About to head to dinner, but regarding the flea, jacket, roller blades game, a veterinarian said it's impossible to make a game starring a flea that is enticing. I think the flea should be a messenger flea between different flea colonies on different animals. It uses roller blades to go faster and the jacket to stay warm when scouting at the top of a dog hair.

@rm-rf. Beg pardon. Do you mean I should check "Create and maintain in version specific front end preferences" in both V9 and V10?

@MichaelHale I think your vet might be biased on this...

@rm-rf. The only code in my Frontend init.m is SetOptions[$FrontEnd, "VersionedPreferences"->True]

@m_goldberg Yup, that's the outcome of that setting, which means you probably did that a long time ago :)

You're all set.

@rm-rf. You're right. So I'm probably all set and should just leave things as the are?

Yup

@rm-rf. Thanks for your advice.

I'm off now.

my first V10 kernel crash.

@OleksandrR. No problem. If something like this comes up again, I'm sure rm or MrWiz will happily convert you post back from CW.

@halirutan hope they won't bring that crazy feature back though!

@Szabolcs I saw your pull-request. I will look at it tomorrow since now between your code and me, there are about 5 Cuba Libre I already had.

That seems like a bad combination.

(Oh I guess in my Zustand I should write like: tath seesm lke abed cmbinaitin...)

Good night all.

fyi, numbers of symbols in V10 counted is 8579, in V9.01: 7919. About 650 more symbols. I could not get the list of all the functions yes, the script I used for V9 does not work in V10, need to find out what changed. So I just now have count of symbols in each context 12000.org/my_notes/compare_mathematica/V10/main.html

any one knows if Mathematica graphics are automatically rendered on the GPU? (hardware accelerated) and if not, is there a way to configure it to do so?

Any one know if there will be a discount for owner of V9 while upgrading to V10?

when i try i get the same price on purchasing a new licens as i would get if i where to upgrade my old. (my license are a Home user licence)

@stian1986 I do not. I leap-frogged v8 and v9.

I'm afraid there is not much of a saving offered when upgrading, at least for Home Edition.

:-/

@stian1986 I paid 95 euros plus taxes...

Also, upgrading from v8 to v9 with promise of v10 included was on same price range in the spring...

Multiple Home Edition licences, I don't know if it makes sense... especially for a home of one. :)

@stian1986 I think there is

Heyas. When I start a fresh kernel and give the command

Upgrading late to v9 had a chance of getting v10 for free, I think.

Names["Global`*"]

the output is

{rhs, z}

Is it me? :P

I've been looking through all my init.m files but I can't find where these variables leak

{"list", "msgs", "rhs", "z"} on v10... ehm.

What is this namespace pollution?

I know rhs occurs as a variable in the codeformatter

yeah, exactly!

before I also had 2 variables in there that I had caused myself

(I'm v10 also)

Thanks for checking

I searched a bit through

Column@CurrentValue[$FrontEnd, AutoloadPath]

What about it?

found nothing, really :P

Ah, ok.

also FrontEnd`FileName[{$InstallationDirectory, "AddOns", "Autoload"}] does not seem to exist (I guess read FileNameJoin instead of FrontEnd`FileName)

I hope no system function depends on the existence of z or rhs in Global. For example if you set

sys`f[Global`xxxx]:=Global`xxxx

and then remove Global`xxxx, sys`f // Information will look all funny, and I think it can lead to trouble if you use Save. The function still works with the definition with Removed in it though

These strange things like z and rhs also occur in WPC. Also, they have some others. I wounder if it's intentional or just an accident.

I notice v10 Classify["Profanity", ...] recognizes some Finnish profanities, but not many. Anecdotally interesting, I say.

After all, these phrases have nothing to do with practically any other language.

Ah you evaluated Names["Global`*"] on the cloud? Nice test

That'd be {"arg","CloudExportFormat","interpreterJavaDir","rhs","UndocumentedTestFEParser‌​Packet","wolframapps","z","$m7res"}

Like, eh. Why such stuff?

@kirma maybe it just got left in. running strings on the executable also reveals random stuff just left in

@kirma ah, I got a lot more stuff, when using CloudEvaluate

btw I've made a question about it:

Ah, haven't tried that. Now that I think of it, I don't know if I have license for that...

@kirma time to try? :P

@JacobAkkerboom Lots of stuff there!

Like dozens of symbols.

Yeah, and after calling CloudEvaluate my Global space back home is also dirtier than before

Hm.. When I give Trace more than one option, all but the first options turn red. But they still have effect. I suppose they become red because SyntaxInformation[Trace] tells us that there can be only up to two arguments. Also seems like a bug to me

@JacobAkkerboom you can do

Unprotect[Trace];
SyntaxInformation[Trace] = {"ArgumentsPattern" -> {_, _., _.}}
Protect[Trace]

if you want

@acl wait.. huh? It seems a bit broken in general.. No red here:

Solve[x == y, x, y, z, y, adf, gadf, gadfg]
Apply[1, 2, 3, 4, 5, 6, 7, 8, 9]

@acl in 9, it used to be

{"ArgumentsPattern" -> {_, _., OptionsPattern[]}}

Ah but Function does turn red:

 Function[1, 2, 3, 4, 65]

no difference between V9 and V10... still I'm confused

Hmm.. I'm probably not going crazy. In the docs of SyntaxInformation, there is an example that should turn red, but doesn't. It even doesn't turn red in 9, and I expect installing V10 somehow interferes with V9.

Nevermind that last sentence.

But if somebody could tell me if those expressions involving Solve and Apply above really didn't use to turn red, that would be much appreciated.

Only 4 more hours until I've got v10. Gotta love DSL in the USA </sarcasm>

Heya @sblom

I find it weird and uncomfortable that applying Normal to a Dataset gives associations, but applying Normal to associations gives a list of rules. Aren't these two effects of Normal going to interact badly at some point? I wish there were two separate functions for these two types of conversions.

Also, Normal@Dataset[...] does convert the Dataset, but Normal@{Dataset[...]} does not convert anything. This is not how Normal works in all other situations I know of (including SeriesData, Graphics, Association, SparseArray).

@Leonid What do you think about this? ^

upgrade seems to be cheaper now i tried my V9->V10 and they quoted me 65£ for home user license, ref my earlier mullings about they asking full price for a upgrade

@Szabolcs I think this is rather logical, at least the first two cases. Frequent semantics of Normal is to reduce some object to its core constituents, and in this sense, the behavior of Normal for both Dataset and Association seems logical. Since Normal is always some kind of reduction, in either case this is not an equivalence transform, but you can reconstruct back both Association and Dataset by using the constructors.

@Szabolcs The last case (Normal[{Dataset}]) - I don't have a strong opinion here. But, in general, I don't think we should expect from normal to perform some inner scan for the constituents which are amenable to it (in other words, I actually don't support the idea of Normal to be recursive, except in certain special cases. The reason is that while one can easily build a user-level version of Normal which would be recursive, from a non-recursive one, it would be much harder to disable recursion)

@Szabolcs So, in summary, I think this behavior is fine.

@Leonid OK, just wanted to bring it to your attention. I also think that it's good that Normal does not recursively process Dataset.

It looks like this withRules function I mentioned so many times on this site is really broken ... thinking about how to undo the damage.

@Szabolcs There have been several internal discussions about it, and IIRC, the consensus was the same - recursive Normal is bad in general.

@Szabolcs Did you find out why? Was it due to the use of :=?

@LeonidShifrin I agree about that for Dataset, but in many cases it's very useful actually. It's quite useful with Graphics, which have nested GraphicsComplex. And also this (at the very end).

@Szabolcs That's why I said "in general" :)

Isn't the real problem that very different uses are all tied to the same function?

In some cases the recursive behaviour is quite desirable. But once it's made recursive, it may affect several different data structures that are part of the same expression. It may e.g. convert both associations and GraphicsComplex within the same expression. Then one has to be vry careful and use ReplaceAll to only apply it where it's needed.

@Szabolcs I think it is. But in Mathematica model, that's what we have - generic functions (rather than say objects). In the OO world, the same method names are Ok because they are always used on specific object instances. In Mathematica, one can only come up with a new function with another name, or leave with strongly overloaded general function.

@Leonid Something related to that. What do you think about the "methods" and "properties" that can be used with InterpolatingFunction, FittedModel, etc.? E.g. if = Interpolation@Range[10] and then if["Domain"]. These are very rarely documented, but are commonly used internally. I think they're only documented for FittedModel.

They still seem to be a good way to tie names to a certain kind of object, and not have too may things in the system namespace.

@Szabolcs will be right back, have to change browser - Safari has some very bad bugs, almost can't type

For example, Length@Names@"MeshCell*" is 9. When these prefixes (like MeshCell) start to proliferate, it's a sign that some sort of object oriented approach that ties methods to a certain class would be needed. Just for organization.

@Szabolcs Ok, first this note (typing fourth time now) about recursive functions problem (like for Normal) - it is all due to the "over=transparent" nature of Mma w.r.t expressions and their inner parts. Often this is a big win, but we also pay for it in many ways (broken lexical scope in Function is one example of this payback)

@Szabolcs They were just some meaningless parts of words - I encountered a pretty bad bug in Safari, obviously. Now switched to Firefox.

@Szabolcs Yes, I agree, they are a good way to do things. This is pretty similar to properties of objects in javascript. But always using strings is a bit annoying (purely on the syntax level), in js at least you can often omit strings (it auto-stringifies)

@Leonid Regarding withRules, do you see anything wrong with this?

SetAttributes[WithRules, HoldAll]
WithRules[rules_, expr_] :=
  First@PreemptProtect@Internal`InheritedBlock[
    {Rule, RuleDelayed},
    SetAttributes[{Rule, RuleDelayed}, HoldFirst];
    Hold[expr] /. rules
  ]

@Szabolcs This is inevitable, because the more Mathematica moves into applied domains, the more there is a need to keep collections of pre-computed properties, which can be easily accessed. Mathematica would certainly benefit from some kind of object model. Recently I have been thinking that prototype model of inheritance (like in js) would suit it better than the classical one.

@Szabolcs What's the purpose of Switch?

The LHS of a rule should not evaluate. The RHS should evaluate for ->.

@LeonidShifrin Nothing, removed. :-)

@Szabolcs You probably want also to apply ReleaseHold to the result?

I used to have Unevaluated in place of Hold. I removed that to make sure the modification of Rule is gone by the time expr is evaluated.

Ok, fine, I see First

@LeonidShifrin First does that. I wasn't sure if ReleaseHold was recursive and I was lazy to look it up ...

Got it. I also use First often for that. Just did not see it at first (pun unintended)

@Szabolcs Why do you need PreemptProtect?

@Szabolcs Do you want to use this inside Dynamic?

@LeonidShifrin This is what prompted my question. What if InheritedBlock gets iterrupted by a pre-emptive evaluation, and the pre-emptive evaluation ends up using a modified Rule?

@Szabolcs Good point. This also applies to other Block-like constructs we often use. I actually never bothered, but your version is probably more careful. I just don't see whether it can ever lead to some kind of locks or blocking. Don't have a good enough understanding of Dynamic etc to see that on the spot.

@Leonid I just broke Association ;-) Look:

@Leonid dropbox.com/s/zsbsyxx8c390gx1/…

(Yes, I was looking to break it, while trying to figure out if keys evaluate further, once they're part of the association)

Maybe Compress needs a separate atomic representation for it instead of decomposing it into a compound expression with head Association and no Hold* attribute ...

@Szabolcs Interesting. This is surely an inconsistent behavior - either Association should update once the value changed, or it shouldn't at all.

This is because if we do <| a->1 |> we get <| 1->1 |> if a had a value.

@Szabolcs I will try not to forget to file this.

But if a gets a value afterwards, the association doesn't evaluate the keys further.

@LeonidShifrin Do you need me to send it to support?

@Leonid I haven't tested, but very likely the same thing will happen if an association is sent through a MathLink connection.

Let me try with Parallel ...

@Szabolcs : This would actually be better, since external reports are often given more weight, at least unless there appears a specific internal discussion thread on the issue.

@Szabolcs I can start such a thread too, hope I will remember to do that.

It will also be a sort of a problem with Save. Saving and reloading a variable will do this. Probably the same for WDX, but I want to try MX too ...

As I expected, MX is fine, but not WDX.

@Szabolcs All right, gtg. Talk to you later!

@LeonidShifrin See you!

I can't get SunPosition to work correctly.

It has the sun rising somewhat before 1 am

AstronomicalData seems to work OK with the same details

@kguler SSSTriangle[3,4,5] gives Triangle[{{0, 0}, {5, 0}, {16/5, 12/5}}]. Have you found a simple way to convert this to a MeshRegion which contains a single triangle only, i.e. mesh = MeshRegion[{{0, 0}, {5, 0}, {16/5, 12/5}}, Triangle[{1, 2, 3}]] ?

If we had that, then there should be a better way to solve this with HighlightMesh. But I'm struggling with that too to get a really simple solution. So far I only have:

HighlightMesh[
 mesh,
 MapThread[
  Labeled,
  {MeshCells[mesh, 1], RegionMeasure /@ MeshPrimitives[mesh, 1]}
  ]
 ]

The simpler HighlightMesh[mesh, Labeled[1, "Index"]] labels with the indices, but I'd really like to use a function (RegionMeasure?) instead of "Index".

Hi Szabolcs ...

Hi @kguler! I'm also just learning this new geometry functionality, and it's really bothering me that I can't come up with a simpler and shorter solution for this labelling problem. I was wondering if you came up with anything better or if you will come up with something better based on the above.

The solution you posted of course works well, but it's not simple and general. I'm trying to find the best way possible with the new functions.

Neat idea .. MeshRegion , HighlightMesh are all new to me -- it looks/feels like GraphicsComplex?. I thought working with the coordinates of the Triangle object directly was simple for OP's specific question. I will to study/play-with Mesh-- stuff more to see we can make something like HighlightMesh[mesh, ...] work.

@kguler Do you know of a way to convert an arbitrary set of graphics primitives to a GraphicsComplex format?

I vaguely remember coming across something like that in one those spelunking expeditions - maybe "?*`*ToGraphicsComplex*" ...

... I get "NDSolve`FEM`GraphicsPrimitiveToGraphicsComplex" -- maybe worth looking inside?

Just tried the code pieces you suggested --- MeshRegion[#, Triangle[{1, 2, 3}]]&@SSSTriangle[3,4,5][[1]]` and the rest work beautifully!!

Any cons to using Sow with tags in a function to provide optional output?

@Szabolcs, thank you for the many TILs re these new functions. If you go with "Index" for labeling it can be also be done using the MeshCellLabel option: MeshRegion[{{0, 0}, {5, 0}, {16/5, 12/5}}, Triangle[{1, 2, 3}], MeshCellLabel -> { 1 -> "Index"}]

Hello everybody

Anybody here?

Well I had a question or two about Mathematica's Web Cloud program

@Rojo youtube.com/watch?v=VXsKIM0_WZU

Hello

@mildused It's better to just ask.

Oh okay Thanks. They say that we get Mathematica 10 when we pay for a subscription to the Web Cloud. If we have Mathematica 10, do we get one year of web cloud? Probably not but just wanted to make sure