.

@bobthechemist I suppose not since there are different hyperlinks. But don't ask me about details :) Clear licensing and products overview would be a nice thing to see. But I have no hope to see it unless someone outside will do this.

@Szabolcs Hehe, this is nice.

@R.M. @ShutaoTANG A simple and readable pattern-match method should be this:

f[x_List] := x;
f[x_] := {x, x};

It is equivalent to what RM uses in his anonymous function, but using a regular definition.

@Kuba Well, not really. Everyone can do it like they want so THREE.js is free to use any coordinate system they like :-)

@halirutan of course, but why not be consistent?

@Kuba Maybe I haven't understood it right: Is THREE.js itself inconsistent?

@Kuba Funnily enough, the OP in the Stack Overflow post is not consistent with what they want... 3.js is consistent w/ the right hand rule but their "what I want" is a left hand rule :)

@R.M. that I know, it's worse :P

@halirutan no, it's great, I just don't understand why they build assuming 'y' is vertical.

@halirutan I would like to know where this convention originates

@Kuba Because of OpenGL?

@halirutan I don't know? :) that's why I'm asking

@Kuba Ahh, OK. Yes, OpenGL uses a right hand coordinate system like this

@Kuba Since 3.js seems to be a wrapper library for OpenGL the only reasonable choice is to use the same coordinate system.

@halirutan Hmm, but what I want is right hand side system too, but with the 'z' being the vertical. So, algebra is the same, the wrapper could have been built assuming vertical 'y', isn't it so? I'm not insisting, I know nothing about computational graphics :) I'm just curious

@Kuba But then it wouldn't be a right hand side :-) Because the order you stick your fingers out is important too.

@halirutan take the picture, +x stays, put +y where -z is, put +z where +y is

It is right hand side

@Kuba Yes. Then you simply need to rotate the coordinate system.

@halirutan I know :)

In OpenGL you could just add this rotation matrix to the stack of transformations. Don't know whether this is possible in 3.js easily, but it should.

@Kuba Or, you set your camera differently which might be easier.

@halirutan no, not at all camer.up = (0,0,1) r something

so quite easy :), the point is I'm curious because all examples and additional packages are building planes on x-z plane ;)

for me, coding this would be unintuitive

@Kuba Well, you should rather get used to it since all the core 3d graphics relies on it.

@halirutan good to know ;)

ok, have to go now, thanks, see you later.

Graphics@{Opacity[.005],
  Line@Map[{Sin@#, Cos@#} &, # 6.0000001^Range[40] & /@
     Range[0., 2 \[Pi], \[Pi]/1000], {2}]}

@kirma What a coincidence, i just saw that shape in my cup of tea few days ago, then completely by chance read about nephroids and said to myself I had to MMA it. And now this !? ;)