@rm-rf, @Murta I voiced my support for this operator, at the point when this decision was considered, but the higher powers decided otherwise :)

@halirutan Not fully sure about the RPi (I think it did), but it certainly did work on the v10 pre-releases. It made working with associations a lot nicer... I'm already missing it :(

@LeonidShifrin Was there any reason from an implementation/semantics/language pov for its removal? On a related note, what do you think of @Akater's comment above re: interpreting (a, b) as Sequence[a, b]? I can't think of anything that could go wrong and it does seem like a meaningful syntax

@LeonidShifrin What a pity!

I'll complain about it in a formal feedback!

@rm-rf I didn't even get to play with that operator

@Rojo it was like this:

<|"a"-> 1,"b"-> 2|>@>a

1

You could extract Association values without a lot of ""

Oh, great to avoid the quotes

Yes! And to do things like: <|"a"-> <|"c"->3,"d"-> 4|>,"b"-> 2|>@>a@>c to get 3

@Rojo It even looked nice in the FE, like an @ with a tiny arrow

:)

Very interesting to make columns reference

Aww, well

Associations are nice anyway

but I can see how going back to something less nice is disappointing. There must be a reason

Yes!... I'm very curious to know how Dataset will work!

The docs are not ready yet

reference.wolfram.com/language/ref/Dataset.html

I get happy about weighted data working natively in Histograms now.

reference.wolfram.com/language/ref/Histogram.html

@Murta Makes sense

@Rojo Not just disappointing... I made heavy use of Associations and @> in a large package that was supposed to be "v10 ready" from day 1 (which is kind of a good use of these pre-releases). Unfortunately, now I need to revisit that and change it up :(

I never happened to use WeightedData anyway

@rm-rf Can I ask?

@rm-rf I have used it in a package too!.. Changed today

@Rojo You mean what the package is about?

@rm-rf Yes

Oh, just for my research work... v10 had some very nice features which made the code flexible and I was writing it in v10 so that I can use it on the univ servers when v10 releases

wahrs mma 10 dammit

Hi, someone familiar with Fourier cos expansion? I have a problem takes me a whole day without progress. Any suggestions would be appreciated.

Why there is a change in the coefficient behavior around 8 ?

For n<8, the coefficients decrease very fast, but after n>8 they decrease very slow and the sign is oscillating. I think this behavior has something to do with the fact that the gaussian is cut at the boundary, but exactly why this happens ?

@rm-rf Could you indicate for what kind of data processing you're using it? I think I understand associations, but I don't see a killer application for the type of data that I most often encounter, which is tabular data for which associations seem to be too inefficient.

@rm-rf Re: reasons - I don't know of any really good one. Re: Sequence - I don't see any problem here either, since currently (1,2) and such are invalid syntax, so it can't possibly collide with anything just on the parser level. However, the language might later get other data structures like e.g. sets, where such syntax (using ()) might also be useful (this is a complete speculation on my side, nothing more)

@Pickett Thank you! But is OAuth the only way to authorize oneself in Twitter [and other private services]? Do I get access tokens underneath when logging in via browser? OAuth seems to serve the purpose of sharing access without sharing passwords but I don't mind using full credentials to connect.

@xslittlegrass You are right, it's the cut. The irregularity comes from the fact that after a certain finite number of approximation by cosines Exp[-x^2] on a compact segment does not resemble cosine in any way. I'm sorry if this doesn't add much but I don't think there's anything else to explain here. You could try other segments (with FourierParameters Option) and see what happens there.

@Akater It's not the only protocol that exists for authorization but it's the one that Twitter and most other services uses nowadays. Logging in on Twitter.com you're not authorizing yourself against the API, you're just creating a session for yourself that Twitter's servers recognize. So, if that's what you're asking, I don't think it makes sense to think about what happens in the browser in terms of how the API works. Twitter doesn't have to connect to itself.

@Pickett Thanks again. Then maybe I should have asked how to create a session. But that would probably be an offtopic discussion.

@xslittlegrass To see the effect of the truncation consider these:

@Pickett I mean, I consider web interface to Twitter as a GUI for some set of commands but after a couple of years of Twitter [and other Internet services] use I never get to those commands (as opposed to other software where I eventually understand the lower-level functionality just by using and studying higher level tools).

 Integrate[f[x] Cos[n x/a], {x, 0, a Pi}]

Manipulate[
 Plot[1/4 I (*E^(-(n^2/(4 a^2)))*) Sqrt[\[Pi]] (Erfi[n/(2 a) - I a \[Pi]] -  Erfi[n/(2 a) + I a \[Pi]]),
  {n, 0, 20}, PlotRange -> 2],
 {a, 1, 10}]

@Akater yeah, we're not meant to understand that innards of such services, I think. The API most surely don't correspond to Twitter's own code. You create a session by logging in on their website. The browser sends the information as a POST form, then Twitter sends back a cookie with an identifier in it that the browser stores. From now on the browser includes this cookie in its requests, Twitter's servers check the information in the cookie against the information it has recorded

about the session to make sure it's legit. The things that the browser does can sometimes be faked, so that you can set up and use a session in a different programming language, but in my experience it is very tricky to get it to work.

@Pickett yeah, we're not meant to understand […] — that sounds a bit depressing. :-) In an attempt to make all this less offtopic, what are the conceptual differences between a web service and a Mathematica package that I download from remote server, or a Mathematica expresssion I evaluate on a remote machine. I don't see ones.

And my experience is not from trying to do this on a major website like Twitter, I can imagine they've fixed it somehow so that it's not possible.

@Akater If it's a client-server relationship I guess it is conceptually equivalent.

I don't mind going off topic btw, this is the chat after all :)

@Pickett it is very tricky to get it to work — doesn't that alarm anyone? If some human activities become difficult to automate [by formalisation], isn't that a bad sign? If those activities are themselves basically formalised algorithms, isn't that a very bad sign?

@Akater Twitter and other web services never intended for you to fake a log in to be able to do the things you can do in their GUI. They created a user-friendly API to enable you to do those things.

@Akater Twitter have limited more and more over the past two years or so how much you can do with their API. This is very bad news, and the reason is said to be because Twitter wants more control over the content so that they can leverage it better to earn more off of ads. This is bad news if you like open data.

@Pickett Well, I don't have a CS education but that sounds extremely wrong. I wish every web service had a console, at least. :-) (I understand this is quite unfashionable, everybody tells me that.)

I think with some experience working with APIs such as Twitter's you'll be more positive. Companies nowadays want users to contribute to their services and they really do want to make it easy. But it just turns out that the API setup is as easy as it gets, as of now, technically speaking.

@Pickett …basically, I love Wolfram language precisely because I'm able to automate things with it, in a very sane manner. But interacting with Web is far more painful, and it appears that it's not because of language's limitations but more because of the design of the Web.

@Akater Mathematica is not a good language for it, because there aren't any libraries made for Mathematica. PHP, Java, Python, Ruby etc. all these languages usually have pre-made libraries. So connecting to Twitter and tweeting turns into just writing login(username, password); tweet("My tweet"); or something like that. But with Mathematica we have to do the things that those libraries do ourselves.

So it's not a flaw of web technology, and I wouldn't say it's Mathematica's fault either. It just happens that Mathematica is not big enough to get all the conveniences.

You could of course use JLink to use the Java library and communicate easily with Twitter in that way.

@Pickett Well, this is a very interesting topic. I understand that the power of Python is in its crowd. But this doesn't make Python a better language. Wolfram, as Liap, is a great language to do anything, it just doesn't have a community. (BTW, this community is probably better than the Lisp community: it had less time to develop, and yet it's extremely useful and responsive.)

It doesn't make it a better language, but there are situations where it's a better choice of language because of the solutions that already exist for it.

@Pickett Sure, but for me the point of writing in Wolfram is precisely to avoid Java, Python, etc. :-) I tried Python 3 times [for different goals], and it was an awful experience. Wolfram was the only language (of 5 or so) that I immediately loved.

@Pickett Yes, that's the right expression for it.

@Pickett Speaking about libraries, Wolfram is heavily standardised in comparison to, say, Lisp (probably because a great amount of source is closed). We are all aware of disadvantages of closed source but it could actually have played a positive role after all. (Still, I wish it became more open, maybe gradually.)

@Akater @MichaelE2 Thanks for the help, I"ll look more into it :)

@Akater It's great if you want to be able to do some image processing one minute and then som control systems stuff the other. I'm not convinced the breadth of naming conventions and programming styles is a big detriment to other languages. When I worked with PHP I worked for a period of time in Wordpress (procedural), then for a period of time in Codeigniter (object oriented, so called MVC) and so on. I didn't dabble in so many different areas as I do now. I think it is typical.

Within Wordpress and within Codeigniter everything is standardized and follows well documented conventions.

Someteimes I know the answer but don't even try to respond because I expect someone else to do it before me (and better than me). I suppose it's a common problem at StackExchange, and has been discussed already, maybe even numerous times?

@Rojo how can I access nest iteration index inside a nest

@Akater I've heard that before here. I searched for a discussion on meta but I couldn't find it. Imperfect answers can lead to nice feedback I guess :)

When will version 10 come? Any information?