Managed to get 5.2 to run, but had to install it on a Virtual Box under windows XP. I do not know why it does not work under windows 7.

@Nasser You might try starting it in win XP compatible mode under windows 7.

@Karsten7. I did. I tried all those options on windows 7, none of them worked. 5.2 comes up OK. But everything I type it reads it as plain text. I think the style sheet is corrupted. So I gave up and installed it under Vbox with real XP and now it works.

I am running integration tests and wanted to see how much it improved from older versions to current one.

If I remember correctly, Win 7 Pro has a Virtual Windows XP. But if Vbox is working, than there isn't much benefit in using that.

Hope you are going to post a link to your result here. Looking forward to it.

@Karsten7. I have windows 7 home edition,. I did not know win 7 pro has build-in Virtual windows XP! I am running now integration tests on version 5.2. Had to change few minor things in code to make it run, but it is running now. Amazing actually I had to make very few changes only in the code. For example. 5.2 does not have NotebookDirectory[] and had to change Import option to make it read the files. That is pretty much it.

@Szabolcs I think we can actually validate your sense that more questions are bad these days than they used to be. If we take closing as a sign that a question is bad, then we can compare the likelihood that a closed question was closed on a given date to the likelihood that an arbitrary question was asked on a given date.

That comparison looks like this:

Now of course we can't rule out that closing has simply gotten more aggressive of late, so maybe the question quality itself hasn't atrophied but rather more frustration has built up amongst the core cohort who make up the site. And I'm not a data scientist so I don't really know how to control for that.

The plot above also closely resembles the one for when a question itself was closed, which comes from the fact that if a question is going to be closed, it will most likely be closed within a day from asking, which we can get from this plot:

@b3m2a1, bear in mind with a larger stock of answered questions there will be more closed for being duplicates. Can you split into different closure reasons?

@KraZug I felt like the number of duplicates couldn't make such a drastic change and I think splitting it up does show that:

Like the number of duplicates shouldn't have changed so much after say the first year, once Mr.Wizard answered all the main questions

Those "bad fit" questions are the too broad, too localized, opinion based, etc.

You can see the duplicates start to track the basic asking pattern by about September of 2012

Note that "off-topic" seems to be mostly "trivial error". I don't think I can cut the data any finer than that (the keys don't seem to be in my dataset).

If you want the notebook where I'm doing all this it's here: wolframcloud.com/objects/b3m2a1/mse_question_closures.nb

Of course, this could be skewed by the fact that these closure reasons may not have existed in the current form until ~June 2014.

Hi, off topic. How can I set entry m[[i,j]] in a matrix to be 0 when j/GCD[i,j] is composite?

ReplacePart[m, {i_,j_}/;compositeQ[j/GCD[i,j]]->0]?

You'd need to define compositeQ yourself though

Can Mathematica solve linear an infinity by n linear system?

@b3m2a1 so the number of questions rises when the school begin. i think students are sharing the link to this site among themselves

I really miss the old help system that was in version 5. It allowed one to have global view of help in structured way, instead of the current flat structure. Here is a small movie

I am sure there is no hope that WRI will switch back to that good help system by now. May be we can start a petition? :)

@Alucard well the number of questions does rise around then, but note that 1) these are probability distributions so they simply describe the probability that a question asked at time x will have been closed and 2) the probability that an arbitrary question would have been asked at time x does peak around the starts of university terms (Sept and Feb or so), but this isn't reflected in the closure probabilities

What calculation can I use for ways to split 21 fruits between 6 monkeys?

If all monkeys are essentially the same, so I'm not counting giving 21 to Monkey A being different from giving it to monkey B.

Hello @Piomicron, this chat is about the software Wolfram Mathematica. You are perhaps looking for the mathematics chat room.

@Piomicron IntegerPartitions[21,6]?

oh, that's also an interpretation. I got caught by the word "calculation"

@Nasser You can be interested in this old question of mine: stackoverflow.com/q/5779835/590388

@AlexeyPopkov thanks, that is very interesting. I could not figure if you meant one can actually generate version 5 look-alike help system on current Mathematica versions? Do you mean one get version 5 look-alike help on 11.2? With all the 11.2 help in there?

@MichaelHale That seems to be oh so close to what I'm looking for. Looking at the table, I can tell how many there are in 6, however, how do I see the list of examples?

Thanks, by the way

May be we can bribe someone who works at WRI to make a 5.0 look-alike help system as build-in for future versions of Mathematica?

@Piomicron You are saying you want only the options where every monkey gets at least 1 fruit?

IntegerPartitions[21, {6}]

Ah, thanks

@Nasser The point of the question was how to use legacy Documentation from version 5.2 in the never versions (which at that moment had the legacy Help Browser, but it was empty; don't know whether it is there now). The Documentation files can be taken from an existing installation of version 5.2.

@AlexeyPopkov I see. I am sure it is all then gone now in 11.2. So it would not be possible to reproduce :( I wanted all the help in 11.2 to show as it was in 5.2

@Nasser Yes, it seems to be missing: evaluation of FrontEndTokenExecute["HelpDialog"] does nothing in version 11.2.0.

@Nasser I do not believe that we can convince Wolfram to give us back the legacy Help Browser. But currently we have something that can help to implement it ourselves and in some sence is a reincarnation of it: WolframLanguageData. In its current state it is just miserable, but I'm sure Wolfram wish to have it to be at least as good in structurizing the language as the old Help Browser was. So what is possible is to convince him to improve the WolframLanguageData.

@AlexeyPopkov that will be too much work for us users to do this I would think? It seems like a major task. WRI I am sure still has the code for the legacy help system. So someone inside WRI could rebuild all the current help pages into it? I do not know how they store the help pages as. Either way, I still think 5.2 help was much better than current help pages in terms of learning about what is there.

@Nasser The key advantage of the legacy help was how clearly the language was presented there in the tree structure form. Current help has the "See also" section on every Documentation page which is an analog, but currently it doesn't contain links to all strrongly relevant functions (but still is useful). If Wolfram would make this section to be comprehensive, I wouldn't too worry about the old Help Browser (which indeed had comprehensive tree of related functions).

In principle re-implementing it using Dynamic on the base of WolframLanguageData wouldn't be too difficult, just slow (and currently WolframLanguageData is miserable, as I wrote above).

I'm sure this is asked and answered, but I can't place my finger on it. I'm trying to create a manipulate-encapsulated graphic that will only change size in the x-dimension. code here. I'm trying to demonstrate how an interference filter works, and would like to enforce that the width of the dielectric plays a role in the filter design.

@bobthechemist what do you mean "will only change size in the x direction"? Like ImageSize wise?

@b3m2a1 In the pastebin version of my example, moving the slider initially increases the distance between the thick black lines, but once it reaches the size of the manipulate window, the aspect ratio of the graphic changes, so that it looks as if the wave is getting smaller in amplitude. I don't want that and would like to have the image keep getting longer (in the x-dimension)

afk for abit

Ah okay. Well you could have the AspectRatio set explicitly from your data. That's probably the easiest as you can just use a CoordinateBounds or something

@AlexeyPopkov you can also rebuild a form of metadata index pretty quickly at the paclet level. I don't have an old version of M- to look at, so I don't want to try to prototype a full browser, but getting the data itself so it could be queried via Dataset is not so bad.

@AlexeyPopkov @Nasser see this for details on how to build such an index. A new help browser could simply import the doc notebook and embed it, so it'd mostly just be a click-through search interface.

@bobthechemist You can use ImageSize -> {(l2/(2 Pi)) 800, 200} on your Graphics object.

@Nasser @AlexeyPopkov I went yet one step further with it: mathematica.stackexchange.com/a/157318/38205

That implements a proper click-through help browser

@b3m2a1 wow, that looks really like the old 5 help system. Amazing. Thanks.

Glad to help. It was surprisingly quick to knock up.

And I think getting a better way to search docs out of it too was a nice bonus

@b3m2a1 is there something I need to do for 11.2? I run your command but for some reason it does not display as you show in your answer. May be there is a conflict somewhere? This is screen shot

Now a new notebook opens:

Let me check. I remember there was a minor glitch in loading package, but I thought I fixed that before commiting to github.

Ok, thank you.

Yeah looks like the index isn't building properly. Give me a moment to debug.

@Nasser Okay that was a problem with contexts and caching. Both should be fixed and you should be able to try that command again without any changes.

If it's building properly it'll take a minute or two to make the index. But it caches that so future loads are very fast.

@b3m2a1 I restarted new kernel and run the command again and now it comes up. Thank you. But I have couple of questions: what commands/help is it supposed to cover? It now has few topics. I assume you only build it for few commands? Not the full 11.2 help system? Would it be possible to make it cover the complete help system? Here is screen shot

Also, would it be possible to make the windows extend using the mouse to see now? Now that you are making us a new help system, we will be asking you for more :) Thanks again, this looks really nice.

So basically it works off of the metadata I can scrape off of all the doc files on the system. From that I can get a doc page type, its context, it's URI, its build date, history, keywords, and I think the pretty much covers the useful stuff.

What would you want me to extend it to?

I could certainly expand that. I picked 150 pretty arbitrarily.

@b3m2a1 I mean, how to use it to see all the commands in 11.2? The full help system that comes with 11.2, all the 5,000 commands or how many there are. Does your help cover all of these?

Oh it should. Go to Symbol > System`

Oh huh that's out of alphabetical order

Let me clean that up.

@b3m2a1 I looked at Symbol-> but do not see System. I see Symbolic

I checked again, and it is missing System`

You're right I just check the core dataset and it's missing. I don't entirely know why, but it's something to do with the paclet loading no doubt.

Should be easy enough to fix, though.

Ok, thank you.

@Nasser the new index takes significantly longer to build (obviously) but it seems like it's building fine. Once I've confirmed that I'll push the new code to GitHub.

@b3m2a1 It is OK if it takes long time to build, since this is one time only when starting mathematica. But if you could make the small windows at top little wider to make it easier to see what it inside, that will be great. Thanks.

@Nasser yeah I bumped them up to 250 from 150

@b3m2a1 I assume the way you are doing this, it is not possible to have user adjust these windows width using the the mouse? Is this done using GUIkit?

It's a DockedCell otherwise they could be adjustable. At least I'm pretty sure a docked cell nixes that. I'll test it.

it looks like the first window does not have to be very wide, but the second and third need to be wide, since the content there can be wide.

@C.E. That did the trick - thanks.

@Nasser Good news, the resize works. The new version should be up and running, too. You'll need to clear your old cache. So reload the package from GitHub with Get then run OldHelpBrowser`Package`clearCachedDocumentationData[].

@b3m2a1 great. Will try it again once you give the green light. I am excited by your new help system.

@Nasser try it now. I just tried quitting, reloading, clearing, and rebuilding and it seems to have works.

Here's what you'll want to do:

Get[ "https://raw.githubusercontent.com/b3m2a1/mathematica-tools/\
master/OldHelpBrowser.wl"];

OldHelpBrowser`Package`clearCachedDocumentationData[]

OpenHelpBrowser[]

How do I see the full partition table? I keep pressing more to see more, but there's over 700 lines. I don't care how long it takes to load...

@Piomicron use MatrixForm or TableForm

@b3m2a1 thanks again. I tried it. It took only 10 minutes to build which is no problem. But the Symbol->System` is still not there? Is it supposed to show there now? Here is screen shot. Also the adjusting of window sizes works very well.

It should be sorted to the top

Was it not there?

Opps, you are right. It is at top. strange.

I did that intentionally

I figured it's the most common thing people would want

I see. Good idea.

@b3m2a1 but why is scrolling down seems to be really slow? I mean, clicking on slider using the mouse seems to be slow. But holding the slider with the mouse and scrolling works, but sometimes that goes too fast. So to scroll down slowly, I use the mouse and click on the bottom arrow. But it is slow that way and it hardly move. Do you see on your end also?

It looks like I have to click 4-5 times for it to scroll down by one entry.

@b3m2a1 This is really great, thank you very much! The only significant but missing thing is the seacrh field where user could enter a command and immediately have it selected in the navigation tree.

@b3m2a1 as I said, now we all will be asking you for more features for this :)

This now will replace using the 11.2 main help pages for me.

@AlexeyPopkov search shouldn't be too bad to implement assuming the InputField and DockedCell play nice. I've gotta do some other work for a bit, but I'll come back to it. @Nasser I'm glad it's useful. I'll try to add requested features when possible. As for the scrolling I don't really see it on my end.

@b3m2a1 another thing I notices. Sometimes the place where one adjust the width of the window goes out of view and do not know how to bring it back. This happens as I was adjusting the windows sizes. Here is screen shot

But it was easy to close the windows and re-open help again and now it is OK

@b3m2a1 can you scroll down the list of symbols using the click the arrow down method and not notice it slow? I mean like this:

on windows 7, I click the mouse on the lower arrow to scroll down slowly so I do not miss an entry. But it does not scroll well. Have to click few times for it to scroll down once.

If I use the bar to scroll down, it goes down too fast.

@Nasser unfortunately all I have is a Mac and it's smooth on there. I tend to use the little bubble in the center of the scroll bar or use the two-finger scroll thing that I think corresponds to a mouse's scroll-wheel (at least in tkinter it does).

@b3m2a1 clicking the mouse on the little space under the bar makes it scroll down too fast and one misses entries. Using the bar itself to scroll down is also too fast and hard to control. But click the mouse on the arrow itself is only way to make it scroll one by one but it is too slow on windows. This mainly happnes on the system` since it has so many entries but that is the one which is most needed. Here is screen shot

@Nasser I'll try to think of things I can do about it. I think it might be a ListPicker issue though. Try this:

ListPicker[Dynamic[a], Names["A*"], AppearanceElements -> All]

That's basically what I'm using here