« first day (912 days earlier)      last day (3429 days later) » 

12:00 AM
@hwlau right L3
the cache in AMD works more "separately" than intel
well either way, I'll check tomorrow. But this seems to be happening at too high level to be that. But who knows
@hwlau well I have no idea, I don't really know much about physical layer stuff :)
But this physical layer really affect the performance
@hwlau Yes I understand. But there is an infinite number of things to wonder about, and one has to be selective
@acl Ya, the mathematica also add a layer there, so it is not transparent. I don't usually have problem when I write C++ program.
But the scaling of large system should always correct, so there is no problem of estimation.
12:07 AM
@hwlau Neither do I in terms of performance (usually), but I consider writing C a problem :)
@hwlau OK I don't disagree of course. That's the main problem with mma, scaling things up needs experience (and may not be possible)
@acl BTW, is it easy to pass vectors and matrices to a manually compiled C code program, and return them back?
@hwlau reasonably, although I don't have as much experience as eg szabolcs and halirutan on this (I just played with it, wrote one piece of code and then just use that since)
@acl I posted this:
Q: Menu bars on Palette windows don't work properly -- any workarounds?

Mr.WizardAfter getting help from Rojo with Menu commands on Palettes and other windows without menu bars? I was able to create a Palette with a menu bar by: Selecting the Palette and opening the Option Inspector with Ctrl+Shift+O Setting Editable -> True and WindowElements -> {"MenuBar"} However the b...

@Mr.Wizard Yep, that is what I had in mind
The real conundrum for me is how one is supposed to easily edit a Palette as I can't figure out how to use Generate Notebook from Palette in v10. (Windows.) Have you figured that part out?
12:18 AM
In any case for now I use Notebooks[] etc
@acl How much are the different between the generated C codes from mathematica, and manually written C code?
@Mr.Wizard it does seem to work, but as I have never used it before, I don't know what it's supposed to do. it literally generates a notebook with the buttons in it (and FullForm works on them)
@hwlau hm, hard to say, but you can look for yourself at the generated C code (it's explained in the docs)
@hwlau paste this into the doc centre
@acl If you've never used it before there must be another approach that I am somehow blind to. If you have a palette that you wish to modify and save, possibly as a copy, how do you do it?
@Mr.Wizard I never modify my palettes. I have notebooks which produce them, and modify those, then save the resulting palette
I didn't even know you could modify an existing palette until you mentioned it just now...
@hwlau search for Inspecting Generated Code
@acl Do you not have the Generate Notebook from Palette menu command?
12:25 AM
@Mr.Wizard I do. I just noticed it now that you mentioned it. I never use menus if I can avoid it, so had not noticed it.
The codes can be understood in my eye. They are splitted into so many piece.
@Mr.Wizard What it does here is this: for a palette with 3 buttons, it produces a nb with the three buttons. FullForm works ok on them
Also the generated codes doesn't seem to allow some flags
compile flags
@hwlau it's not really meant to be human readable I guess
@acl Die-hard old-school? Keyboard shortcuts over mouse I take it? I admit that is faster for those who master it.
12:28 AM
@Mr.Wizard yes, die hard
@Mr.Wizard OS X is nice in this regard. You can assign system-wide shortcuts, assign key board shortcuts to arbitrary menu entries in any application, etc
@acl What OS are you using and when is that command active? Or perhaps I should say where since there is no menu bar on a palette?
also, the basic emacs keyboard navigation shortcuts work pretty much everywhere
@Mr.Wizard This is on OS X, where the menubar of the active application is always at the top of the screen, not attached to any particular window
It does sound easier under that OS (which you just answered). Perhaps some day I'll rise to that level of proficiency in Mathematica.
@acl I want that
12:30 AM
@acl Okay. It's got to be a Windows problem then. Thanks.
@Rojo stick to your vim
@acl I meant the OSX thing
You can keep your emacs
@Rojo hmm, turns out vim comes installed with os x (I use emacs so hadn't noticed). so now I started it
how do I get out?
ah! never mind
@acl Unplug the PC
@Rojo +1
12:32 AM
@Rojo :q
I'll print it and put it on my office wall just in case
@acl don't just click the cross :)
@hwlau what cross?
@acl close the windows
@hwlau ah
It is not the right way to get out, not much better than unplug :)
12:37 AM
@hwlau depends. there is some satisfaction in extinguishing the life out of a vim instance
@acl I think most people do the same thing they enter vim for the first time
@hwlau it's not terribly intuitive.
3 hours later…
3:22 AM
Hello all. Is it only me on Linux or is it no longer possible to copy&past cells from one notebook to the other without loosing the cell structure??
3:44 AM
@halirutan I just copied and pasted a cell properly. Do you have an example?
@halirutan By the way, my f%%2#$% $MachineID changes some times with the Linux kernel updates, which is %#$% annoying. Did that ever happen to you?
@Rojo No. Never.
@Rojo Aehm.. It doesn't happen right now but it happened some minutes ago.
@halirutan Oh, ok. I just had to boot with an older kernel to have mma active
@Rojo This should not happen. Didn't happen for me. I would definitely complain about that at Wolfram because the kernel updates every few weeks for me.
@halirutan Yes. 2 licenses. I can't use the same on Linux and Windows on the same machine. At least I ask to be able to not need both for just one Linux, hehe
4:38 AM
@halirutan I have been having all sorts of difficulties with copy/paste on Windows
It's pretty annoying
@mfvonh Me too on Linux. When I use the normal Google Chrome and I copy code from Mathematica to Chrome, I get additional bytes.
This is the cell I want to copy
and this is what arrives in Chrome
Pretty neat. Now I have to use Chromium to circumvent this.
@halirutan Weird. That one works fine for me. I have been getting frustrating behavior where sometimes it gives me box structures when I copy an expression that shouldn't have them (like just a list or something). But it's not consistent. I'm having to do Copy As -> Plain Text a lot
And then sometimes when I copy it will try to turn the notebook into a template notebook, or it will open a new testing notebook. And then sometimes when that happens Ctrl + C will get hooked to creating a new testing notebook.
I have to shut MMA down entirely
I only started using MMA in v9 so I haven't been through an upgrade before. I guess it is weird to encounter several rough patches all at the same time in a system I like so much
5:08 AM
@mfvonh For this you can use Ctrl+Shift+C, so you don't need the mouse.
(Basically, I always use Ctrl+Shift+C when I want to paste something to a post or here into chat)
5 hours later…
9:44 AM
A: Schroedinger eigenvalue problem in two dimensions (Harmonic Oscillator)

JensIn order to give one possible answer, I'll just take the isotropic harmonic oscillator in 2D and do a finite-difference calculation by discretizing the xy plane with constant spacing a. Here is the construction of the resulting matrix for the Hamiltonian, h. I assume the origin of our spatial g...

Jens' answer deserved many more upvotes... But it's probably hard to appreciate it without knowledge of the problem he's solving
Many more upvotes, in fact.
I mean it has as many upvotes as straightforward applications of MapThread I've posted. Come on.
10:06 AM
A quick question: Suppose I want to set up InputAutoreplacements for some notebook, and do it like this:
 InputAutoReplacements -> {"url" ->
what this does is let me type url and have auto-expand, placing the cursor in the first argument of Hyperlink. This way I can type a label, tab into the next argument and type the url
now, what if I wanted to have an actual label appear instead of the filled square? How could I do that?
I mean like the front end does here:
this must be simple
Or does nobody actually use this stuff?
3 hours later…
1:33 PM
@acl Hello
It looks like the boxes are TagBox[ FrameBox["theLabel"], "Placeholder"]
when solving a differential equation using DSolve with no initial conditions, should not the answer always contain a constant of integration?
InputAutoReplacements -> {"url" ->
RowBox[{"Hyperlink", "[",
RowBox[{"\[SelectionPlaceholder]", ",", TagBox[
"Placeholder"] }], "]"}]}]
@Nasser ideally
@Rojo thanks
@Rojo fantastic! thanks.
@acl humm... so if DSolve returns an answer without C[1], is this a bug or not?
:16655081 You've got no answers to your transvestite-cursor question. You should learn to sing.
1:37 PM
For a first oder differential equation..
@Nasser I suppose. what's the equation?
@acl, good. so I can post it on main board. Here it is:
DSolve[D[y[x], x] - y[x]^2 + y[x]*Sin[x] - Cos[x] == 0, y[x], x]
{{y[x] -> Sin[x]}}
I'd like to know why M does not return C[1]
@belisarius I will wait another couple of years before I give up on that answre
Isn't solution without C[1] a particular solution? and the answer should be general solution?
@Rojo I tried that on my machine and it works.
1:45 PM
@Nasser I don't have any experience with DSolve but yes you're right, there is a constant missing
@belisarius What, the waiting or the suggestion by mwnwwmw on the comments?
@Rojo Regretfully I can't copy my singing as an answer
@acl thanks for your input. Will post it on main board to give chance for someone to see why it is missing in this case.
@belisarius You can
@Rojo I still have a little drop of decency remaining.I won't.
1:49 PM
@belisarius I'll ask again in a few months then
@mfvonh I am happy to answer questions about my answers and methods. Don't hesitate to ask, unless you learn better by working through it alone.
@Mr.Wizard Wizard, about that step thing. I haven't posted another approach because I think perhaps it isn't useful given yours
Meaning, yours advances one step, either argument evaluation, changes because of attributes, or iteration. What goes straight to the next iteration
@Rojo Do you care to post it here in chat? I'm interested. What was your target application and behavior?
@Mr.Wizard Not much, I think I did it when that question was posted, then didn't feel lke posting it OR didn't have the time to try and fix all the possible errors that might come out from people inspecting it, so I didn't.
@Mr.Wizard 1 sec, I'll paste it here
I'll no longer will help in formatting user's code, it is waste of time. Today I did this for someone, then few minutes later they updated the question, erasing all the edits I did when they reposted the code again :)
2:05 PM
@Nasser That's your prerogative. I am inclined to roll back the edit when that happens and ask the OP to try again, this time respecting the formatting. I think I've only done that once or twice however.
@Mr.Wizard Sorry, work stuff. Back. This is what I've been using (barely)
SetAttributes[{Step, stepIterative, stepTrace}, HoldAllComplete];
Options[Step] = {"Attributes" -> False};
Default[Step, 2] = HoldComplete;
Step[code_, (wrapper : Except[_?OptionQ]) // Optional,
   OptionsPattern[]] :=
    stepTrace[code, wrapper] /; OptionValue["Attributes"];
Step[code_, wrapper : Except[_?OptionQ] // Optional,
   OptionsPattern[]] :=
    stepIterative[code, wrapper];

(* MrWizard *) (* Ojo que así como está aplica el wrapper cuando no \
hubo iteraciones *)
stepTrace[expr_, wrapper_] :=
@Mr.Wizard That goes to your version when the option "Attributes" is set to True, and by default to mine. Mine just uses $IterationLimit
to force it to stop after one iteration
(there was an extra Print. Corrected)
f[x_] := g[2 + 3];
g[x_] := x^3;
Step[f[2 (3 + 4)]] gives HoldComplete[g[2 + 3]]. Yours gives HoldComplete@f[14] as a first step
As it is, HoldComplete is the default wrapper. You can put any wrapper as second arg
@Rojo Thanks. I'll enjoy looking at that later. Apparently I'm too tired right now as it's not making sense. :^)
@Mr.Wizard It just sets IterationLimit low, iterates needlessly a few times, and then runs the code,w hich is stopped because it hits the limit
Okay, that makes sense I guess, but I never considered such a thing. See, I'm glad you shared this. :-)
@Mr.Wizard :)
A nice wrapper (2nd arg) could be something like this. Function[Null, Interpretation[HoldForm@#, Step[#, #0]], HoldAll]
that lets you evaluate the output time and time again to get the next step
2:15 PM
@Rojo That sounds similar to this:
If you change the return value from # to # /. HoldForm[x_] :> Defer[step[x]], then the resulting expression is in a convenient form for further stepping. One can repeatedly type SHIFT-CONTROL-L, SHIFT-ENTER to step through the surface evaluations until the expression goes inert. — WReach Feb 8 '12 at 14:10
@Mr.Wizard Exactlyyyy
One more thing to play with. I do need to go however. It's been a long time since we chatted; I should do something about that.
@Mr.Wizard :) We'll talk soon
Good. :-)
2:28 PM
@Mr.Wizard Thank you :) I knew I could ask, I was just trying to figure out what I wanted to ask. Which is: I thought ought to be able to nest more P = corresponding to the TraceDepth setting (so I could go deeper, assuming for now the expression is complex enough), but that does not seem to be the case.
Or maybe I am not implementing it correctly
I don't actually need that functionality. I was mainly testing to see whether I understood how it worked.
3:13 PM
@Oleksandr Do you have a wireless mouse or keyboard? It's the new suspect for the M10 startup slowness.
@TaliesinBeynon I think I have found a Dataset issue. ds has a Dataset, and it changes after DumpSave["file.mx", ds]. I'll try to find a minimal example because this dataset is 1.5GB
Not just the symbol being exported, but others if they are sharing the memory it seems
This is what happens when everyone just looks at the Mma documentation for the definition of a term. Three different people telling the OP that his matrix is not an incidence matrix. There's no standard definition of incidence matrix, and the most common way it's used matches neither Mathematica's definition nor the OP's: mathematica.stackexchange.com/q/55190/12
3:28 PM
@Szabolcs the real problem seems to be OPs who apparently cannot conceive of the fact that not everybody has the exact same model of the world in their head.
Although this is not restricted to this site.
I mean, you'd think it obvious that when you ask a question, you either link to definitions of the terms you're using or define them. But no.
@acl It happened more than once before that people complained that some distribution is not giving correct results in Mathematica because the results differ from R. In all cases the distributions were just parametrized differently or the parameters just came in a different order, but they never bothered to check that in the docs ... (either R or Mma).
Same with the incomplete gamma function between Mma and MATLAB (again slightly different definitions for the same name).
@PatoCriollo Nice new gravatar :)
@TaliesinBeynon, small example (it turns out it's not Dataset related)
Can anyone try this?
fname = FileNameJoin[{$TemporaryDirectory, "test.mx"}];
as = <|"a" -> {<|"b" -> 0|>}|>;
DumpSave[fname, as];
Hi @Undo, what brings you here?
@Szabolcs Thank you! Missing my mate.
3:35 PM
@Szabolcs Ok, same bug
W Community is so bloody slow it's near unusable. The other day I measured how long it takes to load, it was 21 seconds.
@Rojo There was also this
@Murta see you at the Conference. Anyone else attending?
@Szabolcs I don't use it, I couldn't find a "workflow" yet to use it
3:52 PM
@PatoCriollo WTC?
These two are different. Beware
<|"a" -> 0, "b" -> {1, 2}|>[[;; , 1]]
<|"a" -> 0, "b" -> {1, 2}|>[[All, 1]]
4:11 PM
@Taliesin The operator form for ReplaceAll isn't documented either. I really like these operator forms, as I find code written lienarly in an all-prefix or all-suffix notation much easier to read. Eithert read it left-to-right or right-to-left, ne noeed to jump around.
@Szabolcs Yes!
4:36 PM
@Szabolcs Ping me when you discover those things :)
4:51 PM
What's going on here? -->
Q: Why does ImportString["1c", "Table"] eat the letter c?

SzabolcsI have some whitespace-separated matrix data that I read with Import[..., "Table"]. The data contained mixed strings and numbers (the strings are for header names). I noticed this weird behaviour: ImportString["123c", "Table"] (* ==> {{123}} *) Mathematica ate the letter c!! Why? It doesn'...

Why does Mathematica eat my c?
@Szabolcs It's importing it as a number. Perhaps in some format it interprets the c to mean "base 7", "triple precision" or whatever?
@Rojo It's interpreting it as a currency ... just found it ...
"Part 139957468142560 of {<<1>>} does not exist"
When on earth did I ask for that part, haha
1 hour later…
6:28 PM
What file formats are best for storing tables with named columns and rows? The priority here is portability, i.e. I'm looking for something easy to read with many tools.
6:47 PM
@Rojo ah, well spotted. I've reported this bug.
@Szabolcs @Szabolcs agreed! Operator forms have turned out to be a huge win for writing readable code. Unfortunately I can't remember whether it was Stephen or me who first suggested them, so I don't know who should get the credit :). Either way it was a major (and risky) decision, and I had to argue with a lot of people in the company who remained skeptical, so credit goes to Stephen for just pushing it through.
@Szabolcs but they were motivated by the needs of Dataset's query language, which is an interesting historical detail I think
@TaliesinBeynon Can I rely on ReplaceAll as an operator for long-term-maintained code? It's not documented yet, which might be an omission or the syntax/behaviour might not be finalized yet ...
Replace is documented.
@Szabolcs Yes, I'm updating the documentation as we speak.
@Szabolcs another one that is surpisingly useful is Extract, because you can do Extract[5] for example, which is a 'getter' that gets the 5th part of something
like a lens in Haskell
what i would like to do is generalize this to a form of Part that is specifically an operator form, call it "Parts"
then you can write Parts[All, 1, 1] @ data and get those parts
Haven't tried Extract yet. I'm still discovering the best ways to work with the new functions and syntax.
and even better, you can write MapAt[f, Parts[All, 1]] @ data and now you are applying f to the first column. Of course you can do this with MapAt[f, {All, 1}] already, but the point is that we would support Parts[...] wherever we currently support a level spec
@TaliesinBeynon Isn't there a risk of confusion there? Like Count and Counts ...
7:01 PM
@Szabolcs unfortunately, Parts is exactly the right name for a whole bunch of reasons. Plus, people don't usually write Part, whereas they would have to write Parts, so it may not end up being confusing in practice
Regarding Part, what do you think about adding a keyboard shortcut for typing \[LeftDoubleBracket] and \[RightDoubleBracket]? They're much more readable than [[ and ]]. One of the first things I do after installing Mathematica is add this keyboard shortcut (Command-[), but techincally this isn't even allowed by the license (the installation directory shouldn't be modified)
Sorry, in the meantime I need to clean up some data, my prof already asked twice for it ...
General question: is there an optionto quote strings when exporting to CSV? By default they're exported unquoted.
@Szabolcs You should ask John Fultz whether he is willing to do that, I can't inject new keyboard shortcuts willy nilly. Also, I have no idea about quoting strings when exporting to CSV. I mean, it's not hard to write your own CSV exporter that does whatever you want, but you know that already.
@TaliesinBeynon That second one was really a question to anyone in the room. There are workarounds, I was just looking to save time :-) It's the "TextDelimiters" option, which doesn't seem to affect the "TableHeadings".
@Szabolcs we should have a StringEscape["foo\"bar"] that will change the " into a \", so that you can write StringQuote[x_] := "\"" <> StringEscape[x] <> "\"";
@TaliesinBeynon that's exactly what I did as a workaround
7:14 PM
@TaliesinBeynon I think many people end up with versions of that in their init.m after a while
@Szabolcs @acl that reminds me: we now have an official "incremental language development" group that works on continuous improvements to the language. not as big as operator forms or associations, but general cleanup, missing functions, etc.
@TaliesinBeynon that sounds like a fantastic idea
@Szabolcs @acl so of course I would be interested to hear what the community thinks are the most pressing "missing functions" in Mathematica/WL
also I should mention that the new stuff in v10 (Associations, the new operator forms) are really, really nice
@acl thanks! i think so too!
7:17 PM
@TaliesinBeynon Oh, that's a dangerous question to ask here! You may get too many replies!
(the extra library type stuff is also nice of course)
well, what i would suggest is: you each have 2 functions to suggest, so make them count!
choose the things that you really think are most general and that if you lost your init.m you would most miss
i mean, i have dozens of things
but being honest only a few of them are really important, relatively speaking
i mean, we are already going to do simple things like StringDelete, so don't suggest that
@Taliesin But really: what would make me the happiest is to have a release, call it version 10.5 if you like, which focuses solely on fixing bugs, adding polish, improving usability, but adds no new major features. Study how people use existing functionality, what are the most frustrating limitations or problems, and fix those.
in fact, anything in GeneralUtilities i won't count, although it would be nice to know what people think is useful there
hm, this reminds me I can clean up my init.m. it has stuff like getKeys and the rest of my schoolboy implementation of Associations
7:20 PM
@Szabolcs yes, I think you've hit the nail on the head.
it is afterall what Apple does from time to time with their releases.
@TaliesinBeynon Yes some bug hunting would be nice :)
unfortunately i know Stephen too well... there are so many more features coming. a bug-fix only release is probably a year or more away, even if I convince him
we're going to add NLP, for example
and natural language parsing
and a bunch of visualization stuff
text search
graphical models
representations of web pages
web crawling
@TaliesinBeynon as in network models?
that'll keep us busy for a while
@Szabolcs as in probabilistic inference, bayes nets, this kind of thing
that's a more long-term thing, though we're starting to think seriously about it
I think my number one request is still a good Python interface, for calling Python libraries from Mathematica (not vice versa).
7:23 PM
oh god yes, i forgot that
But that's a major feature, not a simple function.
well, if we're going to do python, we should do it properly. PythonLink, virtual-env like functionality, etc
are there not any unofficial projects that could help you in the meantime?
you should know of any better than most
there's Pythonika, but it's not very good.
what it does well is transferring some basic data structures, such as large integers
The library I wanted to call the most was igraph. It also has an R interface (which is actually better maintained thanthe Python interface). I am using it through RLink now.
@Szabolcs what does it do?
let me reprhrase, what do you use it for?
@TaliesinBeynon You mean igraph? It's a library for handling graphs (networks). The functionality is very similar to Mathematica's, but it has some functions Mathematica doesn't have. I'm not particularly happy with Mathematica's graph handling, it's sometimes inconsistent, not well documented, and it gave me wrong results in the past (I'm not aware of any wrong result bugs in v10 though).
Also, the types of results that these graph-related functions give are often difficult to verify. It's not as simple as substituing back a solution into an equation. Verifying the results with igraph gives me some peace of mind. I wouldn't trust either Mathematica or igraph alone, but when they agree, I trust the result.
By "not well documented" mean things like: what a function computes is not always described precisely. E.g. EdgeBetweennessCentrality returns twice the result I'd expect for undirected graphs (not directed ones). Comparing with igraph made it immediately clear what's going on.
Also, it's not documented what works with weighted graphs and what doesn't, so it takes some effort to figure this out (i.e. if weights were taken into account when computing the result).
E.g. EdgeBetweennessCentrality uses weights, but BetweenessCentrality does not.
Ah, it seems the documentation has been updated and now this is mentioned (in v10).
@Taliesin BTW looking here, version 10 took care of many of the requests.
7:40 PM
This gives the letters in terms of FilledCurve in V9, but gives exclamation points in V10, one ! for each letter.
Graphics[ImportString[ExportString["ABCD", "PDF"], "PDF"][[1, 1]]]
The same thing happens with Import and Export. (Export produces the good-looking PDF. The problem seems to be with Import.)
However, this works fine: First@Import["ExampleData/mathematica.pdf"]
@Taliesin So my first request was PythonLink, and my second request: a function to replace the diagonal of a matrix with something else (or zero it out). Is is something extremely basic, and not difficult to implement (I already have it in a personal library), but surprisingly it's something I use a lot in practice. Just to have a small reuqest in addition to the big one :)
Many of my other requests (DelaunayMesh, Managed Library Expressions, MaximalBy) have already been added to 10.
8:29 PM
Actually what I would like right now is: 1. faster histograms (they could be much faster than they are), especially for more than 1D 2. for all the empirical distributions (EmpiricalDistribution and HistogramDistribution), allow supplying already binned data, and allow supplying multiplicities for datapoints. And of course make them a bit faster to construct (same as point 1.)
9:14 PM
Hello hello. Home from work. I must say I'm pleased that my employer is pleased with my productivity even though I spend a significant amount of time each day doing Wikipedia and game design stuff. I never thought I'd be able to get paid for that. I hope it lasts.
@TaliesinBeynon Nice upcoming features list.
@Szabolcs I've kind of wished the front end would collapse part double-brackets automatically. It already highlights the matching pair.
9:40 PM
@MichaelHale That would be the best, but if there's no time to implement that, an official kb shortcut would do well too.
@MichaelHale Do you use Mma at work?
Not yet, just wrote an R script this week. They know I prefer Mathematica though. I convinced one of my teammates to check out the trial. I think I'll be able to convince them gradually to let me use it over 6 months or so.
@halirutan You should take a look at this question if you have v10 and Gnome :) Sorry for pinging you here but you are the only Linux user I see that could have v10. Although, I doubt you use Gnome.. :)
10:07 PM
Only joking :) but I've seen a lot of crashes this week. Bug fix release soon?
@TaliesinBeynon tks for share these good news. Nice to know about the function group for small implementations. BinListsBy a example of simple function that I miss, and have in my personal package. Use a lot in some data analysis.
@cormullion WRI people have indicated that the next point release is not too far off. Nice joke, when at last I realized what you'd done :)
@pickett Perhaps Mathematica's simply telling me to down tools for the week. It's right! :)
10:44 PM
@Murta For a couple of months I worked on a very flexible, multi-dimensional binning system. It's the main tool I used to do blog.stephenwolfram.com/2013/04/… . It never made it into 10, because we didn't know quite how to represent the resulting multi-dimensional count objects.
@Szabolcs thanks for the ideas, and the links
@TaliesinBeynon I would very much like to see a more powerful Trace-tool. And maybe a compiler targeting on Android?
Another hope is more efficient Graphics/Graphics3D render engine capable of dealing huge amount of primitives smoothly. (I was trying to do some real-time 3D ray-tracing.)
@Öskå Yep, Gnome user for years. I'll take a look at it.
11:04 PM
@cormullion Next month was mentioned a few times.
@TaliesinBeynon Ah, I completely forgot: what I'd really like is a general file format for storing arbitrary Mathematica expressions. Right now there no less than four ways to do this: 1. plain text expressions, i.e. Export[..., "Package"] 2. WDX 3. MX 4. Export[..., Compress[expression], "String"]. But all of them have some significant weakness.
1. big files, very slow to read/write 2. very, very, very slow to read/write, and I don't know why! 3. not-cross platform and not suitable for archiving (will it work in v11?, probably not). But it's very fast. 4. This is what I use, but for packed arrays it's still too slow to write. Does seem to be cross-version compatible, but there are no official guarantees frm WRI.
What we need:

1. cross-platform and cross-version (i.e. readable in future versions).
2. store arbitrary Mathematica expressions
3. optimized for common scenarios, i.e. reading/writing packed arrays (and maybe Dataset) should be very fast
4. space efficient (MX tends to waste space)

« first day (912 days earlier)      last day (3429 days later) »