@b3m2a1 Doesn't work on Linux with Chrome. None of the examples you posted.

@halirutan ;____;

@b3m2a1 At least the white screen I'm seeing is fast :)

Lol

It takes a while for mine to load. Here's what I'm seeing:

(sorry converting screen recording to GIF)

gif here

All of the JS and stuff that makes this possible is total trash though. I really did not write it well.

@b3m2a1 Well, that's something.

Does Chrome on Linux dislike ridiculously long URLs? All of the code is base64 encoded and cooked in there :)

I already turned off adblock.

I like the error message though

> ... was preloaded using link preload but not used within a few seconds from the window's load event. Please make sure it wasn't preloaded for nothing.

Lol yeah that shows up for every cloud notebook

Okaaay :)

Yup

Hmm maybe I'm screwing up the request somehow or like botching my requests in some subtle way

If you want to see it here's the JS I'm using there: wolframcloud.com/objects/b3m2a1/javascript/wlfiddle.js

Oh goddamnit it's literally just complaining about the callback loop I set up to wait for the window to be ready...

I guess I need to be a bit better there

@halirutan can you give e.g. this another try and let show me the errors it raises? I tried to handle those errors

@b3m2a1 Now I see at least "click to load Notebook".

Sigh...means it's still not working. That button will not do anything.

I don't get why it thinks I'm making a cross-origin request on your browser but not on mine...

(or on my phone even)

@b3m2a1 The error message is more conclusive

l o l

I'm an idiot

Wait how is it still working for me...

Does anyone here actually know how to write JS? :) I have literally no idea if what I'm doing is good or bad or makes sense or not and do not know how to test it.

@b3m2a1 I don't. I can read it to some degree but give me an empty editor and I'm lost.

Hmm and sometimes it looks like the first time the page loads it fails but then the next times it works.

What this means I have no idea

Ah shit that means I need to add a longer delay time

I bet it's caching

Who knows how you can tell when the notebook is ready though...

Meh I did some updates. I realized the WLFiddle stuff wasn't using the shared JS

But now it is and works a bit cleaner but still not, you know, cleanly

@b3m2a1 Do you have to give me a new URL or can I use the old one?

The old one will work

It's all in the stuff after the URL really

@b3m2a1 Then there is something wrong with my Chrome. Still no output

@halirutan I managed to make it work (i.e. fail) on Safari :)

Both Firefox and Chrome consider www.wolframcloud.com and develop.wolframcloud.com to be the same origin but I guess not Safari or Linux version of Chrome

Oh but now it's working even without that...I clearly don't understand web programming.

@b3m2a1 So I don't need to do anything, just release, and it'll be transferred to the paclet server automatically about once a week?

@Szabolcs assuming the release can be pulled as 'latest'

Your current case with IGraph/M is a bit odd since it's a prerelease

But I can also probably pull that with the GitHub API...

But in any case if you make a proper release I can move it to 'latest' and you'll never have to do anything

I realized it was just too much to ask people to ping me. Like how would they remember? So if you use the release mechanism or a paclet server (the latter for my own stuff) you'll never need to do anything.

@b3m2a1 No, it's not possible to get the latest prerelease automatically. It is necessary to filter manually. But I can guarantee you that the first entry in the JSON will always be the latest (prerelease or not) for IGraph/M. If all goes well, I'll somehow manage to reach the standards I set for a proper release by the end of February ... sorry about the mess of a perpetual prerelease.

I can guarantee this because the auto-install script also relies on it, and has been relying on it for a long time.

@Szabolcs It's not really an issue. For speed reasons I prefer to just query for a single release but I'll put in some special logic/settings for cases like yours :)

Each build I check all my GitHub release paclets for updates against the local cache I have and so speeding up each of those requests is desirable of course.

If this thing ever took off I'd need to be smarter about how often I check individual paclets for updates or something. But my sense is that I'll never be drowning in paclets and if I start to I'll just revamp.

@channel here's a powerful thing. We can embed a fiddle (which is itself a highly embedded thing) so that all the work of cooking the URL can be pushed onto the Wolfram URL shortener. Here's an example of that at work: wolframcloud.com/objects/b3m2a1.testing/fiddle-test.html

There are of course little things I could do to make this nicer. And I chose to use the "Delayed" fiddle version of this: github.com/b3m2a1/mathematica-tools/blob/master/WLFiddle.wl

But you can do whatever you want basically.

Assuming I don't get royally screwed by WRI for this we've just opened up a pile of opportunities for sharing code in our websites :)

It was just too interesting not to do though...

@b3m2a1 It's cool, but putting it up on a website means that visitors can spend your cloud credits, correct?

@C.E. that's where this is really cool. I use a sandbox notebook.

It's free.

This is why I think WRI might get...let's say grouchy...with me

It's using this: develop.open.wolframcloud.com/app/view/newNotebook?ext=nb

I actually poached it from the Wolfram U site

Or well I poached some settings from there to make it nicely embeddable

Long story short. You can do this without dinging my credits (I think).

@b3m2a1 ok, then it is highly usable from our perspective, nice. Only problematic from WRIs perspective.

(and in a sense, if it is a problem for WRI, then it is also a problem for us. But that aside.)

@C.E. in other web dev news I finally included pagination in my web builder framework, e.g.: wolframcloud.com/objects/b3m2a1.paclets/PacletServer/…

The fact that it took so long makes me wish Mathematica could work off of standard templates in Jinja or whatever

But ah well. After I do all of this XMLTemplate work hopefully no one else ever has to.

cool

I'm actually very proud of how good the paclet server website looks now:

And the build process is fast and flexible. Who know this whole scheme actually turn out to work reasonably well?

Is there a native way that supports 'return early' coding style? It looks like one has to go with Throw/Catch, right? I don't trust Return or should I?

@Kuba - I regularly use the 2-argument form of Return

I know it's not documented, but I can assume it won't stop working in the future because I can see how much of our internal functionality depends on it. I actually never use the 1-argument form

@JasonB. Thanks, will give it a try. I always feel bad for throwing when return is needed.

@JasonB. @Kuba What is the two argument Return and what is the problem of the normal Return?

this is why the knowledgebase should be a triple store with wikidata ties mathematica.stackexchange.com/questions/189689/…

I swear, when 12 comes out with the SPARQLLink stuff I am going to figure out how entities work and try and make that happen

when is 12 coming out anyways, surely it's soon. the livestreams have dried up a little which makes me think SW is in marketing meetings or something :P

(Avert your eyes to my last few messages, @JasonB. ;) )

@JasonB. ah, I see. That's nice.

That's really nice functionality. I don't know if that paradigm is in any other language.

oh wow mathematica.stackexchange.com/questions/189446/…

This periodic table shows the country of discovery for each element. I saw it on 9gag this morning and thought that I should try to remake it with entities :)

It is noteworthy that according WRIs entities, Poland has two discoveries. This differs from the version at 9gag, in which these were given to France because Marie Curie had been in France while discovering them. A lot of Polish people were up to arms over it in the comments :)

This would have pleased them, I think.

@C.E. since it is "country of discovery" then I won't rage :)

@C.E. On the subject of "what is wrong with Return"... the scant documentation documentation says that if the second argument is omitted then the return point will be determined using heuristics.

Those heuristics have sometimes changed. I know of a specific example where Return was broken for a time in v7 and v8 but fixed in v8.0.4. I've never had a problem with the two-argument form (fingers crossed).

@WReach Do you see any point in using Return and Throw if there is a Catch @ Module[...?

So if Throw is not meant to bubble up. But only indicate an exception/error as opposed to another branch.

@Kuba I would probably use Return if the exit point was "nearby". For long non-local exits, I would Catch/Throw -- with a localized tag to make sure I don't catch anything by accident. The boundary between the two cases is admittedly vague.

I come from a LISP background so I don't automatically associate Throw/Catch with error handling :)

More like setjmp/longjmp in C.

@WReach I come from here :P so I don't know but I use Throw/Catch for control flow because I trust it, it works and it is short.

But I am revalidating my practices whenever I start a new project.

@Kuba Throw/Catch is explicit and I concur that explicit is good.

@WReach Ok, thanks.

@CarlLange +1. I had the same thought when I was watching the various SPARQL/Entity LiveCEOs. A couple of times of I was ready to start shouting it at the screen ;D