Hey Blanco

@Rojo Hi

@halirutan How's everything

@Rojo hmm, writing paper and want to do so much other nice things.

@halirutan Oh, poor you

What would you like to be doing?

I can't complain, really enjoying these days. In 3 days I'm back to real life however

Compress@Compress@"hi @rm -rf"

Hi

@Rojo There are several options which are better that what I'm doing now.

One is to work on my language plugin for IDEA.

Merry 5K questions

\o/!

At the main page with the questions, when I want to reply to someone @"..." does not work. After inserting @ and type the first letters the name should be appeared above but it does not. And if I insert by hand the nickname after @ then it disappears. However, here is working fine. Any ideas?

@Vaggelis_Z You mean in the comments? Yes it works, but you have to consider some special cases:

E.g. when someone has written an answer (or a question), let's call him Bob and you want to write a comment under Bob's answer telling Bob something, then you don't have to put the @, because Bob get's pinged anyway.

Thats why sometimes it does not work.

But when, let's say Fred wrote a comment under Bob's answer, and you think Fred tells bullshit and you want to start your hate-war by commenting under Bob's answer and saying something to Fred, then you can ping @Fred and you should get completion-suggestions.

@belisarius we also have the word "media" but that could be mixed up with sock and be confusing

I've always thought, more or less, that average<-->promedio, mean<--> media

You think I'm off key?

@Rojo Not sure.

@Rojo Have you ever calculated the sock of a population? :)

@belisarius You're around?

@Rojo yup

I'm probably about to make a small palette out of an answer. How would you share it sipmly? Embed it in a image like you always do? If so, how do you do that?

@Rojo See the third palette here

Hello

@JxB Hi

Anyone familiar with automatic figure numbering with, for example, the journal article stylesheet? I've searched the site and googled, got nothing very helpful...

@JxB What's the issue?

I can get auto numbering by copying the template "FigureCaption" example below any figures I add to a notebook. Yet I'm wondering how to do this from the menus. Adding a style FigureCaption gives a blank template with no auto numbering. Insert->Automatic Numbering... OTOH gives a single number. Seems to get the boldface "Figure X. " has to be formatted manually?

The question was covered at Mathgroup, but the answer was to create a new Figure style, which includes its own caption. Not a bad choice, but I'm wondering if I'm missing something in the Vanilla stylesheets.

@JxB How do you insert the template?

I never used them, want to see how it's supposed to look

New->Styled Notebook... select Journal Article. A little down the page shows the "Figure 1. Figure caption." Copying that below the next figures gives a new figure number.

@JxB Got it

Ok, they got some structure

I'd just dock a button or something that creates a new cell with that structure

Something like this:

If you just create a new style I wouldn't know how to keep that same structure. I'd only know to put the "Figure X" as a cell frame label (which sounds nice however but that's not how the template looks like)

@Rojo Nice! I could extend this to an ActionMenu with a list of caption types (Figure, Proof, Picture, etc.)

Thanks!

@JxB Right. Then install palette and forget about it

@Rojo Your Palette gave me the idea to check the Writing Assistance Palette. Nothing there though..

@JxB If not going to export the nb to LaTeX, I would prefer modifying the "FigureCaption" style in the system's JournalArticle.nb stylesheet, by adding the Figure head to the CellFrame or something. (But of course Rojo's method is much more formal and neat)

@Silvia How can you talk about me in my presence without saying hi?

Thanks @Silvia.

@Silvia Long tiiime :)

wired my chrome can't automatically refresh this chatroom page..

@JxB You're welcome :)

@Rojo Always have endless work to do. So less time left for this site..

@Silvia Pity, you're contributions are always valuable

@Rojo Thank you my friend :)

@belisarius Thanks

@Rojo I want you to tell me "you're contributions are always valuable"

@belisarius No need to, you are always around

@Rojo I insist

@belisarius I'll grant you your wish, only because I wouldn't be lying

You're contributions are always valuable buddy

@Rojo I feel relieved, thanks

Welcome

Who would have thought, the ReplaceAll version with nested lists of rules leaks when you use Unevaluated

Unevaluated@Print@2 /. {{Print -> Pront}}

Need double Unevaluated. I wonder if it's a bug that will be fixed or I can safely use double Unevaluateds

@Rojo A very good observation. Tracing will show what happens - ReplaceAll thread over a list of rules in a way which gets one extra Unevaluated eaten up. Whether or not this is a bug depends on whether or not one adopts a rule that every built-in function does not evaluate its argument wrapped in Unevaluated, anywhere inside its implementation. But if you think of it, this requirement, while probably very sensible, also probably can not be fully formalized (made independent of the

@Rojo semantics of the function in question), partly because some of those functions are themselves implemented in Mathematica internally (even if not in the top-level code), and partly because the way functions evaluate their arguments does strongly depend on their semantics. In this particular case, I am more inclined to think that this is a bug. But in general, this class of situations seems to me to be a strong argument against using Unevaluated at all, substituting them by Hold etc...

@Rojo ...if at all possible of course, followed by ReleaseHold when you need to evaluate it. On a deeper level, I think this situation reflects the fact that the split of the functionality into black-box internal code and user-defined functions is arbitrary and something completely external to the core language design. We know that Unevaluated isn't a robust wrapper for a general evaluation process, since it is hard to tell in advance how many levels of it may be needed, and this may ...

@Rojo ...even be data-dependent. With the built-in functions, then, each and every of them has to provide a "personal guarantee" with respect to Unevaluated - which makes this design inherently fragile given that functions are so many and new ones are added. Of course, such a fundamental one as ReplaceAll is more of a surprise, but it only confirms the general statement I've made.

@Rojo What really should have been done here, at the very least, is that for every function the documentation explicitly specifies its behavior w.r.t unevaluated code passed to it. This has not been done systematically, however, so this just adds more cases of undefined behavior to those caused by the lack of precise specification for other sorts of behavior. Of course, we should also realize that the behavior of Mathematica functions is much reacher than for most other languages, due to ...

@LeonidShifrin Irresponsible overloading? :)

@Rojo ... the complex evaluation model - which also gives the language more power, so it is harder to have all this behavior being extensively documented.

Hi @belisarius! Yes, I'd think so.

@belisarius But as I just tried to express, I think this is generally a more complex problem (although perhaps not in this particular case)

@LeonidShifrin And we shouldn't forget that designing a debugger for a language with rewrite rules isn't trivial at all

@belisarius True. M is very different from almost anything else out there. The understanding and appreciation of this is not helped by the fact that it emulates lots of more standard constructs so that its code looks more familiar to people than it probably really should.

@LeonidShifrin But that is also what allows to lessen the steep learning curve. Think for example of a bare LISP

I mean: M is hard to learn, and could be much harder if the true architecture was exposed

@belisarius I agree. I think that a good way out would be some intermediate (perhaps Python-like) linguistic layer which would allow people to solve 99 percent of the problems without engaging into a fight with evaluator. In fact, LISP evaluator is simpler than M's, and has a better defined semantics, due to the complexities of M's evaluator such as rules, attributes, and infinite evaluation

@belisarius Alas, I have to leave now. I just stopped by briefly to express my opinion on this case found by @Rojo, and was about to leave right then. So, talk to you later!

@LeonidShifrin Splitting layers is a tradeoff. Then, if you want to switch a behavior from one layer to the other you get into troubles

See ya

@belisarius I agree. Perhaps what I meant is the more strongly typed language implemented in the core layer, with its implementation being open and freely available, and extensible.

OK! don't engage again! :)

@belisarius Interesting topic. Let's continue next time we chat here!

Bye!

Bye!