@Szabolcs Nope, not fixed :(

Module[{foo$}, Hold[foo$]]
Module[{foo}, Hold[foo]]

Isn't it inconsistent? Should the first one have double dollar? Hold[foo$$9710]

Is there any topic about dolars as names endings?

@Szabolcs Even I use your IGLayoutKamadaKawai,I cannot get your this graph

@ArnoudBuzing Aside of been available forever, as pointed out by @C.E. there is a need for a way of storing data that can be previewed and with tracked changes and that does not rely on external services or obscure code. Despite the elegance of your solution, and assuming permanency, still there is no change tracking and no preview that could rule out malicious code.

@yode I did that by taking the output of this, measuring the edge lengths, and setting them as edge weights. Then I used Mathematica's builtin GraphLayout -> {"SpringEmbedding", "EdgeWeights" -> True} on that graph to refine the appearance. It interprets weights in a different manner than IGKamadaKawai (but it is also a Kamada-Kawai variant).

Sorry, I'm running a long calculation now so I can't give you the actual code at this moment.

Is there any way to change a key in a Dataset or nested Association without having to copy the entire Dataset?

Or is it not really "reassigning" when you do ds = KeyMap[#/."old"->"new"&]@ds

@Edmund There is no official way of updating an association without making a copy. I don't know of an unofficial way either, but there might be a (dangerous) one out there.

A copy of an association will share many unchanged nodes with the original. You can get an idea for the implementation structure of an association using Internal`AssociationNodes, e.g.:

Range[64] // AssociationThread[#, 10#]& // Internal`AssociationNodes // TreeForm

Crudely speaking, if a key or value is changed then only the path from the root of that structure to the changed element is newly allocated -- the rest is shared.

@WReach What about AssociateTo? It changes an existing association. If it copies it on its way, I don't know.

@WReach So that means if I have a large dataset and I evaluate a KeyMap to some key at some level and then only that path is copied. That is a lot better than the whole thing being copied

@WReach Is that correct

@halirutan To the best of my knowledge, it follows the same copy-on-write policy as arrays.

@Edmund Yes. For small associations, the whole thing is practically copied. For large associations, most of the data structure will be shared.

gotta run to a meeting...

@WReach thanks

@MichaelHale @MichaelHale could you answer me a riddle about the Wolfram Data Repository? I currently use Figshare as it's the only repository I know of that provides versioning DOI (a la doi.org/10.6084/m9.figshare.3761562.v319). Does the WDR provide such a feature?

@MartinJohnHadley I was wondering about versioning as well. If I look at existing entries, they can have a DOI link that then directs to the datarepository.wolframcloud domain. My guess though, for backwards compatibility and simplification reasons that each publication in the data repository will be essentially static. Maybe some minor cosmetic modifications will be allowed, but nothing that would break existing code that references that data.

@MichaelHale Bah. I can't trust myself, or the people I tell to do "Open Access and reproducible research", to put up the correct or full dataset first time. I'll stick with figshare until this is more clear. Zenodo will have versioning one day github.com/zenodo/zenodo/issues/34

That's why I called mine (not approved yet) "Video Games Until April 2017". Obviously it would be nice to just say ResourceData["Video Games"] and I can update it whenever, but I think they are leaving that type of functionality to data sets that are being integrated as official entities.

Hopefully one of the cosmetic modifications they will allow will be include a link to the most recent version of a dataset from the doc pages of the older versions.

@MartinJohnHadley Although it is entirely possible they will weigh the demand for allowing updates heavier than breaking potential existing code. Or perhaps allow them for a window of time.

@MichaelHale I understand the fear in breaking computation, but a working versioning DOI makes this less of an issue. The canonical doi(.org)/10.xxx/wolfram.xx.data is the most up to date version, and may have breaking changes but doi(.org)/10.xxx/wolfram.xx.data.vX continues to exist and computations/API queries can be made against that version of the data. Knowing what I know about where some of the WC lives, they may need to keep the overall size of WDR small before they're confident in it

@MartinJohnHadley Yes, that would be nice.

Maybe they could expand the elem parameter of ResourceData to support versioning.

@MichaelHale that would be lovely. Thanks for indulging me in flogging my DOI dead horse, look forward to playing with the video games (data).

@Edmund @WReach @halirutan This seems relevant: mathematica.stackexchange.com/a/141916/12

@Szabolcs The documentation page issue is still there as well.

Someone with 11.1 or 11.1.1 want restart his kernel on Linux? There is a new syntactic sugar for this:

Block[{$RecursionLimit = 100000}, x = x + 1]

Nope, is not new.

Is there a manner of function chaining that does not result in a messed up indentation in the FE?

@Szabolcs Composition/RightComposition?

How is the little blue box called and how can I turn this thing off?

The Linux front end is messy enough

@halirutan I do tend to accidentally click on the circled "i" a lot. I wish it was a mm or 2 lower.

@MichaelHale I hate this thing

And then the times I do want to click on it it never appears, so I just have to press F1 anyway.

@MichaelHale EXACTLY!!!

It seems to be called "Function Template" but there is no option for it.

I really need to file a bug report. The Linux FE is so broken that I barely can stand it anymore.

@m_goldberg Could I have a look a glance about your answer again?

@halirutan On Windows at least, the top of the box is usually directly aligned with the text baseline. So I often accidentally click it when triple clicking to select sub expressions or whatnot. Then I just investigated trying to get it to pop up intentionally. It seems that if you have the cursor inside of the text you want to get help on then you have to move the cursor outside the text and then back inside for it to show up.

@MichaelHale It seems the only way to get rid of it is to turn CodeAssist completely off, but then, you have to press Ctrl+K intentionally to get auto completion.

@MichaelHale As I said, it's only one annoyance. Even in the new version 11.1.1 they didn't fix that the code shuffles when you only navigate through it without typing anything. Just because Wolfram broke something with the font or highlighting

Can you imagine how it is to edit 50 lines of code when you think there is a parade going on with all your characters?

The thing is, when I turn this off, then I loose highlighting of matching braces as well.

Ah. I haven't seen that but it does seem annoying.

Going to upgrade to 11.1.1 this weekend.

They did fix 4 bugs I reported in 11.1.

@MichaelHale Yes, I know, because they were busy fixing a bug in the doc-center on Linux I reported about 5 years ago :)

Just an aside,they fix my three.