Wolfram Mathematica

acl

12:00 AM

@hwlau Well I use compiled functions in parallel code, but not parallelization in compile

(I wonder why our argentinian friends, @Rojo, @belisarius et al, are so quiet)

lossing something?

possibly crying?

Behave people

aha

@Rojo have you use the parallelization in compile?

Liam William

1:22 AM

Does anyone own the MechanicalSystems
plugin http://store.wolfram.com/view/app/mechsystems/ here?

I'm considering buying it because really would like to simple extend my knowledge of Mathematica.

rm -rf

@LiamWilliam I think you can try contacting them for a trial (even though there isn't a trial link available from the store)

Liam William

1:44 AM

@rm-rf Contacting them now. I'm Hopeful I can find some decent API examples at a minimum.

Michael Hale

2:22 AM

MapAt works on Association parts. MapAt[DateList, data, {All,"timestamp"}]. Nice.

halirutan

3:00 AM

@MichaelHale And Cases can now be used as operator:

op = Cases[_Symbol];
op[x + y + z]

Michael Hale

Yes, I saw quite a few functions now have operator forms. A lot of nice improvements to reduce the amount of #,& syntax when doing maps, etc.

And DateListPlot handled the Association fine. DateListPlot[revisions[[All, {"timestamp", "size"}]], Joined -> True]

Michael Hale

3:43 AM

The Ernest Hemingway time lapse was requested. 130 of 740 stable past versions of the article downloaded at 9000 pixel resolution. Using Run calls to phantomjs for the page rendering if anyone was curious. Off to bed.

Taliesin Beynon

7:04 AM

@Mr.Wizard

Mr.Wizard

Hello @Taliesin :-)

Taliesin Beynon

Hi Mr.Wizard!

You had some ideas for syntax?

Mr.Wizard

I don't know how much time you have but I'll try to be brief regardless.

The oldest modification I've been using is to Fold and FoldList, to work on a two-argument form. Unlike some other changing this is a five-minute implementation in top level code; documentation would be the slowest part I believe.

I am using this:

Unprotect[Fold, FoldList]

Fold[f_, h_[a_, b__]] := Fold[f, Unevaluated@a, h@b]
FoldList[f_, h_[a_, b__]] := FoldList[f, Unevaluated@a, h@b]

SyntaxInformation[Fold] = {"ArgumentsPattern" -> {_, _, _.}};
SyntaxInformation[FoldList] = {"ArgumentsPattern" -> {_, _., {__}}};

Protect[Fold, FoldList]

I'd like to know if a two-argument operator form is planned that would block this idea, and if not if there is something else to stop it being generally implemented.

Taliesin Beynon

we can't add an operator form

because your modification is already the current behavior!

it just isn't documented

so you got your way already :)

and probably it has been like that since the beginning

Mr.Wizard

lol -- really?

Taliesin Beynon

7:09 AM

and we just didn't document it

In[1]:= Fold[f, {1, 2, 3, 4}]

Out[1]= f[f[f[1, 2], 3], 4]

2

Mr.Wizard

No, it was not like that from the beginning!

Taliesin Beynon

haha yes

oh, ok

did you try on an old version?

Mr.Wizard

I had Daniel Lichtblau's ear about this years ago and he said he'd see what he could do. Darned if it wasn't implemented without me knowing. :-)

Well, I'm 99.9% sure it threw a syntax error in version 7, but I'll check again now.

@Taliesin Sorry, I was interrupted.

Yes, it issues: Fold::argrx: Fold called with 2 arguments; 3 arguments are expected. >> and returns unevaluated.

Could you do me the personal favor of finding who implemented this? If it was Daniel I need to thank him directly.

Taliesin Beynon

checking

have to check out kernel sources, may take a little while

but it makes sense that Daniel Lichtblau did that

I asked for some new behavior from MapAt and he added it for me, though didn't document it

MapAt[f, expr, {All, 1}] works, for example

Mr.Wizard

Span too, I take it?

Taliesin Beynon

7:24 AM

yes

it's even nicer with operator forms

MapAt[f, {All, 1}] @ expr

Mr.Wizard

Thanks for asking, and thanks to Daniel for implementing. Great stuff!

Taliesin Beynon

it's almost as good as MapColumn[f, 1] @ expr

yeah

actually

you may be interested in this, i'm not sure

so we have level specs, right

Mr.Wizard

(Surely you jest; of course I'm interested!)

Taliesin Beynon

heh

so, level specs are an adequate way of referring to specific 'dimensions' of nested lists

in some sensee

for tensors, they are fine

but take something like that output of Tally

{{"A", 1}, ...}

if we think of this as a "data tensor"

the real problem here is that there is 'index dependence'

e.g.

{{{1,2,3}, 1}, {{3,2,2}, 2}, {{4,3,3}, 2}, ...}

that's a sensible, structured kind of data

but not one that can be described by Dimensions

or adequately 'selected within' by level specs

because there are really two sub-dimensions there

a level requires a 'path' rather than a single integer to specify it

{All, 1} has dimensinos {n, 3}

level {All, 2} has dimensions n

there is the level {All, 1} and the level {All, 2}

we can think therefore of a level spec as really being a shorthand for {All, All, All, ...}, n times

but as soon as we have non-homogenous types, things like tuples or structs (e.g. "Age" -> .., "Sex" -> ..., "Friends" -> {...}), levels are quite useless

because they don't distinguish HOW you get to a certain sub-expression, only how deep it is

anyway, so the idea i'm really describing here is that 'levels' should be generalized in some way to be 'paths' that describe a family of sub expressions

and in fact that undocumented behavior of MapAt is already approximating that

but we could do it anywhere

any the type system inside Dataset already uses such a concept... where some instances of List and Association are 'homogenous' and other are 'heteregenous'

perhaps this is all a little bit too much of a brain dump, and i am a little tired... it is 3:30 am after all.

Mr.Wizard

I was with you about two thirds of the way but I'm not certain I got the rest. I nevertheless am quite interested in where this is going. But it's only 12:30AM here -- please, get some sleep if it's not too late already!

Taliesin Beynon

7:38 AM

fair enough.. maybe i should just write a blog post about it

2

i really like for example the way that Niko Matsakis is doing this for the Rust programming language

explaining his thinking about the type system

anyway, it turns out that the developer responsible is Oliver Ruebenkoenig

who i had not heard of before now

Mr.Wizard

For both Fold and MapAt?

Taliesin Beynon

Fold, I think it was Dan who did my suggestion for MapAt

anyway, i see you just advertised this feature!

you should mention that the change actually happened in February of 2011.

So this is probably 8 and on.

Mr.Wizard

Very interesting; thanks. Re: Fold, I'll do that.

Oliver Ruebenkoenig contributed quite a lot to MathGroup and our own site as well.

@Taliesin Are you heading for bed yet?

Mr.Wizard

7:58 AM

Okay, I'm out of here. Chat later. :-)

Taliesin Beynon

8:12 AM

goodnight

m_goldberg

8:32 AM

@Mr.Wizard. Did you deduce the answer you made to my question involving Condition from your vast Mathematica experience, or is there an third party reference you can point me to?

acl

10:18 AM

@rojo, @rm-rf, @Szabolcs take a look at point 4 here

Oleksandr R.

12:17 PM

I find Undo doesn't work (what I understand to be) properly sometimes. It undoes the last action only as it always used to, and redo doesn't work. It does successfully undo the corresponding evaluation though. Also, documentation bug: says "shift-enter" evaluation can be undone but should be enter OR shift-return

Anyone else experiencing this? Seems to be intermittent. Or are there conditions under which it should not work that I don't appreciate?

Oleksandr R.

12:34 PM

Another moan about V10... Counts vs. Count, KeyUnion, KeyIntersection... is it really necessary to introduce new functions for the new data type? Packed arrays were implemented without duplicating all the functions applicable to them...

Murta

@OleksandrR. I have a nice experience with Undo. In what OS you are?

Oleksandr R.

@Murta Windows 7. But I also have the problem whereby the front end freezes for extended periods with 100% CPU usage so it could be related to that

Murta

Ok.. I'm on Mac

can't say anything

Nasser

1:04 PM

@OleksandrR. so far undo works well here also. windows 7.

I used it 2-3 times and no problems.

I have been trying javascript and webGL for 3D animation, and its performance in the browser is so much better than CDF dynamics with 3D animation.

acl

@OleksandrR. do you mean that if, after evaluating x=5, you undo, it undoes the evaluation and unbinds x?

@OleksandrR. for this, maybe it's related to undo history length?

Nasser

This ode used to give "Solve::ifun : "Inverse functions are being used by Solveso some solutions may not be found; use Reduce for
complete solution information" right away in V9 (which is OK), but now it hangs in V10. Here it is
Remove["Global`*"]
ode22 = y'[x] - y[x]^2 - y[x]*Sin[2*x] - Cos[2*x] == 0
DSolve[%, y[x], x]

Oleksandr R.

@acl I mean that if you write x = 5; a = x, evaluate it, and then delete a = x, pressing undo will just restore it and then both undo and redo are nonfunctional. This doesn't happen all the time and seems more common in the doc pages.

belisarius

1:44 PM

@acl Cheers :)

mfvonh

Does anyone happen to know the status of Wolfram Data Drop? CreateDatabin[] fails for me because datadrop.wolframcloud.com can't be resolved. But I can find practically zero information online.

brama

@OleksandrR. Will you be around for few more minutes? I need to step out ,but will be back. I need to talk to you about ExperimentalOptimizeExpression

Rojo

@acl Nice

Rojo

2:02 PM

I like that those are high priority problems, and that they seem to aim for more small updates

Rojo

2:12 PM

@TaliesinBeynon MapAt working with All, Span, etc, turned a function I used once a lifetime into one of the main ones. Thanks to all involved

Someone with a no v10 version installed. What does this return? 1[[All]]

belisarius

2:43 PM

@Rojo Is it true that Mma 10 is slower than Messi?

Rojo

@belisarius You're of the Messi-pecho-frio gang?

acl

@Rojo Yes that is much better than waiting for slower, bigger updates

rm -rf

@Rojo Is that red border from all the crying?

belisarius

@Rojo Oh, no! I'm aligned with the "Where was Tevez?" faction

acl

assuming I get them on the home license (last time I didn't, although I also have an institutional one and got them there)

Rojo

2:51 PM

@belisarius He was all over the streets, posing for a polenta ad. "Alentá como los mejores"

@rm-rf Just frog blood. That's how I let off steam

acl

@Rojo Integer

Rojo

@acl Thanks. So it has changed

acl

@Rojo did it not work with All before?

MapAt[f, {{a, b, c}, {1, 2, 3}}, {All, 2}] does work on v9

belisarius

@Rojo At least I'm grateful it wasn't a 7-0

acl

as does MapAt[f, {{a, b, c}, {1, 2, 3}, {1, 2, 3}}, {1 ;; 2, 2}]

Rojo

2:55 PM

@belisarius It was a very nice game against a real good team.

and we can whine about that penalty to higuain :)

9gag.com/gag/aEwrp79?ref=fbp

And that was falta de Higuain, haha

Jacob Akkerboom

@OleksandrR. I'm not entirely sure if I understand you correctly, but the behaviour here is different. If I set x = 5; a =x; evaluate it and then delete a = x, I can undo the delete and redo it, and I keep undoing other things. Of course undo does not change the state of the kernel.

belisarius

@Rojo My only thought is that I don't enjoy that style. I know it's plain Sabella style (like in Estudiantes, winner of the Libertadores) but it's too risky and boring

Rojo

@belisarius Who would you like as the new DT for the next 4 years?

belisarius

@Rojo Pekerman ;)

Rojo

@belisarius That would be ok. Let's hope the tsar of our football singlehandedly chooses him

belisarius

3:02 PM

@Rojo Nah, estan mas peleados que la mierda

Pekeran is too honest

Rojo

@belisarius Ah, i didn't know. At least we know it won't be el diego again

belisarius

@Rojo Of course not. He's mufa.

Rojo

@belisarius Plus, at the rate he is speaking right now, he won't be able to finish a sentence in the 15 minute half time

Ok, it's time to get R to work

I am counting on all of you :)

Rojo

3:22 PM

I am going to explain the situation out loud. Why? Because I can

I am on Linux, Fedora 20

belisarius

@Rojo Bolivar died from an explosive diarrhea and San Martin crossed the Andes in a stretcher. Nobody remember them for that

Rojo

InstallR gives InstallR::fail: Failed to install R. The following error was encountered: Unable to load dynamic libraries >>

@belisarius Because they didn't have youtube back then

After running it (and ignoring messages), running something like RSet["a", 8] gives serevarl JLink errors

This question (mathematica.stackexchange.com/questions/15398/…) shows the errors I get.

The question focuses on Ubuntu, I run Fedora. The accepted solution is the following, but to me it doesn't change a thing

InstallR["RHomeLocation"->"/usr/lib64/R"] gives the same message and fails the same way

Leonid's comment (creator of RLink) to this was " So, for at least some Linux flavors, RLink does work with an external R distribution. Thanks for sharing this.". So, it shouldn't be such a surprise that it doesn't work for me. b.gatessucks, who proposed the solution, was running Fedora 17. This works for him, and also ended up working for the OP's Ubuntu

From a terminal, the command R opens R. Directory /usr/lib64/R has stuff, while /usr/lib/R doesn't exist. So the path seems to be ok

I will now look over the other answers to that question

belisarius

@Rojo What are you planning to do with R?

Rojo

@belisarius I need to run some robust fitting script repetitively to quite a bit of data

The script fell from the sky written in R and using R libraries

belisarius

@Rojo I see

Rojo

3:30 PM

Ok. I have more than one issue with my R. Watching the log (RLink`Developer`getLog[]) shows that the platform is recognizes as Windows and not Linux.

I first noticed these issues in the beta. As soon as I installed proper v10 I tested, and everything still failed the same way, but at least the platform was recognized as Linux properly.

@Leonid said that this "wrong platform" error was peculiar, and that R got that platform from $System (I think), but that points to Linux.

My installation has a peculiarity: I run it with a custom user base directory location, that points to my windows installation. I will now open MMA without that custom UBDir and see what happens with the log

Ok. Without the custom userbase dir, the platform is recognized properly, even though all the previously described behaviour remains. From now on, I'll keep it this way, and deal with the custom UBDir once this works.

This pastebin.com/tenuxXGF is the current log

It seems likely that it's a matter of missing/conflicting libraries. I will now start messing up my system by downloading /slinking stuff around, starting with what they suggest in other answers of the same question

acl

3:49 PM

@Rojo make sure to replace glibc with another version (don't keep backups)

Rojo

@acl I am not even sure how to check the glibc version (I think 2.18), nor which I should install, or how to remove the others if they are being kept. :S

(It should be up to date with the fedora repo)

Ok, news

After installing compat-libf2c-34

the builtin R seems to WOOOOOOOOOOORK

For some reason, the getLog[] seems unchanged. Still FATAL errors even though it works.

The custom R installation doesn't show any errors, but when I do RSet["a", 4], I get

RSet::puterr: Error putting the expression 4 into a variable a in R. The last error message issued by R was Error in sapply(dots, function(x) is.symbol(x) || is.character(x)) :
7 arguments passed to .Internal(identical) which requires 6 >>

I am not sure if I need a custom one or not (I think I do but not sure), so for now I'll leave it. Back to the custom UBDir issue :)

Rojo

4:18 PM

Do you guys have an ApplicationData folder in your $UserBaseDirectory?

Oleksandr R.

@JacobAkkerboom yes, that's as it should be (AFAIK). But for me it sometimes (not for this example) get stuck and both undo and redo stop working after a single undo.

Rojo

After installing that library, it works even when RLink (the log) thinks I am on Windows.

Jacob Akkerboom

@OleksandrR. So far it has been working pretty well for me. I've seen it fail once or twice when copy/pasting expressions that were valid Cells. But I couldn't reproduce it.

Rojo

I seem to have an ApplicationData subdir in my UBDir on Windows with several subdirs (MATLink and RLink among them). The RLInk one has a RLink.log file with window paths in them. I am assuming it is to blame for the RLInk mistaking the paths. But I can't find the Linux version of that dir

Oleksandr R.

@brama I'm here

Rojo

4:30 PM

Does $EmbedCodeEnvironments or EmbedCode work for any of you? Not here

$EmbedCodeEnvironments returns unevaluated. And, for example, EmbedCode[APIFunction[{"x" -> "Integer"}, #x^2 &], "Java"], too

(taken from the docs)

Jacob Akkerboom

@Rojo I get:

{C#\:f3b5,C++-VisualStudio,HTML,Java,Java-Jersey,JavaScript,
PHP,Python,VisualBasic}

Rojo

@JacobAkkerboom Thanks. Then this is part of my personal mess :S

Jacob Akkerboom

@Rojo we're all messi here

Rojo

@JacobAkkerboom :P

@JacobAkkerboom Whats your OS?

Jacob Akkerboom

OSX

mavericks

Rojo

4:37 PM

Thanks

If any Linuxerck can try $EmbedCodeEnvironments

Oleksandr R.

@Rojo works on Windows as well

Rojo

@OleksandrR. Thanks

Leonid Shifrin

5:27 PM

@Rojo It is quite likely that I am switching now to fixing all these issues for RLink. @Rojo, the dynamic library issue should be independent of the location of $UserBaseDirectory. Are you on Fedora-based or Debian-based Linux? Could you also execute RLink`Developer`getLog[] and send the output to my email on gmail?

@Rojo and also, what is your current status? I take is that you've got built-in R to work with RLink, but the external R doesn't? The log file keeps the info about previous runs of InstallR[], so you have to explicitly call RLink`Developer`clearLog[] to remove the current log file. Then, the next time you run InstallR[], it will start fresh.

Rojo

Hello! Fedora. The library issue seems to be solved. The location of stuff depends on the user base dir location. The built-in R works, but not the external one. The InstallR with the external one works, but it issues an error when trying to use it. RSet["a", 4"] gives RSet::puterr: Error putting the expression 4 into a variable a in R. The last error message issued by R was Error in sapply(dots, function(x) is.symbol(x) || is.character(x)) :
7 arguments passed to .Internal(identical) which requires 6 >>

I am about to send you (gmail) the log

Leonid Shifrin

@Rojo Yes, please do.

@Rojo Try evaluating REvaluate["R.Version()"] ,and tell me what the result is

Rojo

@LeonidShifrin Wait. The clearLog[] thing seem to really have changed the story. Now the logs show the proper platform even if the $UserBaseDir is the windows one

Probably it was reading that old log file from windows inside the ApplicationData/RLink dir

The log is clean and pretty now

Leonid Shifrin

@Rojo The thing is, log appends new info on every run, so you probably only looked at the start showing always those failed attempts, while the new info was added at the end and you probably didn't see it.

Rojo

@LeonidShifrin That's likely. It was very long, and now it's short

Leonid Shifrin

5:38 PM

@Rojo I intentionally made it that way, so that one can track the history of several runs.

Rojo

@LeonidShifrin R.Version, from the custom one, gives

RObject[{{"x86_64-unknown-linux-gnu"}, {"x86_64"}, {"linux-gnu"}, \
{"x86_64, linux-gnu"}, {""}, {"2"}, {"14.0"}, {"2011"}, {"10"}, \
{"31"}, {"57496"}, {"R"}, {"R version 2.14.0 (2011-10-31)"}},
RAttributes[
"names" :> {"platform", "arch", "os", "system", "status", "major",
"minor", "year", "month", "day", "svn rev", "language",
"version.string"}]]

Leonid Shifrin

@Rojo By custom one, you mean the built-in one, or the external one?

Rojo

@LeonidShifrin I mean external, RInstall["RHomeLocation"->"/usr/lib64/R"]

I mean I ran REvaluate["R.Version()"] with the same R that issues the RSet error

Leonid Shifrin

@Rojo But that means, that it works, at least you can execute R commands via RLink.

Rojo

@LeonidShifrin Yes

It works

:) everything is way better now

after installing that library

Leonid Shifrin

5:40 PM

@Rojo Can you run this RSet["a",4] again, and tell me the full exact error message you get back?

Rojo

RSet::puterr: Error putting the expression 4 into a variable a in R. The last error message issued by R was Error in sapply(dots, function(x) is.symbol(x) || is.character(x)) :
7 arguments passed to .Internal(identical) which requires 6 >>

Leonid Shifrin

@Rojo I am actually rather surprised about this library problem, on Fedora-based Linuxes it used to be smooth.

Rojo

@LeonidShifrin compat-libf2c-34 is what I installed. Got lucky

@LeonidShifrin But don't get too worked up. I still don't even konw what I need from an external installation. I don't know much R

and that "R.Version()" shows stuff from 2011, which is weird

When in a command line I run R. and then R.Version, dates are of march 2014.

Leonid Shifrin

@Rojo This is quite strange. It looks like the error is somewhere on the R side. Never saw this before. What is the exact Linux version you have? Perhaps, you could send me as much info about your setup as you can (Linux version, etc). I will then try to reproduce this on some virtual box with the same Linux.

Rojo

@LeonidShifrin Sure. I am a little reluctant because there might be stupid issues from my side that I can still work on, but no problem.

Leonid Shifrin

5:45 PM

@Rojo Then it can't be R2.14.0. I suspect that you've got some weird mixture of both built-in R and external one (likely 3.x) at work here, perhaps some libs were loaded from one and some from the other.

@Rojo Don't worry about that. I will anyway have to go through various Linux distros and test things, so having this information is important for me too. I might be able to spend some time to reconstruct your case and see what't happening. Unfortunately, my current logging scheme is not informative enough to give all the needed hints to what's up there.

Rojo

@LeonidShifrin Yes. When in the command line I run R, R.Version gives version 3. But not from MMA after InstallR["RHomeLocation"->"/usr/lib64/R"]

When, from a command line, I run the R executable from inside /usr/lib64/R/bin, it also says version 3

Leonid Shifrin

6:02 PM

@Rojo Ok, so here is what I think is happening. For some reason, the version of R being loaded is still the one that comes with RLink (2.14.0), even when you indicate the explicit RHomeLocation option. But, again for some reason, it somehow picks up a different library for the the core R functionality. What I don't understand is how this happens, but I do have a guess for that.

brama

@OleksandrR. I figured it out....thank you.

Leonid Shifrin

@Rojo What I also don't understand yet is why it still tries to load the built-in R, when you explicitly indicate a different R location. In any case, it looks like I'd need to reconstruct this situation myself, to see what't really going on. So, the more details you can supply about your specific Linux setup, the better. I will then look into that during the next few days.

Rojo

@LeonidShifrin Ok, thanks

Leonid Shifrin

@Rojo Nothing to thank for, yet. In a sense, it is partially my fault that these issues exist at all. I say partially, because on Linux there are lots of incompatibilities between distributions, and it is rather hard to make something work universally smooth for all of them.

Rojo

@LeonidShifrin Also your "fault" that the things that work work. Btw, does this mean you're nearly finished with the other big project, or that it's been pushed one release so you have time to focus on other things like this R fixes now?

Leonid Shifrin

6:20 PM

@Rojo The latter. In my current project, there is still a lot of work ahead. These RLink issues, and also the problem of R3.x on Windows, are important, and I will try to address them for the next point release, which will be soon. Then I will turn back to my main project.

Rojo

@LeonidShifrin Great

Leonid Shifrin

@Rojo gtg now. Later!

Rojo

@LeonidShifrin Later!

rm -rf

6:53 PM

@Rojo @bobthechemist @rolfmertig The vim mma plugin has v10 symbols now. If I find some time in the future, I'll try to improve the highlighting with a pratt parser (like in halirutan's plugin) instead of a simple lexer.

Rojo

@rm-rf Thanks

@rm-rf I want vim-like modes in mma

rm -rf

7:14 PM

@Rojo That would be nice (and probably doable), but too much effort for too little gain :(

Rojo

@rm-rf Depends on who does the effort and how many people gain that little gain

rm -rf

@Rojo I'm not convinced that there are sufficient vim aficionados around to appreciate such a mode in the notebook. I know that I would love it :)

bobthechemist

@rm-rf Nice, it's been very helpful so far.

Rojo

@rm-rf Perhaps one could start small

Btw, switching capslock and escape for vim has changed my life in mma

How could I ever lived jumping around for the input aliases

rm -rf

@Rojo One might have to give up on input aliases...

Rojo

7:22 PM

@rm-rf Never. Just use another key or somethign

rm -rf

@Rojo You can't not use Esc/Ctrl-C for mode switching in vim... Ctrl-C is perhaps less common as well. Maybe you can bind some other unused key for input aliases, but that involves breaking someone else's workflow...

I think I have a good idea of how I would implement this though... it's just a matter of motivation :)

Rojo

@rm-rf Share

I don't think it is such a big deal to change the esc key for input aliases. Nheee

rm -rf

@Rojo At least for basic notebook movements, I'll track (per notebook, stored as tagging rules) a CommandModeQ variable. If true, jk bind to SelectionMove[Previous, Cell] and SelectionMove[Next, Cell]... likewise, d -> delete, y -> copy, and so on. Enter edits the current cell (or moves to horizontal cursor) and sets CommandModeQ to False

Rojo

@rm-rf So you are thinking of jk for movements between cells only

Szabolcs

7:38 PM

@Rojo Keep me updated with any RLink issues like this. I just proposed writing a blog post on WallkingRandomly about setting up external R installations.

@rm-rf should the Mms.SE tag calculus be merged into calculus-and-analysis?

Rojo

@Szabolcs Ok. For now I think all that's important is that I needed to install that compat-libf2c-34

seismatica

8:12 PM

does anyone know when, and how, one could start building Android apps with WolframCloud? There's a blurb for it here wolfram.com/engine but I can't seem to find any documentation for app building.

rm -rf

8:27 PM

@Rojo It's going to be hard, but I think we could get both... perhaps cell movements could be in "visual mode" and normal mode within a cell

@Szabolcs Done.

acl

@Rojo so, how are you finding the linux experience after a few weeks?

good for weeding out the boys and leaving the real men, is it not?

Pickett

@seismatica What "blurb" are you referring to exactly? "Wolfram Engine for Mobile" should provide something like JLink for iOS (I guess it already works for Android?), and then WRI is also pushing the idea that we can request information from the cloud like Siri (the built-in AI in iOS) does for example. But these are mere helper tools. We're not going to be building apps in WL AFAIK.

Sorry, it doesn't work for Android yet. They have to release something that packs MMA onto mobile platforms. In the end I think it will be like using JLink.

Rojo

8:53 PM

@rm-rf This has to happen :)

@acl I definately like it. There are some things that are converging slowly, so i still don't feel settled. For example mma 10. I never got to install mma 9 on Linux :P

seismatica

@Pickett yeah I was referring to "Wolfram Engine for Mobile". So what we build in Android with WolframCloud/MMA? I was excited for a moment there cuz I thought I could try to build Android apps with MMA.

"
Wolfram Engine for Mobile

Native computation and interface for mobile devices, including the complete Wolfram Language and SDK."

Pickett

@seismatica Yes, but you still need to know the Android APIs to make everything work. Sure, you can do you calculations in MMA but calculations do not an app make. I think it will be the same deal as Mathematica+Unity for game programming.

However, I haven't learned how to use MMA with Unity so I don't know a lot about that.

Szabolcs

@acl What I hate most about using Linux is using Mma on Linux. Especially the slow front end.

acl

@Szabolcs I hardly ever use the FE on linux so have not noticed anything

mainly what I don't like about it is tiny things that are different from OS X, which I guess is mostly because I'm used to the OS X terminal etc

but I only ever use linux via ssh nowadays, so I'd probably find lots of other things to whine about if I had to use the GUI

Mike Honeychurch

9:09 PM

Just noticed that TableView is still not an official function!

seismatica

9:24 PM

@Pickett yeah I figured as much. So how can one start building Android apps with MMA? I don't see much documentation on integrating MMA to the Android API (whatever that means since I'm very new to app building).

Pickett

@seismatica I don't think the Wolfram Engine for Mobile is released yet.

Maybe app building is appropriate the way you used it to begging with, I was probably reading too much into it. Sorry.

acl

Lots of questions on Poincaré sections, all with the same variables!

What an amazing coincidence!

Szabolcs

@acl Do you still write C code?

It's not something you'd need for finding eigenvalues. Libraries are better at that.

acl

@Szabolcs no, only very short bits if anything

why?

Szabolcs

@acl I need to profile a program and gprof doesn't work on OS X. I'll figure it out eventually...

acl

9:35 PM

@Szabolcs I don't think I've ever profiled C on OS X

acl

9:51 PM

@Szabolcs when you have some time, could you send me the references for the algorithms used by the graph functionality? thanks

(no hurry!)

Sjoerd C. de Vries

@MikeHoneychurch That's a pity. I use it sometimes for data inspection and while it is very useful, I don't trust it really as it has caused me quite a few crashes.

Szabolcs

10:29 PM

@acl I only know some things about specific functions. What are you interested in exactly, other than connected components?

acl

@Szabolcs connected components should be enough

How considerate! Once you get an answer, remove the data and edit the question just in case anybody ever finds this useful again!

rm -rf

10:53 PM

@acl Ugh... this is annoying. I had already warned this chap for a previous instance.

acl

@rm-rf well, up yours I guess

anyway I wrote a comment, maybe they'll put a small sample in the question

Szabolcs

@acl Finding connected components is quite easy, so you don't need to cite any algorithms. You can just say that you broke the (undirected) graph into connected compoents, and that's enough.

acl

@Szabolcs do you have any references anyway? the audience I'm directing this to will have no clue about this

elementary ones are even better

Szabolcs

@acl I'll find something, let me check this book on the shelf ....

acl

@Szabolcs you are the reference? cool, I'll give them your email :)

ah, it was an incomplete sentence, I see :)

Szabolcs

10:59 PM

@acl BTW if you were to program an algorithm to do this: Select a vertex. Traverse the graph by moving along the edges. Mark vertices as you go. Once there are no more edges to move along, you've marked all vertices in a component.

If there are vertices left, they belong to other components.

acl

@Szabolcs is it really that simple? How do I select the next move? What if I have n options, do I select one at random and backtrack when I hit a dead end? Does this algorithm terminate in a time proportional to number of vertices?

OK, is it really is as simple as you say? If yes don't bother with an explanation I'll work it out tomorrow

Szabolcs

@acl I can't describe it in such a short message, it's not that simple. I'll send you a ref, just give me a min

acl

(I mean this as a question, if it's a matter of thinking for an hour then don't bother)

@Szabolcs don't waste your time, I'll work out how to do this and/or what to refer to tomorrow

Paulo Cereda

11:47 PM

Yay, I won a free upgrade from 9 to 10! :)

It's like ice cream day!

Transcript for