« first day (279 days earlier)      last day (4206 days later) » 

12:45 AM
@Oleksandre Ping..
 
@halirutan hi!
 
@OleksandrR. Hi.
No luck on Windows?
 
That is really weird. For me I was seeing undefined references to MLGetRealList etc. using version 3.
 
I got here the following warnigs:
 
I suppose it could be some oddity in the MinGW libs. qhull itself seemed to work fine.
 
12:53 AM
qh-math.c: In function ‘delaunayQhull’:
qh-math.c:122: warning: passing argument 3 of ‘MLGetByteString’ from incompatible pointer type
/usr/local/Wolfram/Mathematica/8.0/SystemFiles/Links/MathLink/DeveloperKit/Linux-x86-64/CompilerAdditions/mathlink.h:4067: note: expected ‘int *’ but argument is of type ‘long int *’
Did you have a look over the code that all the ML-functions are still called correctly?
 
Yeah I got that too until I switched to version 2. I got a lot of other warnings as well though. Windows is weird because long and long long are the same according to MSVC and not with GCC. It's a real mess.
Which gcc are you using? 4.7.1 here.
 
patrick@lenerd:/etc/X11$ gcc --version
gcc (Ubuntu 4.4.3-4ubuntu5.1) 4.4.3
 
OK. Maybe I'm too near the bleeding edge... though I think probably this isn't a compiler bug. I just gave up in the end (rather than try compiling different versions of qhull) but considering the trouble I had, it's surprising it was so effortless for you.
 
What is maybe worth trying is to setup a small extra function calling qh_version and only returning the string to Mma.
So you can be sure it's not the connection to qhull.
I had some hard-time that dependent dll were not found on Windows.
Or are you linking static?
 
BTW I also had to use g++ to compile qh-math because otherwise #include <string>/<vector> etc don't work and I had to make an extra #define too (per the qhull makefile).
I linked fully static. Saves trouble if I were to send the exe to someone else.
 
1:00 AM
Hmm, I used gcc only.
 
Well, static as far as qhull goes. Not static MathLink as well (I think that's pointless).
 
@OleksandrR. yes, I meant that.
I really wonder why no one made a working V8 package for this. The build in Delaunay is a mess.
The output is unusable
 
TBH I never needed anything like that. I just thought I could do a service for paradox2 by compiling it on Windows. But it wasn't so easy as I thought...
 
How do you interpolate on a non-regular grid then?
 
Well, I don't. :) I never needed interpolation so far.
 
1:03 AM
The easiest thing is to use a triangulation and interpolate the surrounding 3 points to my knowledge.
;-)
 
Gratz on the 10k btw. I remember at one time we were nearly level, and now you're way ahead!
 
Thanks. That's because you can concentrate on your work, while I enjoy the other funny things far too much.
@OleksandrR. I would trade some rep if you write my paper.. interested? ;-)
Almost everything is finished. Data, analysis, ... just needs to be written down.
 
Haha. That's exactly the problem for me too. I really hate writing. Currently trying to finish my thesis.
 
Maybe we should ask @Leonid. He seems to be able to concentrate and to reduce other things.
 
 
1 hour later…
2:33 AM
Hello
 
 
11 hours later…
1:05 PM
@Sjoerd are you around?
 
@Mr.Wizard Well yes, but I can't spend much time here now
 
I would love to see these graphs, among others, for Mathematica.SE:
21
Q: How do I access the StackOverflow API from Mathematica

Brett ChampionI was wondering the other day if StackOverflow had an API I could access from Mathematica, and apparently it does: "Saving plot annotations" What's the best way to get data from StackOverflow into Mathematica? Sjoerd used the information to make a plot. I'm interested in adding SO-related noti...

Is that possible? If it is, are you willing to do the work?
I don't even know if we have an API on this site. Where would I learn that?
 
I could do that. I once made an update to the SE new question alarm, so I know that I can work with the 2.0 API.
 
1
Q: Include brief explanation of or background for questions?

George WolfeSomeone recently asked how to speed up iterated manipulations of a graph/table and included a (failed) link to an article about why he wanted to do it. It's about something important that has been recently discussed in Seed Magazine, the Publications of the National Academy of Science, and in new...

 
Shall I let you ask a question and answer it at your leisure, or would you rather I asked something to kick it off?
 
1:09 PM
@Mr.Wizard I'll look into it tonight or at least this week.
 
Excellent. Thanks!
 
Perhaps best you ask a question when I've finished. I'll ping you by then
 
@Mr.Wizard (for future reference: api.stackexchange.com)
 
Okay. Make sure you include halirutan in the graph as his hitting 10K is what got me thinking about this.
@TimStone Thanks for the link. May I assume the API is available for Mathematica.SE? last I checked the DataExplorer was not, for example.
 
Yeah, the API is available for all accessible sites. The API routes take a site parameter, which for Mathematica is either mathematica or mathematica.stackexchange.com.
 
1:15 PM
Do you know when data.stackexchange.com is likely to include us? I recall reading that there were general delays while it's upgraded.
 
They ended up purchasing new hardware to move it to, apparently, so taking the delivery and post-delivery setup time into account, it'll be just a few more weeks (second or third week of November).
 
Glad to hear it.
 
In the meantime, if anyone has any Data Explorer feature requests, be sure to post them on Meta Stack Overflow and hopefully I can squeeze them in in time for the Mathematica.SE data to become available. :)
 
@TimStone speaking of feature requests I once asked if it would be possible to get a shortcut for inserting links to the Mathematica documentation pages. These have a uniform pattern: http://reference.wolfram.com/mathematica/ref/<FunctionName>.html. I would love to be able to enter something like [help:Sum] and have it rendered as a link to the help page for Sum. Now that the site has graduated I'm hoping this can be considered.
 
 
1 hour later…
2:50 PM
Alex Miller on October 23, 2012

On the show this week are Kyle Brand and Nick Craver, two SE employees who are heading up our systems upgrades and relocations – they’ll dish all kinds of details on our infrastructure, plus plenty of chat about other mildly relevant things.

First up on the agenda: Quantcast! Five minutes before we started recording, we noticed that Quantcast is ranking our network at #100! (or at least we were for a bit)

If all that additional traffic should cause our New York data center to go down, what will happen, Kyle? Great segue, Joel! We are working on a system for failing over to our datacenter in Corvallis, OR. …

 
 
4 hours later…
6:29 PM
posted on October 23, 2012 by Michael Trott

In my last blog post, we discussed 3D charge configurations that have sharp edges. Reader Rich Heart commented on it and asked whether Mathematica can calculate the force between two charged cubes, as done by Bengt Fornberg and Nick Hale and in the appendix of Lloyd N. Trefethen’s book chapter. The answer to the question [...]

 
@halirutan success! So it turns out that for qhull 2012.1 many of the files have been renamed, and qhull.h doesn't exist any more for C (only for C++). Last time I had tried to compile as C++ so that I could include the file, but the correct thing to do is apparently to just comment out the #include "qhull.h" line in qh-math.c.
 
6:49 PM
Hey
@OleksandrR. , do you know if there's a function in MMA that gives you the representation of a number as 10^10^10^10^sth, as wolfram alpha does?
e.g, WolframAlpha["13^12^11^10^9^8^7^6^5^4^3^2", {{"Result", 1},
"Content"}]
 
@Rojo good question. No, I don't know. I suspect not.
 
Oh
@belisarius, whatever the algorithm it uses for that, invesigating it may help in your convergence question
I just tried up to 50^49^48...
and in a few seconds it gave me the answer
and the topmost exponent clearly seems to converge
 
7:10 PM
If that's reliable, then Nest[Log, Nest[Factorial, someNumber, n], n] also converges
 
7:57 PM
@Rojo Hypercalc can do those calculations. See the update in my original question
 
@belisarius Seems like it
How
Don't answer, reading the help
 
@Rojo I used the javascript implementation. Easy, but no programming. One calc at a time.
 
Yeah, I'm wondering how it does the calculations
and reading its help
and now I have to leave. It's an interesting read
Later
 
 
2 hours later…
10:03 PM
@OleksandrR. Very nice. Unfortunately the OP seems to have no experience in C-programming and cannot compile it.
 
@halirutan I posted a comment with the binaries I created. I thought that since the procedure is not exactly simple (at least not the way I did it) and most Windows users don't have any compiler installed or know how to use one, just providing the binaries would be the most straightforward approach.
 
Let me look..
@OleksandrR. Ok, I see. I upload your files onto our git-repo.
 
Ok, great!
 
10:58 PM
Hi @user141603.
@user141603 Can you talk in here?
@belisarius Does one need a certain amount of rep to talk in the chat?
 
@halirutan yep
 
How much?
@belisarius and if I create and invite someone in a room?
 
@halirutan I think yes
 
@halirutan So, yes, that user can talk in here.
 
11:03 PM
@Mr.Wizard Ok.
because I asked him to join and he joined but is silent ;-)
@Mr.Wizard Thx. I always wonder how you find those things so fast.
@Mr.Wizard The bug from his question here is really strange
4
Q: Adding ContourLabels changes contour colors?

user141603I have a set of data s5 such that ListContourPlot[s5, InterpolationOrder -> 6, ColorFunction -> "TemperatureMap", ContourLabels -> True, ColorFunctionScaling -> True] but ListContourPlot[s5, InterpolationOrder -> 6, ColorFunction -> "TemperatureMap", ContourLabels ...

 
I'm not fast at all. In fact there is one page of the Moderator tools that I can never find. Literally, I've never found it. I have to keep a bookmark to get back to it.
 
Ahh..
This data here spans the interval [1,2] completely:
data = Table[.5 Sin[x y] + 1.5, {x, 0, Pi, .05}, {y, 0, Pi, .05}];
Look at this
ListContourPlot[data, ColorFunction -> "TemperatureMap",
 ContourLabels -> False, ColorFunctionScaling -> True]
and then at that
ListContourPlot[data, ColorFunction -> "TemperatureMap",
 ContourLabels -> True, ColorFunctionScaling -> True]
First one is
second one is
 
Jin
I saw the Wolfram conf photo, looks great!
Hope whomever went had a good time there
 
@halirutan In version 7 I get this:
 
@Mr.Wizard You're on Win?
 
11:13 PM
@halirutan Windows, ver. 8: same result as yours
 
@halirutan Yes, Windows 7 64-bit.
 
Really weird.
 
@Jin Which photo is that?
 
@Mr.Wizard The one in Meta..
 
Jin
yes that one
 
11:15 PM
 
Thanks, I hadn't seen that.
 
Jin
@Mr.Wizard bubububut you're a wizard, you should know everything...
 
@Mr.Wizard since you're using 7 on Windows, would you mind giving this a try using the binaries I compiled and linked in a comment? (No keyloggers, I promise!) Although I have version 7 installed, the MathLink libs are updated with each version and I want to be certain that something compiled against the version 8 MathLink works in previous versions too.
 
11:30 PM
@OleksandrR. hehe... says the hacker and places his trojan inside qh-math.exe ;-)
 
@Jin I'm afraid I'm a very bad wizard.
@OleksandrR. Hm... do I take the risk or not. :o)
@OleksandrR. It works fine here.
 
@Mr.Wizard excellent, thanks.
Disappointingly, the MathLink overhead seems to have grown from version to version. In version 8, with a mesh of spacing 0.005, the triangularization takes 5.3 seconds. In version 5.2, only 3.9 seconds.
 
Which part takes that long?
 
Just the triangularization itself. Of course I'm not trying to plot the result.
 
Sorry, I didn't ask that well. What portion of the time is in the calculation and what is in overhead?
Welcome @David
 
11:44 PM
Oh, sorry. Okay, well AbsoluteTiming is 5.34 and Timing is 0.34, so that's not enough to account for it; on second though I think the most probable reason is marshalling data between MLINTERFACE version 2 (used by qh-math and Mathematica 5.2) and MLINTERFACE version 3 (Mathematica 7/8).
 
@J.M. you know you can still use regular stars for things that amuse you...
@DavidSlater are you with us?
 
I am, but only briefly
 
OK, just checking.
 
Thanks for the help with TreePlot, saves me from trying to use Graphviz :)
 
@halirutan you're almost at the daily rep cap. Let me see if I can help with that in the next five minutes...
 
11:55 PM
@Mr.Wizard How can you help me with that?
And the daily rep-cap was because I really wanted to reach 10k ;-)
 
I just mean vote for a post or two. Isn't it fun to cap?
 
Sometimes not, if you see you don't get rep for answers which took you really a while ;-)
@Mr.Wizard But I assume it's fun when you start your day with capping without doing anything.. just from old answers!
 
It seems I've already voted for most of your answers. With effort I did find a couple I'd missed that I also felt deserved a vote.
@halirutan I've never done that.
 

« first day (279 days earlier)      last day (4206 days later) »