@Szabolcs I think one of the problems with graphs is the lack of developer continuity over several versions

@halirutan OK, thank you very much:-)

@PatrickStevens in version 9 these functions all have an explicit Return at the end. It appears they may have been written by someone not yet used to idiomatic Mathematica. This may explain the performance pitfalls, since performance considerations are one of the most subtle and difficult aspects of Mathematica, and far from obvious to anyone used to other languages

@Guesswhoitis. I think WRI has quite limited resources internally and is not in the habit of crowdsourcing anything (maybe since Wolfram personally has advanced NIH syndrome). Implementation decisions will always involve a balance between time allocated for research and for coding. Maybe the right balance will not be found, such as when a reasonably efficient approach exists but is not obvious to someone who has not thought about it for a while.

Matrix permanent evaluation is known to be a difficult problem, so perhaps the implementor quickly becomes discouraged as to whether a more subtle approach will lead to significant gains, and so starts coding naively. If other planned functionality depends on the implementation but in a non-performance-critical way, this at least allows more things to be implemented, even if it is not exactly consistent with Wolfram's vision of Mathematica as a massively synergistic fountain of delights

@Searke I must say that this concerns me. To introduce so many different syntactic sugar functions, some of which as we see have rather pathological or at least highly unexpected performance characteristics, significantly increases the cognitive load in choosing between them. In version 10 especially there are many new functions with very similar names to previously existing ones, where the actual operation may or may not be related.

I actually think it is getting to the point now that it is no longer a good idea to keep exposing new functions of this sort in the System` context by default. A reasonable use of namespaces would improve the situation considerably because at least there would be some degree of categorization of the functions beyond the rather ad-hoc groupings in the documentation. Add a SyntacticSugar` context, for example.

@OleksandrR. Is " massively synergistic fountain of delights" copyrighted? I want to use it!

@belisarius I hereby release it into the public domain

@OleksandrR. Please specify licensing model :)

@belisarius well, that's the problem with public domain... you never know what it really means ;)

@OleksandrR. How true! :)

@ShutaoTang Very nice. Look's like my sister's hair after riding her bicycle on a tornado

@Searke it also seems misleading to call it a syntactic sugar given the actual implementation. If it were implemented like this (fully equivalent, as far as I can tell), then it would be syntactic sugar:

ClearAll[downsample];
downsample[arr_?ArrayQ, spec_Integer, offset_: 1] :=
  downsample[arr, ConstantArray[spec, ArrayDepth[arr]], offset];
downsample[arr_?ArrayQ, spec_List, offset_: 1] :=
  With[{parts = Span[offset, All, #] & /@ spec},
   arr[[Sequence @@ parts]]];

But, as it is, it does unexpected things that the operation does not require and which would not have happened if it was written out long-hand

I just noticed that the actual Downsample also allows the offsets to be a list, even though it is not documented. But it's a trivial change of MapThread in place of Map to do that so I won't write it out again

"Wolfram personally has advanced NIH syndrome" - oh, definitely.

"Matrix permanent evaluation is known to be a difficult problem, so perhaps the implementor quickly becomes discouraged as to whether a more subtle approach will lead to significant gains, and so starts coding naively." - I guess it's a difference of philosophy; my personal belief is that if you're angling for something that difficult, you either solve it well or don't bother.

@Guesswhoitis. this is the attitude of someone who is solving problems for fun rather than because of a project manager breathing down one's neck :)

True, yes. ;) That feeds back into my previous question: who exactly is hurrying our fine developers?

To change the subject: have you checked your email today?

@belisarius the reason must be that the implementor has never heard of "decorate-operate-undecorate"...

"it is no longer a good idea to keep exposing new functions of this sort in the System` context by default." - once upon a time, new Mathematica versions always came with gasp new packages! Maybe a good time to revisit that.

@Guesswhoitis. yes, it was a good scheme

@Guesswhoitis. see this picture

@Pickett Cool idea, @Pickett ! Please also announce on Wolfram Community when it is done ;)

@VitaliyKaurov I've decided to release a sort of a community edited link archive first, I will post that on WC for sure. Soon hopefully :)

@OleksandrR. Yes. Namespaces are underused. Intentionally.

Sorry... Contexts.

They're beyond the average users ability.

I hesitate to say anything like that in this chatroom. Most of the people who use this chatroom I think believe they are pretty run of the mill users of Mathematica. This isn't the case. Pretty much everyone here is relatively a pro

@Searke The sad thing about the "beyond users ability" is that we now have to deal with millions of string options, because the namespace is just not big enough to handle everything as a symbol.

@Searke I think anyone ought to be able to load a package, though, right? To hive these off into packages would not be so bad

@OleksandrR. No.

@Searke, Besides, you know that packages were how it went in the old days, yes?

Let me try to think. I would guess that 25% of the customers who called me could have loaded a package without live help by email or phone.

(concomitantly with some users making the mistake of invoking the function before loading the package, resulting in shadowing)

Putting functionality for new users in a context or a package is a death sentence. Might as well have never been made

@Searke this seems crazy to me. What are these people, who cannot read documentation or apparently think for themselves, hoping to get out of their use of a complex and subtle tool like Mathematica?

@OleksandrR. I have helped people install Mathematica on Windows in phone conversations lasting well more than an hour

For my own sanity don't ask me that question :)

But, to hear such things certainly puts the predictive interface in context

"more than an hour" - o_O

@OleksandrR. When I first saw the predictive interface I thought it was insanely stupid

It's @#$ing genius

After seeing how it helped people get started using Mathematica, I had to change my opinion and I basically lost a bet with a colleague

I guess it helps beginners, much like training wheels on a bike…

The predictive interface is also... very problematic. You've likely noticed the number of bugs it creates when on

…but I've also heard "I know what I'm doing; don't defy me!"

But overall, it seems to have been worth it

Whenever I click in a cell, it selects the code before it, going back a line or two. I can get an insertion point by using the right/left arrows. I can select text in the middle by using SHIFT plus arrows. But I can select with a mouse. Quit the FE and reopened the notebook with no change. Anyone know what it is?

I wonder to what extent Web 2.0 and so on has fostered this culture of helplessness. We see users asking questions on SE sites as a first resort rather than attempting to do anything by themselves. I am not that old but it doesn't come naturally to me to ask others in such a way; this possibility simply didn't exist until very recently. So, I still take the position that if one wants to know something, one had better try to figure it out independently.

Coming from Usenet and really old computers, I learned self-sufficiency pretty quick. I prolly shouldn't be, but seeing people who ask first before consulting the docs still makes my blood pressure shoot up…

@Guesswhoitis. I was like that too when I was young. It's cultural, like you said.

I can still find my first posts on the Ubuntu forums because I couldn't figure out who to properly use apt-get

You can understand, to some extent, the users' motivation for this. They have learned that by asking they will get a good or at least reasonable answer quickly, and with minimal effort expended. Maybe the answer will even be better than they could have done themselves after applying much effort.

The shame. The utter horrible shame. documented forever online

@OleksandrR. I often try to take a socractic approach to answering questions. Requiring the person to read something or do a very simple task of combining two concepts together

I'm glad that you do, @Searke.

This isn't very popular with most people, but really helps with people who abuse help.

Tangentially related: back when I was still teaching, more than a few kids apparently came from high schools where spoon feeding was the norm, so college was a total shock. Ugh.

@Guesswhoitis. well... programming and such is particularly ... I didn't go to a highschool that spoon feed me. But when I started with programming and comp sci, it was a culture shock

I had a lot of trouble learning how to learn to do the things I wanted to do. I would have killed for a reasonable book on how to use the terminal when I was 17.

People new to programming don't understand the questions they ask. Or what the fundamentals are they can break a program down into. This is why you almost always have to engage new programmers in a dialog to determine what their question should even be before they ask it.

Image processing questions are good examples. Even relatively experienced programmers often are at a loss for what to ask when they want to do something with image processing. It always requires a conversation to ask them what their project is.

Maybe that's an option now. Back then, it took reading a page at least thrice, and another book if need be. Whether people today are better for not having to do that, I'm not sure…

@Searke I first wrote a program at about 11 years old, IIRC. I had the BASIC reference manual and that was it. Nobody to ask. Fortunately as a kid one doesn't have aspirations to solve any great problems! (But I managed to teach myself a bit of information theory, through my attempts to write an encryption program.)

@Searke yes, image processing always struck me as quite an art, where one cannot really make much progress without considerable experience. nikie's answers are tours de force in this respect

On the RPi version of MMA, a pallete is created at startup with a few links but nothing really of interest. I'd like to make it stop showing up (and think I've done it before, but can't put my finger on the solution). Where might palletes/notebooks like this be stored?

I know that the file is SystemFiles/FrontEnd/TextResources/MathematicaNavigator.nb but I don't know where/when it is being loaded.

ahh, what an awful world is one without grep.

@OleksandrR. It's happened more than once that i was submitting a question and worked it out. Reducing what i think i see to the simplest case to engage other people's assistance is what usually does it for me.

what does infinity mean in Mathematica? How Mathematica treat or manipulate infinity? For example, when you have an integral with infinity limit, how Mathematica substitute the value of infinity?

@xsk8rat yes, definitely. I think clarifying an issue in one's own mind enough to explain it to others gives one a much better chance of solving it. This is sometimes called "rubber duck debugging" although I think the name/concept is quite silly

@barznjy People figured out rules to evaluate integrals with infinite endpoints. Mathematica can implement those same rules. I'd start here:

khanacademy.org/math/integral-calculus/…