@halirutan Looks like a sign of M10 coming.

@Szabolcs or a server maintenance :-)

I do not see anything new in V10 in Manipulate and controls :( is WRI ignoring this part of Mathematica? No new controls added for long time now.

At least they are better than Matlab. In Matlab, GUIDE (GUI making in Matlab) was not updated since 2004 or so. !

But in V9 there was nothing much new except for some progress control widgets, and in V10 nothing new at all I can see in Manipulate

@Nasser Well they can't update everything at once. And it works well enough

@Nasser we have TouchPosition

@Murta that is nice. But I do not do notebooks nor ipads nor iphones, so do not think I'll ever use this function :)

@acl good point. The Cloud takes more priority these days.

@Nasser reference.wolfram.com/language/ref/IntervalSlider.html ?

@Szabolcs oh yes, I saw this and forgot about it. It is nice. Ok, one new control :)

@Nasser It's not here. No mention of Dynamic stuff there.

@Szabolcs I did not see anything really new in Dynamics/Manipulate/Controls. Resources are going towards Cloud computing. Cloud computing is the future.

@halirutan @bobthechemist It looks like a bunch of WRI marketing folks were competing for "suck up of the year" award...

@rm-rf Btw, on a complete different topic: is it a must to be on linkedin?

@halirutan Depends on whether you want to stay in academia or transition to the industry... I don't think it is a necessity for purely academic tracks, but boy, the industry world LOVES it and if you're not on it, you might as well be invisible.

@rm-rf OK, then I'll go through writing up my profile..

@Nasser DynamicModule has some new features, and now options like SaveDefinitions are documented.

@Murta I see. thanks. But looking at connect DynamicModule variables to another DynamicModule instance makes me so scared now :)

@Nasser I believe that this is a method when you want, for instance, to click inside some DynamicModule and create another that brings varibles seted by the first one.

@Murta I sure hope in V10 they fixed the hanging kernel problem. I can't tell you how many times I suffered from this. Right now, I did something silly, and M hanged again, and I am afraid to shut it down else I lose my changes. Not responding. and no way to click on anything to bring it back.

Now I sit, waiting, hoping it will come back. sometimes I wait for 30 minutes.

@Nasser, I'm a V10 beta tester. Now I can say, because it's public, you will be very happy to use Undo! John has told me that creating Undo, the underlying mechanism for notebook recovery can be implemented, but not in V10.

(@JohnFultz)

@Murta it is not the undo issue. I was all this time not able to use my PC, since M starts eating more and more memory and the whole PC hangs, thrashing, which is what just happened. sometimes I wait for one hr, unable to do anything, since the PC does not respond as M took all the RAM (I have 16 GB). This is the problem.

Mathematica is the only program I know that can do this. Eat all the ram so fast before I am able to do anything, its brings windows to its knees so fast.

If "Wolfram Desktop" only runs in the Cloud, then why is it called "Desktop" ?

@Nasser The cloud runs on a desktop

@Murta All of those worked in v9, right?

(not the Undo related ones)

Hi, someone know how to use remote kernels if I'm at home and inside a router? My router has DMZ feature so it should work. Actually I made it work before but after reinstall the system, the setting in my Mathematica is gone.

This video seems to show Mathematica Online (haven't actually watched it, just noticed a new video): wolfram.com/broadcast/video.php?c=105&v=622

@acl FYI, just finished first cut of the symmetric top spin gyro. Seems to work ok, will add finishing touches on it tomorrow. Used disk as the top. Was not sure what else to use. Based on 3D general Euler equations.

@Nasser now if you really want to bring down Mathematica on Windows with windows, try this:

WARNING: Non-beneficial code to follow. Do not try this in a production enivronment. Save your work. The author is not responsible for any damage or losses of any kind caused by the use or misuse of the following code :D

With some fancy typesetting, this was my entry for the oneliner contest 2012

@Nasser where will you put this? On the demonstrations site?

@Rojo I haven't used SaveDefinitions, but I know it exist in V9. I can't confirm if it was stable.

Does anyone have experience with using a ServiceConnection to Twitter? I'm wondering what (if any) the format of the ServiceExecute command would be to find recent posts with a given hashtag (say #raspberrypi).

Since you're all dying to know: ServiceExecute[conn, "TweetSearch", "Query" -> "#raspberrypi"]

@acl I put all my demos on my page, here 12000.org/my_notes/mma_demos/index.htm it is still beta, will finish it by end of today and clean it up. Might change the spin top to something else, but a disk was easy since its moment of inertia is easy. Also need to check for when it falls and hit the floor, now it does not do this.

Arg-Switz. starts in half an hour- Stay tuned

@belisarius if I have an array a={{1,2,3},{4,5,6},{7,8,9}} is there a function to get {3,6,9} directly instead of doing #[[-1]] & /@ kr

@brama Last/@

@belisarius excellent. Thanks..

@brama or a[[All,3]]

or Transpose[a][[3]]

@acl great!!

@Nasser Unfortunately it doesn't like the OS X UI :-(

@Szabolcs I'll fix that now. This has always been a problem. Mac bottons render different from windows. So Manipulate on window will look and can render different from apple computers. I complained about this before, but I do not think WRI can do anything about it. Will make the buttons bigger now. I do not have a mac to test with.

@Szabolcs I've just uploaded a new CDF 12000.org/my_notes/mma_demos/gyroscope_2/index.htm please see if the buttons now ok. (might want to refresh the page, sometimes it keeps old CDF in there)

but this demo is not completed, but the gyro precession and nutation effect can be clearly seen.

@Nasser Looks good now. Hopefully it doesn't matter whether I check it in the browser or as standaloine CDF. This time I checked in the browser.

@Szabolcs yes, may be I need to make it wider. This is only HTML setting. But button look good now. Thanks.

One day I want to buy a mac, just so I can see how the CDF's look like on it as I have no idea now if any demo I write will look ok or not on a mac, I use windows only. Will be nice if Mathematica will at least has a feature to show how a Manipulate would appear on a mac, may be an online service or such.

@Szabolcs Well, that was ridiculous.

oh well

The wonderful world of open source. Spend hrs building and configuring and downloading different sources that do not work with each others, but it is all open source and free, so it is ok, right?

julia> using Winston
ERROR: TclError("error initializing Tk: version conflict for package \"Tcl\": have 8.5.15, need 8.6")

After 5 hrs compiling and building. I just love open source.

if I have lots of free time, I really would enjoy it, I am sure.

@Nasser aren't there precompiled binaries?

@Nasser here?

@acl I am on Linux mint 17. I used github. From source, this gives you the latest 0.3 version.

@acl build it from git? 0.3?

@Nasser hmm. I had no trouble on OS X

I think so, don't really remember

no it's not 0.3

The fun of open source, is that there are so many choices to choose from. But I really need one that works :)

Now it's more clear http://www.wolfram.com/desktop/

There is a Desktop version.

...formerly known as "Mathematica"? (I'm sorry, I've forgotten the arguments for why this won't be so.)

Is it possible to use use Wolfram Cloud in command line mode, so that I can use it as a remote kernel to my local frontend?

@xslittlegrass No, but you can deploy functions to the cloud and have them evaluated there. However, how do you plan to use this? The cloud interrupts any calculations that take more than a set amount of time.

The cloud interface is good, but coding in it sometime not as convenient as in a local notebook interface. I'm wondering if it's possible to just use it as a remote kernel, so that I can access those new functions in V10 but still use my local V9 frontend.

@Pickett :)

@xslittlegrass ok, no that's not possible :/

@Pickett OK, that's fine, I'll wait for the V10. Thanks for the information :)

I believe that the limit to launch V10 is jul-20.

Based on some presentations, that mentioned that it was a question of weeks..

@Murta I'm wondering how did you get the V10 beta version?

I was in the WTC 2013

This year I'll be there again. It's cool.

Just noticed something strange. In making Graphics3D, when I add Opacity as option to Polygon, it slows down things so much. When I remove the Opacity, it speeds up again.

 {(*Opacity[0.2],*)Polygon[{{-4, -4, 0}, {4, -4, 0}, {4, 4, 0}, {-4, 4, 0}, {-4, -4, 0}}]},

I think "Opacity" does something to the Polygon data. The slow down is very much noticeable. So for now, I removed the Opacity.

@Murta Cool! What's the picture in your avatar? Is that a detector ?

Noo!.. it's a product category sales chart (done in MMA off course:). each level is a category group, and each sector a part of that group. Here is a picture of it community.wolfram.com/groups/-/m/t/86916

The core are products perishable and non-perishable.

Great!

@Murta Have you tried to search for the word "Wolfram" on that single page? 78 hits.

@Nasser It doesn't do anything to polygons. It's just the Opacity that slows down rendering.

(You have to think about the rendering: Where with opaque polygons, you can stop at the first wall you hit, with opacity, you have to follow the light through)

@halirutan The entire site makes me cringe

It's getting more and more hyperbolic. Incredible!

Doesn't anybody with common sense look at it?

@halirutan By the way, I used your intellij plugin today for actual work for the first time.

I just want Mma 10 ...

and I want decent graph/network functionality and responsive graph/network developers

It's neat

I want usable and fast graph-processing functions. Not something that works on paper, something that works for real work.

@acl I don't know. I guess it is bad for users with Mathematica experience to read this, because everything is known for a long time. If I look at this:

I want verifyability. I don't trust any graph related functions any more unless the result is much easier to verify than to compute, so I can verify it myself. I want interfaces to alternatives so I can compare.

I just see the usual Mathematica with OpenCL/CUDA and MathLink/LibraryLink..

Also, I don't want everything to be just Automatic with no option or no documented Method options. Some things just can't be done automatically.

@acl I have to do some other stuff, but I'm almost done implementing the StructureView to be a useful tool. After that, I push a new version on the repository.

For example, I don't think there is a practical method to sample graphs with a given degree seuquence that guarantees uniform sampling with a finite computation time. Yet there's no mention of this on DegreeGraphDistribution and no special options to control the precise method it uses.

@halirutan If you want help beta testing ask

@acl @Szabolcs fyi, just updated the gyro demo, now it checks for crash and stops. Fun part it to try to guess when the top will fall over ;) 12000.org/my_notes/mma_demos/gyroscope_2/index.htm

A couple of months ago there was a WRI presentation at our university. There was almost nothing demonstrated that is actually useful in Mathematica, or that would be used by people using Mathematica for work. Just a lot of showy nonsense features like natural language input and built-in datasets that we don't use anyway.

I think V 10 will come out in october, time for the conference !

@Nasser It'll come out very soon.

@halirutan I guess intellij can't show me the function's signature when I autocomplete? or did I miss it?

@halirutan I took out the OPacity and took out Plot3D, both slow down animation so much! I need to make a list of things that slow down Manipulate so I remember.

@Szabolcs same on the website. look at the examples they give

@acl I haven't implemented it that way because there are almost always many call patterns. What I might consider is implementing a Show Parameters function for built-in stuff.

@halirutan OK, was just wondering if I'd missed it

@acl If you don't mind, then you could de-install your current version again and switch to the latest dev-version which provided StructureView and SurroundWith.

@acl Just download Mathematica-IntelliJ-Plugin.zip from here and after removing the old one, you click on "Install from disk".

@halirutan ah OK didn't know it was on github

@acl hehe, where else should I put it?

People who work on core areas of Mathematica tend to show up on these forums from time to time and give good answers, about stats, probability, symbolic computation, special functions, system-modeler, image processing, dynamic stuff, I think even control systems ...

OK, will try it. let's hope I don't manage to bomb my newly-reorganized code

I think graphs/networks is the only popular area where the dev has never showed up ...

and when one tries to use those functions seriously, there's just so much wrong with them! My colleagues use it more than I do and whenever they try something new, a problem or inconsistency comes up.

BTW it appears to be based at least partially on this library: lemon.cs.elte.hu/trac/lemon Which hasn't been updated much recently. This is based on digging around in the installation folder.

@Szabolcs This is one point I was thinking about. The whole demonstrations website, as nice as it may be, does it really show whether/how Mathematica can be used for real work projects? All the funny presentations and impressing one-liners, do they really have something to do with every day work where you probably have packages of several thousands lines of codes and need a UI with more then just a few controls?

Especially, the building of a responsive, non-trivial UI for a larger projects is really hard. Compare that to click and draw it inside a Java Form-builder or QT, where you only have to fill in the functionality later.

@halirutan @Szabolcs exactly. There's nothing substantial on there.

Sorry about the rant. I was just too annoyed today with Mma.

@Nasser Python works the same way.

sort of like a pointer in C

@halirutan overall I find the demonstrations site to be completely useless for anything real world -- and actually is a negative in terms of giving potential customers the wrong impression of what can be built for them. The exceptions are some useful code fragments by some of the developer contributors. The CDF examples are closer to what real world apps need to be like

@Szabolcs I see. I do not use Python. I find this so confusing and error prone. I will never be able to program in a language like this. I was trying it out. I am not impressed at all with Julia. It matrix operations also are confusing. Matlab is much more natural for that.

@MikeHoneychurch Exactly my point. It's nice to see how some of the more complex things are implemented but other that that it might give the wrong impression.

@Nasser Python does the same

@Nasser I think Java does the same. Though I don't know Java.

@MikeHoneychurch Some demos there are really useful and illustrate useful things. They are not meant to be real world applications. Just to illustrate a point or an idea. I agree that many are also trivial and do not do anything useful.

@halirutan my 2 cents worth is that Manipulates are useful for school homework etc but really have no place for real world apps

@MikeHoneychurch There isn't a single demonstration of a large numerical application on their site.

@MikeHoneychurch They might not be useful for building GUIs that others will use, but it is fantastically useful to quickly analyze some data you've just produced interactively

@acl demos are meant to be small and illustrate single idea or usage. They are not meant to be full-scale actual apps.

for quick and dirty stuff

@MikeHoneychurch Yes, in the academic area, this can be most helpful, but now, Wolfram reaches out for real customers.

@Nasser yes, sure. But they don't have any serious examples anywhere

@Nasser It makes more sense for general purpose programming, I guess ... You always have to make a copy explicitly. Otherwise it just assigns a reference to the very same object.

@acl yes quick and dirty stuff for prototyping and so on. But you never use them for real world apps and customers are underwhelmed

@acl Well, the blog comes closest

@MikeHoneychurch That's what I am saying; it doesn't look robust and flexible enough to produce something to sell. But to explore things quickly, it saves a lot of time

my overall problem with the demonstrations is that while I agree with @nasser about the intended purpose, by limiting to manipulate it has really killed off demonstrating creative interface building

@Szabolcs that is not how Matlab works. It is copy-on-modification. A reference is first made. But if the data  changes, then a copy is made. Here is Matlab

EDU>> A=[1,2]; B=[3,4]; A=B; B(1)=7; A

A =
     3     4

@MikeHoneychurch but forget interface building. compare to matlab's file exchange

It is in my opinion a very useful tool, that's why I use it daily. But it's not shown clearly for what purpose it's most useful. And the uses that are showcased most often are not serious.

what's the equivalent to mathematica? Look, let me put it this way. Apart from the stuff that comes with mma, the largest packages I've seen have been written by myself. They're not even that big.

@acl for me interface building is what it is about for commercial customers.

@MikeHoneychurch OK

Suppose I wanted to show someone what can be done and how easily, what do I show?

anyway

@Nasser but other languages do it differently

@acl of the publicly accessible stuff I'd suggest the CDF examples. It is a shame they haven't added to these over the years. I think there are only about 7 or 8

@acl the key with CDF examples is that they demonstrate that interface layout and appearance is not constrained. So you can layout as you choose, use company logos branding etc

@Nasser I'm not knowledgeable enough on the topic to be able to put together a coherent argument ... Here's a related thread I found: groups.google.com/forum/#!msg/julia-dev/kPtAZOc2sUA/… It seems the Julia developers think that choosing the behaviour you describe makes the performance model too opaque. (I.e. you can't easily predict what's fast and what isn't, without a deep understanding of the language.)

@MikeHoneychurch but that's useless to convince people they can use mathematica for numerics with a few thousand lines of code

I guess that's not their target market anyway. Although we periodically get someone at work to demonstrate eg how to produce fast code, etc

@acl we need to split the "selling" of Mma into two sections. What can it do and how can it be deployed. My comments above address the second. For that it is hard to break into markets because apps like Tableau dominate

@MikeHoneychurch No I think whoever is doing the marketing should ask an unbiassed friend to look at the website and offer an honest opinion

@acl so i have two discussions with people. Firstly about getting the job done (the calculations) and what can be built to make it a nice looking app to be deployed and used by non programmers. Got to tick both boxes to get work.

@Nasser But it's faster. What if I want to deal with matrices which I don't have enough memory to store twice?

@acl one can do copy-on-write, as Matlab does it.

Then performance is unpredictable. Just like mma

ie. A=B, does not make a copy. Only when B is changed.

Fortran is very fast. Ada is very fast. In both, when I do A=B, it is copied.

@Nasser Well, I don't use either.

C++ is very fast. C is very fast. In both, when I do A=B, it is copied.

@Nasser Wait, are you arguing that copying isn't slower? Or that it doesn't make performance harder to predict?

@Nasser Yes. Look. I am just saying, if copying happens as needed, you'll run into situations were it is not easy to predict performance. At least, this is how I understand the argument.

@Nasser Well OK there is a bit more to C than that

@acl before I look at performance, I am interested in language semantics I can understand. fast but confusing language is no good to me. That is all. When I write A=B, I want A to get the value of B, and I do not want A to change if I do not explicitly write A=.... later on.

@Nasser Fine, I see your point. I am just saying that one major problem I have with mathematica is that sometimes I can't work out why it slows down when it does. For example, I am now running some code that's not very large, about 500 lines in a text file. When some parameter is, say, 10, it takes 90 seconds. When it's 1000, it takes half an hour.

Why?

I have no idea

This is a problem I always end up dealing with.

@acl dynamic languages are just slow by nature. Mathematica loves to eat RAM. I do not know why. If you want speed, check out Fortran. I do not know what you are working, if it is numerical or business applications. For science/speed, Fortran is the best.

Look. I can get mathematica to an order of magnitude of C for the stuff I do. The problem isn't speed, it's unpredictability

@Nasser I do large(ish) scale numerics for physics. I've been doing this for some time.

@acl then why are you not using Fortran? modern Fortran has everything you need.

gfortran is free and runs on Linux. I have installed. gfortran now supports many things in 2008 version even.

@Nasser What I have been using in the past for speed is C (and C++ but mostly C).

For what I do now it's not much faster, except when mathematica randomly slows down.