@acl How do you manage univ license when you're at home? The kernel has abruptly quit on me several times now because it couldn't connect to the server... this is while I'm on a bus back home or as soon as I connect to my wifi at home (I don't always use VPN unless I need to). It's annoying, because there's no warning as to when it's going to quit. In contrast, MATLAB gives me warnings before quitting, and when it does, it saves my workspace to a file so that my work isn't lost.

I just got sent a test notebook that showed V9 date and time speeds to still be 2 orders of magnitude slower than Excel and virtually no different from V8. Anyone else noticing similar?

comparisons were on a quad core mac pro.

@rm-rf I don't know, I use a home license on my laptop (and from what I understand, if I ask for it I can get a license from the institute that is not tied to the license server; but I could be wrong here, I didn't bother)

@acl I'm getting those, too, in my message window

@EricBrown does not seem to be happening always, but often

it's a notebook where i have copied text from safari into mathematica

i'm going to run a screencapture and send to wolfram

it also happened with code I typed in myself

and once after cmd-/, the cell with the commented out code got corrupted. when it happens again I'll send them the expression

StringJoin::string: String expected at position 1 in StringJoin[ExportString[XMLElement[math,{xmlns->http://www.w3.org/1998/Math/MathML,{wolfram.com/XML/,form}->StandardForm,{w3.org/2000/xmlns/,mathematica}->http://www.wolfram.com/XML/},{XMLElement[mn,{},{906609}]}],<<5>>,AttributeQuoti‌​ng->']].

@EricBrown that looks like what I had

i wish that i knew whether someone has reported a bug. it would save me a lot of time. Also, I wish that they could send out smaller bug fixes, but I also understand that could be mayhem

@EricBrown yes me too, both

@EricBrown are the errors you are finding in the docs manifesting as pink errors boxes?

@acl some of them. Others have errors about Machine Precision

e.g. IndependenceTest and friends I get: NumericalMath`RaisePrecision::precbd: Requested precision Automatic is not a machine-sized real number between $MinPrecision and $MaxPrecision.

Brett identified Ticks -> None as the culprit in one of them.

Not to rip on WRI -- Mathematica is my favorite program in existence. But I wish that they would have someone sit down and "Evaluate Notebook" on all of their examples, on all the supported platforms and see for themselves, rather than me writing, and them writing back and saying "not on my machine" and then me having to send a screencast.

I came across pink boxes in the v8 docs when 8 was released. It is unclear to me why Wolfram can't do a batch evaluate of all docs notebooks (temp options setting to saveable) and then do a batch screening for error boxes, messages etc. and output the incidence of an error with its cell id and then fix it.

Mike, you took the words right from my mouth

I believe that their QA is better than this, and they may have let a few slide to get out the 9.0.0 version. In a way, I'm glad the bugs are in the documentation, lest I "discover" it in my own code.

But the new stats tests are a disaster on the Mac.

Maybe @Arnout is the guy to discuss this with but I recall emailing tech support after the pink boxes in V8 docs and making this suggestion.

(* bomb *)

Anyway, this new version rocks. I love RLink, syntax highlighting, completion, and all sorts of stuff.

@acl I was consulting at El Salvador, with very bad Internet access and a tight schedule :) en.wikipedia.org/wiki/El_Salvador

Hope you had some good pupusas at least :)

@rm-rf Yeah, nice pupusas with loroco, of course :)

@rm-rf Have you been there?

@belisarius No, I haven't (yet). It was on my list during a trip to Cancún, but didn't work out. Some day...

@rm-rf "Pupuseras"

Now you're making me hungry

The pupusas are a piece of art

I love them

@rm-rf filled with cheese and jalapeño, they become ambrosia

Stop!

@rm-rf hehe

@rm-rf Sadly I spent only four days there, and very busy. Perhaps there will be a second trip in January

@rm-rf But not a place for the faint of heart :)

@belisarius If they have to have that notice up, that's a sure sign there are bigger problems... :)

@rm-rf Well, violence seems to happen on a daily basis in El Salvador. But it wasn't that way until some US educated salvadoreans returned home

Generally or some specific individuals?

@rm-rf Just a few gangs. Perhaps 10^4 individuals, although 2 10^6 salvadoreans are "studying" at the US.

That's a lot.

@rm-rf The money sent from expats to Salvadorean families is a noticeable part of a family income

I'll bet plenty of blood money makes its way back to the US too... through questionable channels.

@rm-rf Well, I haven't had the opportunity to ask much, but your intuition doesn't seem off the mark to me. In other countries (like Panama) it's so clear you don't even need to ask

anyway... have you gotten a chance to try v9 yet?

@rm-rf Nope. I was preparing my presentations for the trip when v9 was going out. I haven't download it yet. The worse part is that I'm quite overflowed, so I don't know when I'll find the time to try it out

Ah, you might want to wait then. I did waste quite a bit of time playing around with the new features

Time that would've been better spent doing real work

@rm-rf My time to "play" is quite constrained nowadays, and my current business needs are far away from Mma. Perhaps I'll just read the Q&A on the site and for now, and give it a test drive later

BTW, if you like fishing, google for fishing in El Salvador before going there

amazing beasts

Never fished, but it looks pretty impressive from Googling it. Pretty big catches, most of them (although, that's probably what they choose to show)

@rm-rf I need some sleep. The jetlag is still bothering me. See ya!

V9, WhenEvent docs, Applications, PowerElectronics, first NDSolve line (sol=NDSolve) doesn't work and exits my kernel here. Does that happen to you people too?

perhaps just a Table[f[pi,p(i+1),p(i+2)],...] is best

@ssch Developer`PartitionMap

@rm-rf That's what I was thinking of! Thanks!

@belisarius ah, well, good that you're back then

was getting boring around here

@Rojo yep ... I get a kernel crash as well :-(

This amazes me that the example from the docs are not part of the Unit tests ... I have found so many examples that don't work ... or give errors

@Rojo Crashes here too (Mac)

@Gabriel Definately. It shouldn't be hard to build unit tests automatically from the docs I guess

is this not a clear-cut "read tha manual" question?

@acl yes, pretty much

@OleksandrR. good, I was wondering whether I was missing some subtlety

by the way, what are you doing nowadays (having submitted the thesis etc)? straight back to work?

My supervisor is employing me temporarily as a sort-of postdoc. That is, I get paid as a postdoc with funds scraped together from various grants but have no official research project. What I'm actually doing is showing a new student what I was doing so that he can carry on with it if he likes, and writing papers on the results that I collected for my thesis. And various other bits and pieces as well.

@OleksandrR. So now you can include all those grants in your Cv :)

Yep :) Actually it is a nice excuse to work on a few different problems that are not related to what I was doing before.

how long is that for?

@acl For as long as they take to realize he is using the grants to learn V9 :)

6 months in the first instance although it was made clear to me that I can quit sooner than that if I find something else, or stay on for longer if I don't. So it's quite versatile.

@belisarius beta-test it is a more appropriate description :)

@OleksandrR. sounds good

@belisarius M9's ParametricFunctions are quite useful when fitting kinetic models. :)

@OleksandrR. what sort of kinetic model are we talking about?

@OleksandrR. I haven't had the time to start playing yet. Hope to do it soon (trying to hide my worse envy look)

it has some nice enhancements to the frontend, but template completion doesn't work (literally); so if you used that...

@acl this would be one describing the dissociation and reaction of various chlorinated organic materials in the condensed phase, as probed by transient absorption.

@OleksandrR. so we are talking about this sort of thing, right?

@acl yes, that's it. The complication is that multiple processes occur simultaneously so you need to decompose the spectra into something kinetically meaningful and fit a model with a lot of rate constants undetermined.

@OleksandrR. I know it can get complicated, I was trying to work out if you meant rate equations, or kinetic theory, or something else. but I guess since you're a chemist I should have guessed you meant this.

I remember van Kampen going on and on about this stuff in his book. I sort of glossed over them at the time I was reading it.

@acl oh, yes, sorry. I feel stuck in the middle of these two fields... when I use physics terminology it confuses the chemists, and when I use chemistry terminology it confuses the physicists...

Speaking of decomposing the spectra into kinetically meaningful results... anyone bothered to implement independent components transforms in Mathematica? If so it may save me a bit of time doing it. Currently I just use the principal components transform and decorrelate the result.

@acl Another issue with the template completion — completing the template for a function in a context not on the path will remove the context prefix

For instance, try completing for Developer`PartitionMap

@rm-rf works okay for me

@rm-rf behaves as you say here (ie is broken)

@OleksandrR. Hmm... Wonder if it's another one of those Windows/Mac things... mine simply changes to PartitionMap[...], which won't work

Oh, yes, sorry. I misunderstood what you meant. You're right, it's broken.

Huh... apparently, this is how it was in v8 too. Just never noticed it

Another thing: what is supposed to happen when you push the "send feedback" button? For me, it does nothing.

@OleksandrR. It opens up your default mail application (if you have one setup) and fills in an email to foo@wolfram.com and the command you just typed

You just need to add your suggestion and hit send

Ah, that explains it. I don't have any default e-mail program. I had expected it would open up the web form you get when you pick "send feedback" on the help menu.

@acl actually trying to SaveDefinitions a ParametricFunction crashes the kernel. So, yes, in a very real sense this is beta testing too...

@OleksandrR. I wasn't joking

@acl seems to me that the n.0.0 versions are always quite buggy. This is just the first time that we've had a chatroom to complain to each other about it. :)

@OleksandrR. I don't remember v8.0.0 being so buggy (or at least, the things that were broken were not so obvious to me)

@acl truly. I find v9 almost unusable ... whereas I used 8.0.0 without any major problems

does anybody remember what version 7.0.0 was like? This upgrade has really put the fear in me

@Gabriel 7.0.0 had real problems doing a Fourier transform

really ... ugh

must be a strange build/test process at wolfram then I guess ...

I also recall 6.0.0 was terrible, but it eventually got ironed out

One thing that never got fixed in version 7 was that the Method -> NMinimize option for NonlinearModelFit didn't work. That was fixed in 8 and I was glad of it.

okay good to know. Basically x.0.0 seems like a beta release then. Are most things fixed by the x.0.1? As I noticed some people said they needed to pay to get 8.0.4 ...

In the past the n.0.1 release has been a significant improvement, yes. The people who had to pay for 8.0.4, unless I'm mistaken, would have been those who bought 8.0.0 with Premier Service and didn't renew it for the release of 8.0.4 a year later. Although the status of the Home Edition versions I'm not sure about.

@OleksandrR. I got a home license for v8, then the upgrade to 8.0.1 was OK but I did not get the upgrade to 8.0.4 (although I could have got it via a premier license)

the people who came up with the upgrade rules are probably the same who thought that producing a pdf saying mma is 20000 times faster than maple is a good idea

or at least have the same worldview

@acl lol

those documents make me sad ... I love mma so much, but the press is so painful to read

@Gabriel that particular pdf is cringe-worthy

@acl which one is this? I don't think I saw it. Who started it? (Because Maple is going on about how fast their polynomial manipulations are in comparison to Mathematica as well.)

@OleksandrR. I am pretty sure mathematica started it ... saw that polynomial graph, which follows the same format for the next maple upgrade

guess stuff like this turns into an arms race of propaganda

@Gabriel it's all silly, I agree. But I suppose we are not the people they are trying to convince with this.

@OleksandrR. this

true ... but it really doesn't seem necessary. Does matlab does this type of thing? Mathematica press tends to Hyperbole way, way to quick I think.

@acl ha. What hurts me is the legends they are forced to use ;-)

@Gabriel ha yes

though when I was a big python user, even that community which has no marketing department suffers from some insane claims about how great python is, and how totally, utterally, laughingly crappy everything else is. Always hurt me to have to read that crap

I haven't seen the python community claim anything of the order of "in-line interface to wolfram alpha is a revolution in computation", though

but anyway

@acl touche .... that blog post was beyond crazy. Put a smile on my face....

though I have to admit anything that Stephen Wolfram himself seems to say suffers from a touch of the crazy

@acl TBH that is not that bad. Maple seems to have been asking for it with their claim that "Maple 15 is second to none in terms of scalability and performance." And these operations are ones that many people would consider useful (for example, knowing that Maple's sparse linear algebra concerns just the representation and not the algorithms is quite important if you care about that).

@acl on the other hand, the page that it's linked from is simply offensive about Maple, calling it poorly integrated, shallow, cheap, and a dead end system. That's not acceptable IMO.

@OleksandrR. also I would take claims in such comparisions about the capabilities of maple with a grain of salt. It would really blow my mind if the representation vs algorithms was true. Why implement the representation at all? Just for for space ... until you need to use it ... then you lose all value?

@OleksandrR. yeah, so let's accuse maple of forcing us to use special datatypes for speed. the only difference with mma is that in mma it's undocumented :)

anyway I like mathematica (obviously), I just find this sort of thing offensive.

@OleksandrR. oh man ... that linked webpage is amazing ... the cartoons ... the text ... I was not ready

@Gabriel yeah, I don't know. Ask Jon McLoone, I guess. I have no experience with Maple to be able to tell whether the code used for comparison is reasonable or not.

@acl so true. I really, really like mathematica, it is my favorite environment I have ever used ... but I can't stand the crazy comparisions

true ... my experience is that code comparisons are never fair ... I have seen to many in my day where the person writing the code is an expert in one system, and barely literate in the other. Many Scipy/R/matlab comparisions come to mind

@acl haha, yes. I think we've all got to the point now that we use undocumented features out of habit. It came as a shock to me when someone (possibly you?) recently mentioned that the proper usage of packed arrays is not documented.

of course, but there's a difference between it being a blog post and a document produced by a largish company. I accept that it will be designed to present its product in the best light, but come on

@OleksandrR. yes, search the docs :)

(well actually there is info in the docs, but I think it was not there in 8)

but it would have been nice to have some document telling us how effectively to compile etc. of course, v8 and later are much, much better than earlier (eg, the compiler is documented now)

and I guess it's better to have undocumented functions than not to have them at all.

maplesoft.com/products/maple/new_features/sparse_matrix.aspx

this suggest they did something wrong. I wonder if they way they are making the matrices (in a for loop) somehow breaks the sparsity

@acl most useful information about the compiler is still undocumented, such as what functions are compilable. I don't mind too much if a feature is undocumented because it's usually not too hard to figure it out, and much of the documentation is rather shallow anyway... but it's not attractive to new users

@OleksandrR. it's shallow once we've spent a few years digging into the undocumented details. imagine someone handing you this huge thing and you trying to figure all this out from zero.

and as far as I know, things like CompilePrintdidn't work in v7 and earlier, so you had to work out what the bytecodes meant

and this changed from version to version (or anyway it changed from 5 to 6)

here is the maple propaganda ...

maplesoft.com/products/maple/new_features/…

@Gabriel hmm... looks like that may be the case. The NAG libraries should be pretty good.

Hey, can someone test this in V8?

With[{d = RandomReal[{-1, 1}, 10000]}, Dynamic[Print@3; Histogram[d]]]

In V9 I'm getting constant updates

Seems like Histogram forgot to protect some internal symbol it uses from triggering updates, and I need to use explicit TrackedSymbols...

@Rojo I get only 1 3 in v8 and 2 in v9

Humm

oh I see everytime I try to close the window in v9 it pops up again

Hehe

Make the messages window bigger

@Rojo I don't want to ... all these bugs in v9 make me sad :-(

Yeah, but let's just wait for 9.0.1, and be friendly with Wolfram. I'm sure one day they'll appreciate that and give us lifetime premier service for free.... or not

I'll just use safeHistogram = Refresh[Histogram[##], None] &

but it's sad, yep

@Rojo yeah v 9.0.1 can't come soon enough. Love all the new features ... when they work

sadly I'm all in at this point ... Mathematica for my forseable future ... warts and all ;-)

maybe they jsut need to change there marketing to mutts ... "all mathematical software sucks, ours just sucks less ..."

Nah

Marketing people have no shame

I've already given up on that

@OleksandrR. Wow... that oozes with butt hurt

@rm-rf ?

The mma vs maple web page and several linked pdf files

ah ... yeah that webpage was crazy

Looks like it was an intern project...

so if I want to play with this, I need to copy each cell individually by clicking on it, clicking inside, pressing ctrl-a etc? Or am I missing some way to copy the whole thing at once?

@Rojo V8.0.4 on mac I only got one "3" printed.

Thanks!

I reported 3 things already today...

I have been holding off installing V9 but am starting to think i might wait until 9.0.1

Overall I'm still sticking with v9.0.0 versus v8, but there are clearly issues

But at least the plot legends, I have already used them too much to go back

yeah but I already have several code blocks for all the types of legends i use -- 9 offers convenience but I can already produce those legends

On the contrary, I have been simply avoiding legends :P

before v9

If date and times speeds were fast I'd switch to 9 straight away

I've read that you seem to use those much. I barely ever needed the date functions

Most of my routine usage involves date and time series data

I see

I think Wolfram fail to appreciate the importance of speed since they probably view usage as prototyping rather than deployment. For prototyping slow date and time calcs are ok. For deployment it is a deal breaker IMO.

The room for improvement is infinite, they try to cover so many areas

personally I would like them to get what they have right before moving on to other stuff. We still cannot do all the plots that Excel can do!!!

...the slow date and time calcs also show up when you do Dynamic DateListPlot. If you have a custom tick function you can convert to absolute time and use ListPlot and render your plots nearly an order of magnitude faster. You notice this difference if you make the plots dynamic

it would also be nice if outstanding bugs were fixed with each new release. Some linger for several versions.

That's true

I only checked up on a bug I reported some time ago on a FE memory leak (there's a question in the site) and it wasn't fixed. Same goes for Unique symbols created in Initialization not being tracked. They need more people. They should hire all of us

re: Mathematica vs Maple. Please don't blame marketing people -- if there are any left -- at Wolfram for the hubris.

@Mike fair enough :)

(I just lost a cell due to lack of undo :) )

These sort of feature comparions are basically a pointless game in any case. I have never met anyone that switched. The switching costs for established users are huge so as long as the software you use does what is needed why would you switch? There has to be a compellling case made -- meaning the opposition have to provide something extraordinary and none of the products in this field have that big an advantage over others.

@acl Ctrl+s ctrl+s ctrl+s ctrl+s

So it comes down to getting new users ...