@MikeHoneychurch In the product these work (open dialogs and palettes). I've reported the web issue.

Is there anyone here running Windows or non-Gnome Linux? If yes, could you tell me what order this gives? --> ChoiceButtons[{"OK", "Cancel", "Third"}, {Print["a"], Print["b"], Print["c"]}].

I mean what order it shows the buttons in.

@MikeHoneychurch does bibtex really work? I seem to be able to insert footnotes only

and did it work for v8?

@F'x it does sound like a bit of an overreaction doesn't it

I don't know what that really means though

ok I diffed a couple of files in AuthorTools against those in v8. the changes seem to be a) modifcation dates like above, b) change of names, eg NotebookNameAT being replaced by NotebookName

I only checked 3-4 randomly-chosen files though

well, I can make tables of contents for instance

did this sort of thing work in 8?

Nice what you find if you don't search for it:

f[x_] := x^2;
Language`ExtendedFullDefinition[f]

@halirutan That was there even in v8 (possibly earlier)

@halirutan It's used by the parallel tools to transfer definitions between kernels

I think I can officially declare myself "happy" with v9!

@Szabolcs On Kubuntu 12.04: "Third", "Cancel", "OK" in that order

@Szabolcs "OK", "Cancel", "Third" on my Windows 7 x64

Strange.. When I type Plot[Sin[t], {t, 0, T}, Prolog -> {Black, Rec}], I can't use Ctrl+K getting an auto-complement of Rectangle.. But typing Rec alone in a fresh cell do give me Rectangle in the pull-down list..

Have posted a question on it.

@Silvia Hmm, one has millions of options for this CodeAssist (Check Edit->Pref->Editing Options->Priv Ed Opt->CodeAssist Opt) but with none you can turn off the context-sensitive completion..

@halirutan I'm really sad to hear that :(

My English is so bad that I can't remember most of the Function names :O

@halirutan Maybe if we understood what CompletionParameters does...

@rm-rf I'm trying it, with no hope.

Random question about SE chat: Sometimes people comment on how many people are in this chatroom at once... but where is that info displayed?

@AndrewMoylan It's shown if you view the transcript, for one.

Ah I see a button there, "Join 12 Users in this room"?

Yup.

Also, you can count the icons on the right (if you're not on a tablet)

@rm-rf Where's the fun in counting?

"How can I use Mathematica to count the number of people in chat?"

StringCases[Import["http://chat.stackexchange.com/transcript/2234", "Text"],
     ___ ~~ "Join " ~~ x : NumberString ~~ ___ :> x]

See, isn't that much better than counting? :-)

Excellent!

Excellent+1

'Night All!

Night

Anyone around?

yep

I am wondering if I should trim questions 3) and 4) on my question here: mathematica.stackexchange.com/questions/15507/…

The criticism m_goldberg makes seems apt.

I am still pretty curious about both of them, though, so I was always hoping to raise them here (if I do trim them)

*also

yay. Usually it is frowned upon to have multipart questions like that as well. Better to try to break them into separate questions instead of a thematic larger question

probably better to ask them earlier in the western world ... much smarter people than me are around at that time!

It should circle around ;)

How does it look now?

Is two parts still too much? They seem fairly closely related to split it into two questions...

looks good to me

I think those questions seem more tightly connected

have you check out Leonids and Maeders books?

I found them to be the best source of mathematica as a language itself

not a formal definition, but much less about applications and more about the language proper

That sounds very interesting

The name Maeders seems familiar

I think I saw that in my search before asking the question

The only two books I got were ones on rule-based programming, though

it is such a great book, old but amazing

what do you mean re rule based programming? Something more general than mathematica, or concerning mathematica's implementation of these ideas?

A little of both

"On the Implementation of a Rule-Based Programming System and Some of its Applications" by Marin and Kutsia and "Rule-Based Programming with Mathematica" by Marin and Piroi

ah

Is the one by Maeders you mentioned available freely?

No, only Leonids book is

Probably at any university library if you have access as it is really old

mathprogramming-intro.org

This?

yep

Is this the Maeders book? amazon.com/Programming-Mathematica-3rd-Roman-Maeder/dp/…

yes

sorry should have sent the links!

It's okay, my google fu is strong ;)

Thanks for the info!

Maeders book gives a great discription of the package/module structure of mathematica and leonids book of the data model

Those both sound useful, but I am a little more interested in the unique technical features

leonids book in parts (especially the beginning) has more of a language definition quality to it. Nothing at all like a formal definition, but the closest you will probably get

yeah ... for that I think you are out of luck

That's what I'm worried of :/

though ask again when people like Leonid are around! He is a serious resource

I think I will

you mean something like how languages like python have formal discriptions of the semantics etc? Yes/

?

Yes, exactly

As I said in the question, I'm interested in making my own programming language, and Mathematica is very interesting to me for getting ideas

Yeah sounds interesting ... I have never seen such a document in my searches

do you have mathematica to play around with?

I find it feels a lot like languages like F#/ML with stronger symbolic/macro properties

Heh, funny story... I meant to download the trial version today, but apparently the email address I gave them caused their system to choke. It has a '+' in it, and going to the download page and logging in resulted in a fatal error

I have used it before as a student (both math and CS)

ha ... using the gamil feature?

Yes

I find the + syntax chokes a lot of email scrappers ... makes me sad ... love the idea of having focused email addresses for such things

I'm still deciding whether or not to buy it. $300 is a lot of money for such a simple cause, and I'm not even sure whether I can use it for application development

what kind of application are you trying to build?

I don't have anything specific in mind right now. The main purpose of building applications would be to explore the unique semantics of Mathematica and how they might benefit real-world development

I do have several simple applications that I usually tackle when learning new languages. I 'd probably start with those

I really need to go to bed now, though. Thanks a lot for all the great info! I'll definitely be back later :)

good night!

Anyone aronud?

@Szabolcs Thank you @Szabolcs

@acl whre did you find AuthorTools in M8? I do not seem to have it (in windows)

To the moddy people: when should one post separate answers (both by yourself) to the same question, instead of offering them all in the same post?

For those interested: 82 pages and counting. Still shy 2 chapters and some obligatory front/backmatter, so I expect an additional 20 - 30. Must sleep now.

@Rojo I see mr.wizard doing that quite often. I'd say it's ok if there are two distinct lines of thought involved and/or when both answers are large.

@Rojo As Sjoerd said, it's ok when the answers are very distinct in style and substance or quite large. Some good examples that come to mind are JM's answers on interpolation/fitting using 2 different approaches, Oleksandr's Nelder-Meade answers. However, it's better not to do if the approaches can reasonably be posted in one answer (e.g., 3 answers by the same person using Cases, Pick and Select respectively is not appropriate, even though they're "distinct")

@SjoerdC.deVries Quite often? I didn't realize I'd made a habit of it. I thought I'd only done it a handful of times at most. Anyone know how to set up a Data Explorer query to check that?

@rm-rf My motivation for posting multiple answers has been when I thought there was value in separating the votes for different methods, e.g. I'm not sure one of them works right or something.

@Mr.Wizard data.stackexchange.com/mathematica/query/87549/…

@rm-rf Thanks. Let me try it.

So much for my "handful" :-O

@Mr.Wizard :D

Now I have to review each one of these.

The first few I checked look OK but I can't help feeling now that I've done that too much. Sadly another manifestation of my poor memory. :-/

Hello @MrAlpha

@rm-rf do you know of any "crowd funding" sites that can be used for arbitrary things, not only e.g. creative projects?

@Mr.Wizard In any case, my point (every time I've bugged you about it), was that things need not always be left to vote. I mean, you're very knowledgable about the performance aspects and you needn't need a community vote to say, decide that using Inner is cleaner and faster than some other clumsy aspects... I'd probably trust your judgment than votes alone (which may or may not reflect things correctly)

@Mr.Wizard How arbitrary are we talking about? Also, is there a deliverable?

I appreciate your confidence. Honestly I had no idea I'd posted double answers that many times.

@rm-rf Let me not be vague. I'm feeling increasingly obsolete as far as this site is concerned as a v7 user but as a hobbyist with certain circumstances I cannot justify getting v9 now. I'm wondering if I've been helpful to enough people on this site to merit a communal v9 Home Edition purchase. I hope that's not too egotistical of me; I really do want to remain helpful.

Certainly "send money to xxx" isn't appropriate, but I've seen sites that track dollars toward a specific goal and refund contributions if the goal is not met. That seemed at least plausible to me.

kickstarter used to be like that, but they've since tightened things after some missteps... They now need some "prototype"

But I certainly think something of that sort could be worked out... I know, I certainly would pitch in.

@Mr.Wizard That wasn't meant as a negative, more as a label of approval. Also didn't mean to say that you're doing that often in relative sense. Given your huge production a low rate will give you larger absolute numbers than most others. In that sense did I say you do that more often than most other people, but again, that's not meant to sound negative.

@rm-rf That's very gracious of you.

@SjoerdC.deVries Okay, thanks. Still I had no idea I did it that often, and I'll try harder in the future to combine answers. I'll rely on comments to let me know if one of several methods fails.

I've also been asking Aarthi about having a site-wide contest (sometime in Jan)... we'll have to work out the details later (they're now busy with the hats thingy). Perhaps v9 as a few "grand" prizes would work.

I don't really feel like competing. Despite having the top ranking I've always said I'm far from the most knowledgeable member, and personally I don't want to make my reason for posting be about selfish gain anyway.

@Mr.Wizard There are sometimes good reasons for providing several answers. I tend to combine them, but these answers often end up as a huge mish-mash; quite unreadable.

@SjoerdC.deVries Do me a favor, will you? Look through my multi-answer posts and as a third party tell me how many you approve of, please?

@Mr.Wizard Pfft... just post "normally" and you'll still end up at the top =)

@rm-rf I know that's not true, but I do like the mystique. :o)

Trust me, I've joined a site new and won an iPod in 10 days and never participated there since... (stackexchange.com/promos/5/the-new-ipad-is-here)... I have no regrets :D

hehehehehe...

Okay, things to do. Sjoerd, R.M, have a nice day.

Bye, and take care :)

Heya

@Mr.Wizard I could do that, but really, I don't think that's necessary.

I have two questions related to mathematica.stackexchange.com/questions/15507/…

Actually, I included them in the question at first, but then edited them out because they weren't as important and kind of took away from the main questions

But maybe someone here knows something

First: Has there been any news since this? reddit.com/r/IAmA/comments/qisot/…

Wolfram mentions that he's interested in making a "pure" subset of Mathematica freely available, and increasing its accessibility as a language

That was 9 months ago now and there didn't seem to be any indications of heading in that direction with Mathematica 9

Second: Is there any way at all to get just the core kernel, without the frontend, W|A access, all the bells and whistles? Hopefully a little cheaper too... The Home Edition isn't completely out of my budget, but with that I'm also concerned that even with the home edition I wouldn't be allowed to create, license, and distribute arbitrary applications (that don't use proprietary APIs like W|A, of course).

I realize that this is probably a better question for their sales staff, I just wanted to clear things up beforehand if I could (and save myself a little embarrassment)

@Kiyura I would not hold my breath

:(

@Kiyura but to run anything, you need a running kernel. even if the home edition license does not forbid whatever you want to to with your application, a kernel needs to be running for it, using up a license. (eg, running things in batch on a cluster uses as many licenses as there are running jobs)

Even if you compile to C?

And/or use the CDF Player/Player Pro?

@magma in AddOns/Applications/AuthorTools (which is inside the mathematica bundle on my mac, /usr/local/math/AddOns/Applications/AuthorTools on a linux machine I just checked on, etc

@Kiyura compile what to c? an entire mathematica program?

you can compile routines and link them back in, but you cannot make standalone executables

I thought that I read differently somewhere

I am not that familiar with the Player stuff, but you need to pay for them too if you are going to package and distribute applications (and get them signed? but I am not sure, better ask others)

@Kiyura well no, you can't create standalone executables

wolfram.com/mathematica/new-in-8/integrated-c-workflow

"Generate standalone C code functions from Mathematica-compiled functions."

Oh, that's just what you were saying

yes

I was also looking at this wolfram.com/solutions/industry/software-engineering

well OK, just make 100% sure that what you want to do can be done before spending money on this. eg get the trial and try

Among the capabilities listed is "Automatically convert Mathematica programs into standalone C code with code generation functions"

The link goes to the doc for the CCodeGenerator module, which reads "Code generation from Mathematica involves converting programs written in the Mathematica language into other languages and then supporting them so that they can be executed. The Mathematica compiler provides a system for code generation into the C language."

Can I take that to mean that you still need Mathematica or a Player to actually run such code?

Wouldn't be very "standalone" if so

but why don't you get a trial? you'll work out what it'll take me 30min to explain

I am trying, my email address made their user portal choke :)

I used a gmail + address

I'm assuming support won't get back to me until business hours Monday

oh yes

I'm not sure what I would find out by having the trial that I couldn't find out from the documentation, though. And the more I read, the more I think you might be right - what I'm trying to do may simply not be possible. Which would suck, but wouldn't be a game-killer... I can still study the language "from the inside," it would just be a lot more convenient to be able to work independently from the frontend, just using the kernel to run and compile code

I really need to chat with someone official to clear this up as well as the licensing issue

Licensing also isn't a game-killer but it would be nice to be able to use any code I write (and the compiled result, if that's possible) freely

well, there's this, but this is not what you have in mind I guess

I saw that, but like most of the docs relating to C compiling, it asks more questions than it answers.

That particular module seems like it's geared towards analyzing C code rather than compiling or linking with Mathematica

Actually, maybe that module is how Mathematica code IS compiled to C

yes, I think so

There's also this reference.wolfram.com/mathematica/CCompilerDriver/ref/…

But it looks like a thin wrapper for whatever C compiler is on the system or bundled with Mathematica

i.e. still doesn't answer a lot

but you can also automatically produce C code and link it back in; for example, to produce code to make a list of sines, you can do ExportString[Compile[{}, Table[Sin[x], {x, 0, 5, .1}]], "C"] and you get (this will be a bit long):

@Kiyura if I'm not mistaken, some Mathematica C compiled functions can really run as "standalone", together with a couple of distributable libraries. I think that this was not clear on M8, since it was not ready at release time. I was also interested in knowing more on the subject. If I was told correctly, the runtime covers basic functions (similar to what old compiler did without calling the Kernel), and not advanced Mathematica stuff, and so, it is not like MATLAB runtime.

I am fine with a very basic subset

but of course you cannot produce C code to use NDSolve

@Kiyura OK could you describe what it is you'd like to do? Then maybe it's easier to answer (if I know)

Study and understand the underpinnings of the language semantics

@P.Fonseca which old compiler?

@Kiyura by producing C code?

I mean, the C code here is produced by routines written a long distance above the base language. it's not like everything you do gets internally compiled to C code

Yes, that would be one purpose for being able to compile to C; the other would be to get practical experience with Mathematica as a software development language to get an idea of how its semantics affect practical development. (Since I'm interested in possibly using some of the ideas to make my own language)

Right..

I do not think analyzing generated C code will get very far either, which is why my main tack was to try to find a language specification

I can probably derive most of it from the core language docs

It will probably be most productive for my purposes to just study the language from the "outside" by going through the docs and experimenting.

@acl that's correct. It cannot compile NDSolve, but, for example, you can implement your own RK code, with simple functions, and have it compiled. (I mean the Wolfram Virtual Machine compiled code)

Is it possible to link with arbitrary C libraries?

@P.Fonseca yes that sort of thing is more or less what I use mathematica for. but I did not know I can compile the code and drive it without mathematica. Maybe it is possible.

@acl @Kiyura check this post (and next in thread...): forums.wolfram.com/mathgroup/archive/2011/Mar/msg00448.html

@Mr.Wizard Contact me at arnoudb@wolfram.com.

This is linked to, and seems to demonstrate compiling a Mathematica function such that it can be used in a standalone executable, without Mathematica: reference.wolfram.com/mathematica/CCodeGenerator/tutorial/…

@P.Fonseca, thanks a lot, that definitely answers one major question. Theoretically, then, if I only used functions that could be compiled, I could write a mostly complete program in Mathematica and then compile it link it with a minimal C main-loop to get a standalone program

@P.Fonseca ok thanks, that is a good point

@acl @Kiyura you are welcome. I only wished that the generated libraries could link to the player (and specially, not the Pro, since this is what mathworks is doing), so that I could call more advanced stuff from other programs (as I used to do with MATLAB).

@P.Fonseca yes; I thought that was what @Kiyura wanted to do

and it would be fantastic.

Well, I am not interested in nearly anything to do with the extensive math APIs

Pretty much just the core language

So that should be fairly useful to me

in what sense?

M-expressions, rewriting and rule-based programming, its approach to functional programming, etc.

Its most basic semantics

I am probably being thick, but how does machine-produced C code help with understanding that?

It might not. There is a chance that I could learn something about the semantics by looking at the generated C code for certain functions, but I am expecting it to cut off with a black wall fairly quickly when you call into library functions. Generating C code is more closely related to my tangential goal of writing practical applications to learn the language from a first-hand perspective.

But that may be just as well served by writing applications purely inside Mathematica

On a side note, I'm also very interested in the natural language input... but I'm sure that falls very squarely under "trade secret"

And that would be a lot more opaque to a reverse-engineering effort

@Kiyura it's also done on a server, not locally

You probably need to play with a trial version for a few days to see how feasible what you want to do is

@Mr.Wizard well, if you'll sign a contract to not use infix for a month, I'll pitch in

@acl, Ah, good point. I know that but forgot about it

@acl not fantastic... Game changing! Look here: readwrite.com/2012/03/12/… . So, I was waiting for a better integration on M9... But when I see the Enterprise CDF Deployment strategy, I honestly don't believe we will ever have Mathematica integrated into other technologies.

I have several times wondered how many Player pro licenses were sold, and how it would compare with a free strategy that would second effect as more people knowing Mathematica, and buying Mathematica.

Yes, I think they're missing out on a ton of opportunities by not having a more open platform

Just imagine where Flash would have gone if we needed to pay for the player...

Their model works fine for enterprise and academia, but it seems like with W|A, 8, and 9, they've been trying to cater to a much wider audience. I don't see that working out very well without a truly open platform and a different monetization strategy

@rcollyer I got to 165 pages all in. I thought you were handing in yesterday? Or there is not so much of the "do it NOW or we will throw you out without a degree" aspect where you are?

@acl I have never used BibTeX. My interest was Endnote. I'm not sure why this would not be in the Mac version. Endnote make a Mac product and I actually tried a Mac prototype of this stuff as far back as 2009.

@MikeHoneychurch right. I was just asking whether this worked in v8 as I thought it did not (and can't see either endnote or bibtex working now)

Also, my main question is, did the things I show done with the AuthorTools above (making TOCs etc) work in v8?

@P.Fonseca I think what it meant by that article is that he has an expectation that other companies will come to Wolfram.

@P.Fonseca For integration into other technologies you wouldn't use Player Pro or CDF. You would link the kernel to whatever the technology developer was building e.g. if you made some commercial scientific equipment the kernel could be linked to the proprietary software that drive the equipment and analyzes the data. People tend to leave the company before those relationships get developed to the point of being implemented.

... As a separate discussion Player Pro doesn't make much sense for third party app deployment but why do you see as the problem with CDF Enterprise? The end user just needs the free player.

@acl I only tried a couple of things in 8 (AuthorTools). Some stuff works some doesn't. I think the last development on this package was V5.2. My brain tends to archive data after awhile but I just kind of remember some problems and decided my needs were not strong enough to pull apart the code and figure out what updates were required. I think others have commented about it on this site as well.

@MikeHoneychurch yes according to the comments in the code, that was the last time stuff has changed seriously.

Dumb question: What happens in cases like this here where the OP answered his own bounty question? Does he loose and gain the rep for the bounty at the same time? Or does he only looses the rep he used for the bounty since the FAQ says The reputation you're using for the bounty is immediately and irreversibly deducted from your reputation.

@halirutan I don't know. But if we wait for 5 days we will find out.

You lose the rep for the bounty when you create the bounty, and there's no way to get it back. You also can't award a bounty to yourself.

@SjoerdC.deVries are you around?

@Acl On my iPhone now

@halirutan Yes, the lose it as Kiyura mentioned. Mods can refund bounties, but don't do so except in certain circumstances. In any case, he has more or less recouped his bounty amount in upvotes...

@acl ...and now on my iPad for some speedier conversation. How can I help?

@rm-rf Aha. So the bounty goes either to the accepted answer or to the one with the most upvotes, but never to your own.

@halirutan yes. If I remember right, he probably won't even see an option to award it to himself

@OleksandrR. I was trying to turn it in yesterday. Didn't work, still writing. My committee isn't happy and likely to give me a poisonous snake for the snake wrestling segment of the defense, but it will get to them tonight ... by mid-way through the night at least.

@halirutan Actually, let me clarify my "yes" better for the first half of the question: The accepted answer is not automatically chosen... the OP has to specifically choose one (and it can be different from the accepted answer). If the OP defaults (i.e. doesn't award the bounty), then the most upvoted answer (≥ 2 upvotes) posted after the bounty was set will be awarded half the bounty amount. In case of tie, the older of the two will get it. Failing these, no answer is awarded.

@rm-rf Thanks.. I just deleted my text, because I wanted to ask you exactly this but thought it's maybe a dumb question because the accepted answer is always the best one for the OP.

@rcollyer well, best of luck. Having to write 2 chapters in a day plus printing is hard work!

@halirutan There are cases though... for instance, you can place bounties to "reward an existing answer", so while they might have accepted one that solved their problem, they might bounty another because it helped them in some other way. Another instance would be where they accepted an answer that solved like 90% of the problem, but the hardest 10% was in a different one...

Hmm, I see. I think yoda rewarded my CodeHighlighter answer after I gave it.

It was a 500 bounty before and I got 1000 in total.

@halirutan You got two bounties... if you hover on the +1000, it says so (also recorded in the question's history)

@rcollyer Someday when we meet, and we have some beers on the table, you will have to explain all this. good luck!

Ah, yes. Now I see it. Anyway, I regret, that I did not look the answer to this up. It was the follow up to the Cell to LaTeX question. But I never liked to write things in DefaultKeyTrans and was to lazy to check it..

@rcollyer I dropped one of the chapters from my thesis, so I had only one results chapter. Then I dropped half of that. Then it got to the point that I simply couldn't write any more within the available time and dropped half again. That chapter was 90 pages.

@OleksandrR. :O... My entire dissertation probably won't be 90 pages.

@rm-rf mine got to 165 in total (total printed pages, I mean, including title page etc.). Which is short by the standards of previous theses in my group.

But, nobody wants to read an 800 page thesis. One of the professors in the group is notorious for having had a 2-volume thesis. Examiners were not happy.

Everyone (university permitting) should take the stapled thesis route...

Would be nice. Unfortunately I got all of my results one month into the fourth (final) year and immediately had to start writing the thesis, analysis of the results be damned. So I had to write and analyse at the same time and had no publication record.

@SjoerdC.deVries it will have to be a lot of beer because while I could use one now, it would not be conducive to getting this done.

@OleksandrR. a friend walked into his general exam (candidacy exam wherein you propose your research topic) with 100 - 120 pages already under his belt.

@rcollyer we don't have this here. But, yes, it is very nice when you can use something you did in the first 3 years rather than having to throw it all away as useless.

@OleksandrR. That sucks. Me, I just lost access to most of my data, and my alternative computer ate some more!

@rcollyer yes, I heard that... really sorry to hear it. Can't you have an extension for extenuating circumstances? I was so worried I have my data backed up on 3 geographically separated RAIDs and a portable hard disk.

@OleksandrR. truthfully, you shouldn't have a tremendous amount done going into that, just enough to justify to your committee that you have an inkling of what you are getting yourself into.

@OleksandrR. my adviser recovered it, but I was working through my back-up strategy at that point, so I'm still going through what was there.

@OleksandrR. already working on the extenuating circumstances: my wife broke her ankle while I was at the mma conference, had to return at the end of the second day. :(

Then, my data went kaput!

And, now I'm writing, or at least trying to.

@rcollyer I can honestly say I had almost no idea what I was doing until I started to write the conclusions. Then it all fell into place and I (thought I) understood what everything is all about.

@OleksandrR. I can understand that. I was staring at a couple of numbers for a week "because they just didn't make sense," and then they made sense and I felt I wasn't being dishonest when writing about them.

Gotta to run. I need to finish this tonight. No matter how thin it is ...

@rcollyer Okay. Well, very best of luck. I'm sure you'll pull through!

@SjoerdC.deVries I remember you mentioning that AuthorTools did not work (some time ago). Now, I tried it in v9, and I do not find anything broken. a) do you also see it on win? b) do you remember what was wrong earlier? I am trying to work out if it's worth my while to delve deeper

@OleksandrR. don't you write a transfer report?

@OleksandrR. that's nothing, I still have no idea what I was doing!

@MikeHoneychurch that's exactly why the technology shouldn't be build on the expectation that a specific relationship is needed (specially in the current fast paced society). Lets suppose that I'm adding some special functionality to the CAD system my company uses. It's clear that it is not a case for a Player (the GUI will be added to the CAD interface...). There are 100 users of the CAD software. Should I use Mathematica for the calculations needed?

If the compiled library had access to a free runtime kernel, it would be pretty easy. Since it doesn't, it means I have to see with WRI what is the best way of doing it in order to avoid installing 100 licenses; but for that, it needs to justify the contractual and technology complexity (justifications on why that particular partner/technology, what are the alternatives, what about the future, etc),

and most of the time it doesn't (in my cases, 100% of the time, since they are small apps that cost only some weeks of work). So things are done without WRI technology, and Mathematica remains unknown.

@P.Fonseca I think we are talking apples and oranges. What i believe the context of the article is about is OEM type agreements. So a e.g. Bruker, Perkin Elmer, Philips, Seimens etc. uses the kernel in stuff they sell. Hence relationship development with those types of companies.

@acl no, as there is no transfer. One project (nominally) from start to finish.

...or Autodesk but last I heard they were developing their own stuff rather than use Matlab, Mma etc.

@OleksandrR. the "transfer report" (where I was at least) isn't about two projects, it's about somehow indicating that you've made some progress. it's just a formal thing, a thesis and a mini viva with two examiners

@acl my university mandates a dedication. I didn't want one because I didn't think anyone would appreciate being associated with my shoddy work, so I had a quotation instead, and I chose one by Karl Popper. That helped to keep me honest with respect to making non-falsifiable hypotheses and always arguing a point from several perspectives to make sure that it is internally consistent.

@OleksandrR. I had a quotation at the beginning of each chapter

@acl oh, yes, I see. Yes, I had those. One each year, in fact. But of course if you feel the project is going nowhere and change your direction after writing those then they are of no help at all. Initially I was working on mass spectrometry and my first report went into a lot of detail on the physics of the quadrupole mass spectrometer and the application of the Mathieu differential equation to it. It took me a long time to do but the only thing that came of that is my gravatar!

@MikeHoneychurch understood. But I don't develop Autodesk products. I use them (not from that particular one). And when I add things for internal use, I'm not in the same league as a Philips or a Siemens, and I always felt that for cases like mine WRI doesn't have a practical solution. How can I do it?

@OleksandrR. oh! well, I just had to do one after my first year, had a viva and that was it.

@P.Fonseca so u want to make something that links/integrates with a CAD package? Have you looked at Maple? They have had pretty good CAD integration for quite awhile. Come to think of it MapleSoft tend to do most things years before Wolfram ...the Wolfram implementation ultimately seems to be better though.

@acl seems to me that most people take these far too seriously. But you do have to give the impression of making progress, even when you know you aren't...

@OleksandrR. impression is literally the only thing that matters if you think about it.

(not just in academia)

no doubt you can think of particular people, but, on second thought, those people are just less good than others at hiding their weaknesses (or care less if they're not hidden). of course, many if not most also are good at what they do, but the important point is that they are perceived to be good.

but I guess all this is trivially true so why waste bandwidth

I was lucky that I had a very good supervisor. Both extremely intelligent and a thoroughly nice person... he didn't mind that what I eventually did was hardly related to what I was supposed to have been doing, not to mention an area (plasma physics) that neither of us was familiar with to begin with.

@OleksandrR. mine was similar. we wrote a paper together and then I switched to stochastic processes, of which he knew nothing. didn't ask to be a coauthor on the next paper (and did not read the thesis before submission, which I guess is a mark of either trust or gross negligence)

and extremely intelligent, yes! I thought everybody was like that until I went to do a postdoc elsewhere

@acl yes, I know what you mean. Most of the papers in my field seem to be pretty well meaningless, bordering on unscientific. If I had done work like that I would not think to publish it. But, everyone has pressure to publish, no matter what they've done or not done, so I suppose it's understandable.

@OleksandrR. It's just two different things what you have to and what you want to write.

it's not even that. the whole thing is set up so that the opinions of a few people, and a few external metrics, are the only things that matter. this usually works, but can backfire in a number of ways. but anyway, I don't see how else it could be set up.

This science thing today is a joke. Everyone is fiddling to publish small and meaningless things. I'm really annoyed by this (and by myself that I have to join this crap).

@acl I know instances of both explanations... my supervisor did not read my whole thesis before I submitted it, which was in that case a mark of trust since he is extremely attentive otherwise. But a friend also submitted his thesis without the supervisor reading it, which came after essentially telling the latter to shove his project and doing something else... I gather the supervisor still has not read it to this day.

@halirutan no not everybody, I only publish deep and important results :)

(it's meant to be a joke)

@acl ;-) And I mean of course not everyone but when you compare how many scientists we are and how many are joining this publish-perish thing..

@halirutan yes I know what you mean and agree. but then again in something like my corner of theoretical physics, what's important is more or less arbitrarily set by a small group of people, so it's somewhat arbitrary anyway. so I just set my own directions and don't worry about it too much after that.

guess I've been lucky in this sense

@MikeHoneychurch I'm not talking about CAD specifically (it just came to my mind as a clear example where a Player interface wouldn't be fitted). I'm talking about using WRI technology on small projects to be distributed to a certain number of users. The Flash comparison I made before seems a good one. Flash has/had a huge success even though the Adobe Flash builder ins't cheap.

I'm almost sure that if flash player cost 1$, no one would know about it, and adobe wouldn't have sold the amount of builders they did. Probably they could have sold it to big players, but history shows us what they would have missed. This is, IMHO, what happens with Mathematica. No runtime available (like MATLAB has) means a lot of missed opportunities (outside academics or R&D, very very few know what Mathematica is).