Wolfram Mathematica

12:00 AM

Nope. I'm also not so active these days. I need to study a bit.

Got hold of the Combinatorica book, I'm reading that

I see. I have the book too, for a long time, but only read parts of it

At the time I was reading it I did not have the experience I have now. What I'd do now is to start reading the code of Combinatorica from the source, looking into the book only when something is not clear. This would be much faster.

Szabolcs

the book also contains some useful theory, not just a descriptions of the algorithms

I mean the newer book, from 2003

Leonid Shifrin

@Szabolcs Yes, right. But I am not sure that in this capacity it is the best / most effective source (yes, I mean the new one). So you need it for your new job?

Szabolcs

Yes, I need to learn a few new things. The only problem is that I constantly get distracted and keep reading about things that are not essential

Leonid Shifrin

@Szabolcs Tell me about it:). But, you know, I miss those times in academia sometimes. Now I don't have much time for distractions, although I still distract a lot

@Szabolcs Speaking of distractions: I seem to have finally found, as a side product, a cheap but effective way to implement and use OOP in Mathematica, if we don't care much about performance.

Szabolcs

12:11 AM

Ah, that's always the catch: if we don't care much about performance :)

@Leonid Did you work around the problem with Temporary symbols not being removed sometimes?

Leonid Shifrin

@Szabolcs I used to think the same way, and for many applications this is true. But what I also realized is that a lot of problems stem from the impedance mismatch between immutability and OO as we know it in other languages (C++,Java, etc)

@Szabolcs Symbols: I tend to avoid mutable state im my approach as much as possible. For object instances, I use the same approach as JLink, having something like OOBlock - so that I collect stuff created inside

@Szabolcs What I found interesting is that we can combine instantiation, inheritance and convenient method call syntax (like JLink) with immutable expressions and objects based on behavior much more than state.

Szabolcs

@Leonid If you have time perhaps you could do a short blog post on this. Or maybe a series with a number of short posts.

But aren't instantiation and immutability somewhat contradictory? Or do you mean something like With when you say instantiation?

Leonid Shifrin

@Szabolcs I was thinking about it. But no time now. I came to it as a part of the code exchange mechanism for us, that I am working on.

Szabolcs

I mean, create once, and keep it until the end of the OOBlock?

Leonid Shifrin

@Szabolcs The way I approach this is that I create an object by assigning some type wrapper to some expression, like MyObject[{1,2,3}] out of {1,2,3}. Then, I assign behavior to the head, like we usually do in Mathematica

By instantiation I mean that any created object has its own identity, so two MyObject[{1,2,3}] are different instances. In fact, I need this for purely technical reasons, initially I was hoping to do without even that. But it turned out to be actually useful

@Szabolcs So, the catch is that if I need to manage some state, I create a closure with the behavior I need, and attach it as a new method to my object.

So, I don't deal with the state explicitly much. This allows me to do all this in a way which seems to me quite natural for Mathematica

Szabolcs

12:23 AM

What did you use this for? Can you give a (simple) example?

Leonid Shifrin

Yes. I have a structure which describes the structure of some project, and is Mathematica analog of JSON - nested list of rules. It looks something like this:

{
"author"->
{
"name" -> "lshifr",
"email" -> "lshifr@gmail.com",
"url" -> "http://www.mathprogramming-intro.org"
},
"name" -> "LocalDataExtrema",
"mathematica_version" -> "8.0+",
"description" -> "New description",
"url" -> "https://gist.github.com/4238023"
}

@Szabolcs I want to have an easy way to manage this structure, in particular change some properties, extract them, and also - which is really important, persist it on disk and read from disk in an automated manner

@Szabolcs Basically, I want to make persistence on disk, and a number of other operations, such as mapping between string form and expression, automatic.

The reason why I needed objects is that I have more than one such structure, and they are different but share many common properties and behavior, so I wanted inheritance

Also, I wanted to be able to bind a particular object to some file, and then not drag the information about the file around. And many otherthings. To put the story short, I found out, rather surprisingly, that my Mathematica code for this kind of application I need to make now, really sucked. It weas cluttered and mixing concerns.

@Szabolcs When I started investigating what's going on I realized that standard Mathematica coding practices do not provide the level of encapsulation / information hiding that I needed, so I gradually realized that I need some form of an object model, to organize the code better.

Here is how my type definition currently looks iike for this structure:

DeclareType[RuleTreeInfo][
Get[prop_?propertyQ] :=
With[{result = getProperty[$content, prop]},
result /; result =!= $Failed]
, Get[prop_] := $Failed
, Normal[] := $content
, ToString[] := toCodeString[$self@Normal[]]
, Delete[prop_?propertyQ] :=
resetContent[$self, removeOptions[$content, prop]]
, Replace[changeTree_] :=
resetContent[$self, mergeOptions[$content, changeTree]]
, read[] := read[$self]
, save[] := save[$self]
, bindToFile[file_String] :=
Module[{},
Evaluate[$self] /: read[$self] :=

Szabolcs

let me copy that to a notebook

Leonid Shifrin

@Szabolcs it uses a number of auxiliary functions which are needed to fully understand what is going on, but this is just to give an idea

@Szabolcs I should have used Read and Save instead of read and save as well :)

Szabolcs

12:39 AM

@Leonid Can you make sure that there'll never be a conflict with defined symbols? (Normal, Get, etc.)

rm -rf

I finally got hold of David Wagner's book!

Leonid Shifrin

@Szabolcs If you use System symbols, then yes :). This was a tough cookie to make :) If you use user-defined symbols, then I am thinking of insisting of creating them in one special context devoted for them, for all types. But I did not automate that yet

@rm-rf Big congrats!

Szabolcs

@rm-rf I just ordered it half an hour ago ;)

Leonid Shifrin

@rm-rf Now you know who my teachers were (Wagner and Maeder) :)

Mike Honeychurch

Is the Wagner book still in print?

Szabolcs

12:42 AM

I ordered it from a library, not a new one.

Leonid Shifrin

@rm-rf, @Szabolcs But you guys are already on a level where it may not be as helpful as for someone less experienced

@MikeHoneychurch It has been out of print for at least a decade by now

rm -rf

@Szabolcs Same here. I'm definitely going to make a copy for personal use

Mike Honeychurch

Yes I think I found it in a university library many years ago.

Szabolcs

@Leonid Is this going to be in the next version of ProjectInstaller?

rm -rf

@LeonidShifrin Is Maeder's book still available?

Leonid Shifrin

12:47 AM

@Szabolcs It is a different project. ProjectInstaller is just a bootstrap installer which can install a wide range of projects and packages, since it has almost no restrictions on the project's structure. This one will be tied to GitHub gists, have a versioning system, dependency tracking, and dev.tools to make it easy to publish and install code. The ProjectInstaller will likely be used to bootstrap the installation of this one.

@rm-rf Yes, it should be, but not sure if it is still in print. But buying it used was always an option, and not an expensive one. I did not check recently, but I don't expect any significant changes in that.

Mike Honeychurch

maeders books are still in print

Leonid Shifrin

@Szabolcs I already can create and update some test gists from Mathematica, e.g. this one, but I still need convenient versioning system, and developer tools to publish code easily. This is where I am now.

Szabolcs

@LeonidShifrin So creating and updating the gists will all be possible from within Mathematica?

Leonid Shifrin

@Szabolcs Yes, it is already possible :)

@Szabolcs But I still need to drop the dependency on curl in favor of either new functions of V9 or, if we decide to support V8, some Java code. I also need to improve the authentication / security.

Szabolcs

@LeonidShifrin What do we need to have on the local machine for these packages? Typically a package .m file and another .m file with the metadata? Is it going to work without needing git?

@rm-rf I often don't get the -ding!- sound on OS X when I get a reply here in chat. Do you have any idea what might be wrong? Something OS X specific? Let me try Firefox ...

Leonid Shifrin

12:56 AM

@Szabolcs Yes, that's the entire idea - without needing GIT (but since gists are also git repos, those who want it will be able to use it - but I will add this functionality later). On the local machine, there will be a Mathematica client, which is basically the package manager for user's repository and the isntaller/ loader. The loader will have some runtime component

rm -rf

@Szabolcs Coming soon: GitLink` :)

@Szabolcs Maybe... I've heard others complain, but I usually have my notification turned off

Leonid Shifrin

@rm-rf I have it it my plans (Git link), but later on. There is JGit, you know

rm -rf

That's for the WB, right (EGit)?

Leonid Shifrin

@rm-rf no, this ispure Java Git implementation. Sweet thing to link to Mathematica. Also, I don't put WB in the center of what I want to do. I want a FE-based set of dev tools.

rm -rf

Oh, nice :)

Leonid Shifrin

12:59 AM

@rm-rf Well, words are cheap :)

@rm-rf But you know, most folks would actually like a server-based simple versioning intergated with FE workflow,without really using git - this is what I am trying to make. I think Github gists are an ideal format - I did not find any feature they would not support, except nested directory structure, but I bet most code people would like to share will not be like that. And we can add that support for full-fledged repos later

@rm-rf The biggest problems were (and are) to design it right, to have the right set of features - just enough to work effectively 90 % of the time, without turning this into a full-fledged VCS

Szabolcs

@magma Sorry for the late reply. I am really not sure what happened there ... could you please try to remove all versions you have an re-install the newest one? Simply go to $UserBaseDirectory/SystemFiles/FrontEnd/Palettes and remove all versions you see there. Then download the version from the link I posted above and copy it there. Do this while Mathematica is closed, please.

I'm wondering how to easily discard the front end version information from a notebook. Apparently, Export["a.nb", Import["a.nb", "Package"], "Package"] will break the file. Not sure why yet.

Leonid Shifrin

1:16 AM

Ok, guys, have to go. 5 a.m. here, time to get some sleep :)

halirutan

@LeonidShifrin Sleep well.

It's 2:30 here and I just got up.

rm -rf

@LeonidShifrin Yes, that seems like the right approach. If they need a VCS, then well... they should use a VCS!

Leonid Shifrin

@halirutan Thanks. Got up?

halirutan

Fell out of bed..

Was so tired yesterday that I went to bed with my daughter at 8:00.

Leonid Shifrin

@halirutan Are you all right then?

@halirutan falling out of bed can be painful

halirutan

1:19 AM

@LeonidShifrin Was a joke ;-). Now I had 6 hours and I can work at night when everything is silent ;-)

rm -rf

@halirutan Is this one of those lost-in-translation things because of different German words for waking up and physically getting up from bed?

halirutan

@LeonidShifrin I would fall on the dog.. that would be painful.

Leonid Shifrin

@halirutan So, I am not alone in these harmful habits :)

halirutan

@LeonidShifrin Nope

jfklein13

@halirutan, do you still have issues with the MINT_32-related code?

Leonid Shifrin

1:21 AM

@rm-rf That's the point, it is very easy to fall into a trap and start re-implementing what essentially is git. The trick is to exclude certain sweet possibilities

@jfklein13 Hello Joel

jfklein13

@L

Hi Leonard

Leonid Shifrin

@jfklein13 Well, I dont' have such grand ambitions - so just Leonid

halirutan

@rm-rf I don't know.. Does "I just got up" doesn't mean I was coming from bed?

jfklein13

sorry, my fingers messed that up :)

Szabolcs

The answer to why Importing and Exporting as "Package" breaks a notebook is that the InputForm of RowBox[{"1", "+", "1"}] is $1 + 1$. This is how it writes the expression back to the .nb file, and the FE can't read that.

halirutan

1:23 AM

@jfklein13 It seems to be no real issue. Just bad documentation. In fact in WolframLibrary.h there is absolutely no documentation which is a quite nice feature for a header file.

Leonid Shifrin

@rm-rf But I will provide basic VCS features, such as rolling back to a certain revision

jfklein13

Sorry for your troubles on that one.

halirutan

@jfklein13 So the essence is, you cannot use the MINT_32 switch. I wanted to post a question about this and had already the introduction text but then I didn't know what to ask. For an explanation? It's maybe not appropriate here.

rm -rf

@halirutan yes, but I could very well say "I woke up at 6 am and made some tea" and it'll be perfectly clear that I actually got out of bed, walked to the kitchen to make tea. However, to a German, it could mean that I stayed in bed (but awake) and somehow made tea ;)

Szabolcs

Isn't it 5:30 AM there Leonid?

Leonid Shifrin

1:26 AM

@halirutan I do keep in mind the formatter, b.t.w.

halirutan

@rm-rf Somehow.. Wife, I need tea! ;-)

rm -rf

sudo make tea

Leonid Shifrin

@Szabolcs Yes, and so I really have to go. I did not plan to stay that late tonight. See you all!

Szabolcs

See you, have a good sleep!

halirutan

@LeonidShifrin No hurry. I have a lot to do and I'm really addicted to this "distraction" thing.

Leonid Shifrin

1:27 AM

@Szabolcs Thanks!

@All Enjoy the rest of the day, folks!

rm -rf

Off he goes in his sleigh

halirutan

Cool, and I thought I have hallucinations seeing a sleigh..

@rm-rf About the Wagner book: I was writing the phd-mentor of Wagner because I couldn't find someone else who may know how to reach him. I was thinking about asking him or the publisher whether it's possible to make this material public.

Lately, I asked the Futz about it. You might have seen it:

Dec 4 at 15:19, by John Fultz

@halirutan I haven't seen or heard from David Wagner in years. Last I heard, he got a permanent consultant gig, and then he just completely dropped off the map. He hasn't been to any conferences, and I'm not aware of any exchanges with WRI personnel. I just checked, and his last bug report/TS exchange was in the 90s. His old consulting website has been snarfed up by a domain name troll.

Szabolcs

@halirutan you're both up at this hour? how much sleep do you get per day?

rm -rf

Yes, I saw that. It's surprising how one can disappear completely (the only logical explanation would be that he's no more, but no one knows)

halirutan

@Szabolcs Today from 8:00 to 2:00.. so 6 hours. But if I'm really tired (because sometimes you just don't manage it to go to bed) I just sleep. Even on my chair in the office.

Szabolcs

1:37 AM

I simply can't live like that ... which probably means I'll never make professor, based on my past professors' habits

They didn't sleep either

halirutan

@rm-rf I haven't got a reply since 5 days know.. I hope I still get one.

Szabolcs

:7219176@halirutan Try searching on Google Scholar. It brings up a profile. But I don't know if it's the same person.

halirutan

@Szabolcs It just works because I don't have to get up. It would all be different, when I had to be at 8 in the office or something like that.

rm -rf

@Szabolcs No, it's a guy in Berkeley CS

I saw that already

halirutan

@Szabolcs I have a patent entry from 2004

google.de/…

Szabolcs

1:46 AM

@halirutan You do have to take your daughter to kindergarten, no?

halirutan

@Szabolcs Yes, at 9. The next reason I like to get up in the middle of the night.. You don't have to get up early at morning, because this really sucks.

rm -rf

@halirutan His PhD advisor's page says that he's now at Google (someone else mentioned it here too)

halirutan

@rm-rf I thought this means "google him!".

But obviously not. It makes much more sense that he is at google now.

rm -rf

Dec 2 at 2:16, by Oleksandr R.

@P.Fonseca Principia Consulting is defunct. Wagner went to work for Google. No idea what happened to him since 2004 though...

Ok, so probably Olek is referencing the 2004 patent too...

Szabolcs

Have you ever burnt a CD with both audio and data tracks on Mac?

halirutan

2:01 AM

@Szabolcs I use Burn

rm -rf

@Szabolcs What's a CD?

Szabolcs

@halirutan I managed by using Burn and writing CUE sheet

@rm-rf I need to burn some CDs for my wife with her music. I'm trying to make one CD that everyone will enjoy, both old fashioned people who really prefer their CD player (which is the device hooked up to big speakers), and people who'd just rather copy it to a mobile device. I did it before on Windows, but not on Mac. Finally I made a cue sheet.

rm -rf

@Szabolcs I like the music!

halirutan

2:28 AM

@Szabolcs Sounds like your wife likes piano music. When she is not completely against mainstream movies... johnny depp.. pirate of the caribbean.. etc you should show her this here: youtu.be/NzGgX1DihPw

rm -rf

@halirutan She's the one playing

halirutan

@rm-rf I thought I read something like that from Szabolcs. That's why I posted it ;-)

Mathematica StackExchange (meta)

0

Q: Can I make two similar posts---one for graphical assistance and the other to display the result cleanly so I can cite it on OEIS?

A while back, I discovered a convergent sequence OEIS A219733 that on other forums I argued that was fractile-like or self-similar. I have found an illustration of the Droste Effect on this sequence. However, I have poor skills when it comes to creating images. So, I want to ask a question t...

discussion

halirutan

4:37 AM

In Linux it seems the line-wrap in the frontend is broken. Can someone test this? Make a long line like the following function

testFunctionWithAVeryLongName[argum1_,argum2_,argum3_,tfunc_]:=
With[{a=1,b=2},a+b]

Important is, that the first line goes almost to the end of the notebook-window. Then append something to tfunc

for instance another parameter

testFunctionWithAVeryLongName[argum1_,argum2_,argum3_,tfunc_,anotherverylongparameter_]:=
With[{a=1,b=2},a+b]

What happens here is, that the line is not wrapped:

Can someone confirm this?

(cc @JohnFultz)

Nasser M. Abbasi

8:16 AM

@halirutan on windows, I do not have this problem. BUT, on windows, last night, suddenly the lines were not wrapping in my notebook (V9). I did not know why. Happend only one time. But simply restarting Mathematica fixed it. You might try to restart Mathematica on your Linux while holding the CTRL-SHIFT key while MMA is going up untill it is up. This way you can reset all things to factory setting and try to see if this problem is still there.

@acl, ok, after 24 hrs debugging, I found why book solution was not matching MMA DSovle under some input. It turned out I was writing ArcTan[y/x] as written in the book. When I changed this to ArcTan[x,y] to take care of the signs, now my solution matched MMA. Book does not mention such things, and assumed y was positive always. oh well. 24 hrs gone on this, but I learned something :)

soandos

10:39 AM

Does anyone find that RandomPrime[{2^2047,2^2048}] always generates the same number?

Nasser M. Abbasi

10:51 AM

@soandos not on V9 under windows 7, attached screen shot

acl

11:26 AM

@NasserM.Abbasi I think this arctan thing is a standard way to waste a couple of days :)

Elfet

11:43 AM

Hello!

Can any one help me?

I try to do vedio tracking.

How to create histogram of color for image window?

halirutan

11:55 AM

@Elfet And ImageHistogram does not help?

reference.wolfram.com/mathematica/ref/ImageHistogram.html

Elfet

12:30 PM

Yes!! Thanks!

Szabolcs

3:39 PM

@halirutan Not on OS X. I'll try on Linux when they upgrade to Mma 9 on the shared machine test.

@halirutan Oh, we've seen that one! :-) It's very manly playing

Gabriel

4:19 PM

mma 9 is breaking my will ... suddenly plot styling from old notebooks is different ... it is so much work to figure out what has changed from one version to the next ...

I really think a change log would be great ... so I can know what behaviour has changed ... without having to trouble shoot every fricken document

Brett Champion

@Gabriel Can you give an example of a change in plot styles? It may be a bug.

Gabriel

@BrettChampion for some reason I have some code that when I do PlotStyle->{Thick, Red} it is just thick so I need to do PlotStyle->{{Thick, Red}} which I imagine means I have some extra nesting in my code ... I just can't spend time right now trying to pin it down ... might be a bug on my end, but man it seems all my code is full of such errors in the upgrade ...

@BrettChampion okay the rage has died down ... here is an example that works in v8 but ignores Red in version 9 ...

ListLinePlot[
Transpose@Table[
{
{r,
r^2
}
},
{r, 0, 10, 0.1}
], PlotStyle -> {Thick, Red}
]

or with less brain dead formatting

ListLinePlot[Transpose@Table[{{r, r^2}}, {r, 0, 10, 0.1}],
PlotStyle -> {Thick, Red}]

from a much larger piece of code ... I imagine it is the extra nesting ... so not sure if the bug is in v8 or v9

rm -rf

4:41 PM

@Gabriel Yup... red in 7 and 8, blue in 9

Gabriel

makes me want to cry ... the new NDSolve features make me a v9 lover ... but man I have spent the last week "fixing" all my code to be usable ...

Brett Champion

These will be red and thick, and aren't ambiguous:

ListLinePlot[data, PlotStyle -> Directive[Thick, Red]]

ListLinePlot[First[data], PlotStyle -> {Thick, Red}]

Gabriel

5:16 PM

@BrettChampion yeah went that route ... but is this a bug fix? Or just that the code base was changed and this happened?

All these changes in what seems to be default behavior are tough to handle when you are maintaining code ...

magma

@Szabolcs thank you , I am travelling. I will continue testing next week.

Szabolcs

5:49 PM

@Gabriel I'd say the 9 behaviour is quite consistent. Directive was introduced in 6, but could still often be swapped for List (the notation we used in <= 5), so it was a bit messy...

Having to rewrite code, that's not so much fun though ...

Gabriel

7:20 PM

@Szabolcs agreed the new behavior seems better, but without a list of such changes it makes the user experience brutal

Transcript for