@Kuba also did you know "DefaultText" is a possible second argument to ExportPacket? It pulls the format from TextClipboardType, which can be "Package" in which case Mathematica actually respects CurrentValue[$FrontEndSession, {"ExportTypesetOptions", "PageWidth"}]

Another IDE-related blog post this time on making tutorials and stuff

@VitaliyKaurov I showed how to use the BTools MarkdownToNotebook function to turn MD into a notebook in that blog post too.

I am looking at this question, the first drawback that @AlexeiBoulbitch mentions is that Mathematica does not have a nonlinear FEM solver. Since, V12 that is no longer true. I put a note under Alexei's post about a week ago asking for permission to strike through that sentence. Alexei, unfortunately, has not logged on since about two weeks.

I could just make the change myself but I absolutely want to avoid the impression that I am editing posts to the liking of WRI. What do other people think: should I wait a little longer for Alexei's permission or should I just go ahead and make the change?

@user21 I'm sure he doesn't mind. It is an objective information and no one would think that there is a hidden WRI agenda behind the edit.

I would say go for and make the edit. It's far more important that people know that this is possible in 12.

Additionally, I might have an subjective opinion on this, but you and Leonid appear to me as two WRI employees that often seek the critical opinion on their work here. Not sure how often I have read something along "I worked on XXX, I would love to get feedback" from you. I consider you an important asset of this community although you are working for WRI.

@halirutan, OK, I made the change. Better to ask and make it public then to be sorry later. If anyone objects that change let me know and we will find a solution.

@user21 Nice.

@halirutan, thanks for the kind words. I see my interaction here as my part of a direct customer service.

@b3m2a1 how is DynamicLocation not an obvious flashy feature they should have intention to document? Interactive dashboards where many plots are supposed to be created automatically rather than polished manually for publications it is very useful.

@Kuba yeah not exactly sure. If I had the time and energy I'm sure I could think up lots of fun things to do with it. Lots of cool potential for animations or for embedding plots inside other plots and things.

I'm completely stuck

How the hell do I do evaluation control in Mathematica/

The documentation is dire on whether: once a function is evaluated its arguments are evaluated once (?). Or how I can simply evaluate a function aat a depth n of evaluation

@1010011010 could you be more specific in what is your goal?

@halirutan @JasonB. too!

@ChrisK Ahh yes, our beloved chemist! How could I forget that?!

Two livestreams today, guys! The first Twitch Talks (developers demonstrating new functionality) is at 12p central. Then Stephen will have a Live CEOing stream today at 1:30p.

@Kuba Do you know the Lieb-Liniger Bethe Ansaz model?

Basically I'm solving each state by punching holes in some N particle ground state one by one

So the idea is that I'm solving for some N particle state for Nh holes, which leads to N over Nh combinations. I'm doing it in a bunch of Do loops, which I wish to create using Fold

Fold[Do, <stuff>, <iterators+bounds>] works fine, until I want the iterators to depend on each other

So the next step is to evaluate the iterators from the outside all the way in, so that at every point all the values attached to the iterators (inside/assigned by Do) are known

I thought that maybe Nesting Hold's and ReleaseHold's, or Evaluate and Unevaluated would do something like that for me

But boy was I wrong

And did I go down the rabbit hole

FYI, ListLinePlot with many points and dashed lines crashes MMA 12 under Win10. I confirmed with support. For now I'm not using Dashing when plotting large datasets. The following reproduces the crash for me. You may need to resize the output to trigger it:

@gdelfino Could you narrow down the bug? After all, ListLinePlot is only creating a normal graphics with lines. Would this be an issue, if you created the line-strip by yourself or is it a bug in ListLinePlot?

No. It is not a bug in ListLinePlot. It is in the front end (Windows) when rendering long LineBox with dashings. I had this notebook that just opening it would crash the FrontEnd and that is how I found it. The code above produces something similar. Also notice when you resize the resulting plot that it is much faster without the dashing.

@gdelfino - is it possible to reduce the number of points? ListPlot and friends ListDensityPlot and ListLinePlot do not resample the data, they will create a Graphics object with as many points as you give them. In your case, I think the front end is having trouble drawing 40,000 dashed lines

even without the dashing, while you do get a result, it really slows down the front end just to have it on screen

I agree that the FE hang from your input with the dashes is not good

Well. The thing is that it crashes in version 12 but it doesn't crash in 11.3.

interesting - in version 9 I get it returned instantly but in 10+ it really gets my CPU fan going and I just force quit it out of impatience

you should report it as a bug

@halirutan what resolutions should I try to get this to look good at?

I currently have this at the size of the base image:

It is reported under [CASE:4225009]

@JKlug Thanks as always for the heads-up!

Really enjoyed this tech dive into Asymptotics: twitch.tv/videos/418476758##

@b3m2a1 What I really want to try and cannot do with the standard SVG export is:

Give all polygons a small edge of the same color as as its face and if possible the caps-style of the lines should be round.

The size of the image is not important, as I would combine it with the font in Inkscape or Adobe.

@halirutan ah I see. I can definitely make that happen. Had a few new plot-size-calculation bugs to clean out but it's working fine again. I can add a quick rule to handle CapForm as a directive too.

Oh turns out I already had one

@b3m2a1 Nice. When I use the standard SVG export, there are clear visible artifacts when I open the SVG in the browser

And it's absolutely OK if I do the conversion myself, when you fixed the SVG.m on your GitHub repository.

@halirutan okay the current version should work. Here's how I've been doing my conversion:

svgString =
  ExportString[
   ToSVG[
     CloudImport@CloudObject["user:b3m2a1.testing/se_logo_sample.wl"] /. _AbsoluteThickness -> Thickness[.003],
     EdgeForm[CapForm["round"]],
     ImageSize -> 40
     ] /. {
     {r__Rule} :> {
       r,
       "mix-blend-mode" -> "hue"
       }
     },
   "XML"
   ];
Export["~/Desktop/test.html", svgString, "Text"]

The EdgeForm[CapForm["round"]] is simply a directive inserted into each processed element

And I opted to use a percentage-based thickness instead of an absolute one

@b3m2a1 Let me play wit that. Thanks!

Unfortunately it's not really possible to get a size-adaptive Thickness but I found .008 looks good at small size and .003 looks good when large.

At one point I also toyed with resizing some of the interior polygons to make them smaller. This reduced the skew, but I think there needs to be a different resizing factor for lateral and vertical.

@b3m2a1 Yes, that is an issue I also found. I have researched this in the past and there doesn't seem to be a really good solution to this all

One other thing I was thinking about it it might be able to use the Region stuff to create a single background color (say red) that could be used and then smaller polygons could be layered on top of that

This way if you made things smaller the lines would be less apparent

@b3m2a1 Yes, I also thought about a layered approach, where you draw polygons onto another, but then I'd had to touch the logo code and I'm not sure how hard the problem (esp. the distance-related coloring) becomes.

Oh boy I can't wait to argue, once again, that the Knowledgebase should be just Wikidata during this livestream :D

@CarlLange Have you gotten any response?

@C.E. Nah, not really. It's not really the scope of the meeting unfortunately.

But it should be

ok, it would be interesting to know what they think about it.

@C.E. I agree, but I think they've gone too far down this road and can't see the forest for the trees :)

I can only hope they'll add EntityStore[RDFStore[...]] at some point

or... I can make that happen some evening after a few beers. we'll see.

@CarlLange They certainly have invested a lot, and they've made a very successful product out of it, Wolfram Alpha, that is probably a sustainable company on its own by now. I'm actually surprised to hear that Wikidata could replace what they've created (although this isn't the first time I've heard you say it).

Well, I don't think it could replace the whole thing. The differentiator is that the Knowledgebase has computable properties, which Wikidata doesn't.

Does anyone know how to work these SPARQL functions in version 12.0

@CarlLange ok, just getting the data from there seems reasonable.

@Edmund You asked at the right time :) I'll give it a look in a little while :)

@C.E. Yes, exactly! What's particularly cool is that you can already link from Wikidata to the knowledgebase wikidata.org/wiki/Property:P4839

So in theory we could get some properties from the knowledgebase and others from wikidata, or something, if you want. It's really worth them exploring this...

@CarlLange Thanks so much! Looking out for your answer.

Since I'm about to start walking again, those are the 7 trails (800km or so) of the 42 trails (4000km) that we have left to walk at toughsoles.ie! Just thought it was cool :)

GeoImage is nice :)

@Edmund I think you may need to use named subgraphs for what you're attempting (the second usage example in the docs for RDFStore)

But I'm not completely clear on what your intention is (probably my fault and not yours)

actually, you can Join the two RDFStores, so I think the query needs explanation. Do you know what you want to do in bare SPARQL?

Ah, I see, I understand now. I believe you need to do a subquery for this, but let me read the spec

@halirutan how is this: wolframcloud.com/objects/b3m2a1.testing/se_logo_sample.html

@b3m2a1 I'm already trying things out. In the logo you linked, one can see that the white triangles are very small and that it looks too "overdrawn".

It's a pity that we need to make sacrifices here and play with parameters when the geometry is plain simple.

@halirutan that's probably coming from the path I layered in the background. Used RegionDifference to construct a GraphicsComplex but I think the region doesn't have large enough gaps.

Here's the notebook I've been working in: wolframcloud.com/objects/b3m2a1.testing/se_logo_nb.nb

It's also got some code to shrink various elements and group them by color to give more targeted approaches to the tweaking

@halirutan just did an update where I just reintroduced the white triangles you had in there and layered them on top

Not sure how that'll look against the tiled background but it's probably small enough not to matter...

Here's my current version in context:

@b3m2a1 I just read SVG docu and as it turns out, we need stroke-linejoin='round' instead of line-cap because none of our lines has a "cap".

I hacked this and it looks pretty solid.

Nice

If you want to feed that into SVG.m turns out I support JoinForm too.

@b3m2a1 Ahh cool. I used StringReplace since I was too lazy to look if you had this

@halirutan I'd forgotten if I had it too

This is what I support directly:

$SVGFacePropMap =
    KeyMap[SymbolName]@<|
        Filling -> "fill",
        Opacity -> "fill-opacity",
        RoundingRadius -> {"rx", "ry"}
        |>;
$SVGEdgePropMap =
    KeyMap[SymbolName]@<|
        LineColor -> "stroke",
        Filling -> "stroke",
        RoundingRadius -> "stroke-round",
        CapForm -> "stroke-linecap",
        Thickness -> "stroke-width",
        AbsoluteThickness -> "stroke-width",
        JoinForm -> "stroke-linejoin",
        Dashing -> "stroke-dasharray",

But I think you can also pass any string and I'll just feed it in

@halirutan I think "bevel" might work even better than "round"

Is there a Wolfram SPARQL endpoint for entities? I tried to query the entities with the SPARQL functions and it appears to be downloading the entity graph store.

164.6k values

@halirutan Also I figured out what I couldn't embed my .svg anywhere. I had the wrong xmlns or whatever it is. Now it works:

@b3m2a1 Would have been my next question as I got an error when I tried it to import into Inkscape

Yeah pull the latest version off GitHub

I changed how the viewbox is set too, as it didn't seem to like having a negative minx and miny

@Edmund Not that I am aware. You should post on the wolfram community for this.

I was not actually aware you could download that as a file though.

Here's another take on our logo:

@b3m2a1 The downside to this approach is, that you cannot make the upstrokes of the font pixelperfect.

@halirutan SVG text doesn't line up on the pixel grid?

Ah here is my final proposal:

@b3m2a1 Well, usually with such logos, you correct each character individually.

Nothing like a bit of Comic Sans to show we're serious

@halirutan what do you mean by correct each character? I guess I don't quite understand enough about fonts on the web to figure out why plain text would pixellate.

@b3m2a1 We know the final pixel-size of logo on the web-page. Usually, I create a grid in Adobe or Inkscape of that size. Then, I layout the text so that it has good size and that spaces between graphics and text look good. Finally, I'm looking at each character separately and move them tiny bits so that e.g. the upstrokes of the "M" fit into full pixels.

We had this issue with the first "new logo" that SE provided where the bottom line of the "Mathematica" text did not end exactly on a pixel leading to ugly antialiasing artifacts.