« first day (2555 days earlier)      last day (1919 days later) » 

b3m2a1
b3m2a1
12:07 AM
@Kuba Yeah I don't think it buys us much. It's a pretty vague tag...
@channel I spent a while polishing up the Paclet Server interface (it looks a lot better now) and I included a sidebar on the right. Because it's traditional I'm including some external links in one of the panels and would like input on what links are good to have. Here's what I have right now:
That's a link to the "pitfalls" question, a tutorial I wrote, Leonid's book, and the WRI fast intro. Anything else I should include?
I also have this for the site links:
And if people think there are better things for those I'm happy to include changes there too.
 
 
2 hours later…
b3m2a1
b3m2a1
2:41 AM
@JasonB. can you think of any fun chemistry tasks that would also serve as a nice intro to programming (i.e. stuff that will reward OOP, code reuse, encapsulation, etc.)?
I'm training a new computational chemist
@ChrisK how does this sound as a pitch: 
A number of people in the broader Mathematica community have enjoyed the Live CEOing Twitch streams and have often though of questions that would be interesting and useful to have discussed by WRI in a livestream dedicated to end user input. If this is in the realm of possibility, we can come up with a finalized list questions that WRI could vet.
Or something along those lines. I feel like to coax them into doing this it will need to seem like it's not just going to be a "why doesn't this work" type thing.
Or maybe suggest it to WRI that they solicit input from the broader community for a stream.
 
b3m2a1
b3m2a1
3:19 AM
@Kuba what do you think of this idea: we use the undocumented cloud notebook API to make the equivalent of a "let me Google that for you" website but with Mathematica
2
We can do it 100% (I basically did that for to some OP by sending him here: wolframcloud.com/objects/b3m2a1.testing/…)
 
b3m2a1
b3m2a1
3:46 AM
Here we are:
It's a little bit meh but it works. Just format your query in URL format and submit it to https://www.wolframcloud.com/objects/b3m2a1/LetMeMathematicaThatForYou.html?que‌​ry=...
Here's a more complex example:
 
 
2 hours later…
C. E.
C. E.
5:27 AM
@Kuba I would never have thought to put the syntax tag on a question about a web service API. I also would prefer to use the webservice tag instead of web-access. But I agree with you about the API tag, I think it is simply too broad to be useful.
 
 
1 hour later…
KraZug
KraZug
6:48 AM
Yay, got my Fanatic badge.
 
 
1 hour later…
C. E.
C. E.
7:52 AM
@KraZug Congratulations :)
 
Kuba
Kuba
8:20 AM
:)
@C.E. @b3m2a1 thanks for feedback.
@b3m2a1 I think WRI should be very interested in supporting this idea.
 
b3m2a1
b3m2a1
8:54 AM
@Kuba I realized I could turn this into a JS fiddle type thing, too. Here's a preliminary implementation:
 
Kuba
Kuba
@b3m2a1 While I think it is in WRI best interest I also think you should consult it with them first. It is a thing a legal dept can have objections to, don't you think?
 
b3m2a1
b3m2a1
@Kuba I figure if they complain I can just remove all that stuff.
All I need to do is a DeleteFile and it's gone :)
 
Kuba
Kuba
Assuming this is all they will want you to do.
 
b3m2a1
b3m2a1
Right they might have crazier requests but I figure probably not
 
b3m2a1
b3m2a1
9:55 AM
@Kuba @CarlLange why is web programming so terrible...if I wanted to really have a decent JS fiddle like thing I'd need to do a lot of handling of asynchronous requests and responses in/out of the cloud notebook API.
I do not have the bandwith for that for a pure-amusement project
In any case, here's my implementation of a WL fiddle: 
cellToString[c : (Cell[b_BoxData, ___] | Cell[_, "Input", ___])] :=

  First@FrontEndExecute@
    ExportPacket[c, "InputText"];
cellToString[c : Cell[_, s_String, ___]] :=
  ExportString[
   <|"style" -> s,
    "content" ->
     First@FrontEndExecute@
       ExportPacket[c, "PlainText"]
    |>,
   "JSON",
   "Compact" -> True
   ];
makeWLFiddle[cells : {__Cell}] :=
  With[
   {
    cc = NotebookTools`FlattenCellGroups[cells],
    key = StringJoin[ToString /@ RandomInteger[10, 15]]
    },
   StringReplace[
And here's the fiddle for that:
One ridiculous thing about this is that all of the data is passed in the URL. So I you have to base64 encode it before passing it in. But it works! That was very cool to see. I can also have it load on request rather than load immediately.
 
Kuba
Kuba
10:16 AM
@b3m2a1 just create a WL fiddle within a JS fiddle :P
 
b3m2a1
b3m2a1
10:28 AM
@Kuba Lol
Honestly if the sandbox cloud notebooks didn't wipe themselves after a bit that'd be a great approach
 
 
8 hours later…
b3m2a1
b3m2a1
6:39 PM
For anyone who wants to contribute to the paclet server, it's even more automated. As long as you're down to always have the server point to your GitHub repo's latest release you never even have to tell the server to update it (unless you want to force an update more than the ~1 to 2x/week I'm planning on doing).
I got the update logic working so it's fast and automated for the latest release of a GitHub repo. @Szabolcs this in particular is pertinent to you I think (and maybe @halirutan for Rubi).
 
Carl Lange
Carl Lange
interesting, this is the first time I've seen this on SO mathematica.stackexchange.com/help/serial-voting-reversed
 
b3m2a1
b3m2a1
7:28 PM
@CarlLange Yeah it comes up sometimes. On MSE I don't imagine we'll have much negative serial voting, but I've definitely logged in to find like 150 rep gone and that notification.
 
Carl Lange
Carl Lange
7:47 PM
@b3m2a1 That's what's just happened to me also. It makes a lot of sense. I like it!
 
 
2 hours later…
b3m2a1
b3m2a1
9:47 PM
@Kuba wow look how well this Manipulate is performing in the cloud: wolfr.am/AFSiGUta
It's still laggy but much better than I expected
Wow I'm really liking this fiddle thing: wolfr.am/AFSOQXnY
 
Kuba
Kuba
10:13 PM
@b3m2a1 yep, really impressive. Somewhere around v1.48/49 there was a huge change in cloud notebooks. I am seriously considering using them instead of FormFunctions for few features I am working on. Forms feel more natural for that but they are more or less abandoned.
 
Craig Carter
Craig Carter
10:26 PM
Perhaps a fun idea: Today the Mueller investigation published a redacted document on the Manafort case. Looking closely at images of the redactions, (I think) I can make out forms about the size of letters--so I wondered if I could do a bit of image analysis to extract more info. I tried the usual suspects: Manipulate[TextRecognize[Binarize[image, b], "Word"], {b, 0, 1}], Manipulate[
GaussianFilter[image, r, {n1, n2}] // ImageAdjust, {r, .1, 20}, {n1,
1, 4, 1}, {n2, 1, 4, 1}]
and got nothing that seems useful. Somebody who is better at this may be more successful. (caveat, I could be fooling myself that I am seeing letter-like forms)
 
Jason B.
Jason B.
10:55 PM
@b3m2a1: "Those fun chemistry tasks depend upon what area of computational chemistry the person is moving into and their coding background. If they have never coded, then using WL to do a variety of the standard coding 101 tasks ("Hello World", List manipulation, Multi-demsional lists, File I/O, and algorithmic text extraction/scaping) are critical."
"MD folks will need to know about large data formats such as HDF5 are needed. Nearly everyone will need to interact with queueing systems, so creating simple wrappers that spit back info on current jobs, etc will be tools that go quite a long way. Electronic structure folks would be best served coding up a simple Hartree-Fock code and reading pubs.acs.org/doi/10.1021/acs.jchemed.8b00255. A good source for fun activities is pubs.acs.org/doi/abs/10.1021/acs.jchemed.7b00078";
^ from my new coworker, with a background teaching chemistry
 
b3m2a1
b3m2a1
11:11 PM
@JasonB. I'm teaching them in python :) it's what everyone else I work with uses and what will serve them best going forward I think.
 
Jason B.
Jason B.
you are dead to me now
:-P
I kid of course
 
b3m2a1
b3m2a1
But I like the idea of things like File I/O and text scraping as a warm up. They've never coded before and those will at least be something tangible they can care about.
I'm hoping I can show them that Mathematica is useful and nice, just trying not to tether them to it :)
 
Jason B.
Jason B.
if using python, are you using RDKit - or is it not useful in your subfield?
 
b3m2a1
b3m2a1
@Kuba I revamped the "Let Me Mathematica ..." site so now there's an input field! wolframcloud.com/objects/b3m2a1/LMMTFY
@JasonB. I think it'd be good to use that. I think I can find a use for it in a lot of places.
Generally we don't use external libraries (outside of Gaussian)
But I would like to move people toward doing so. They're useful.
I am in a long-term "convince people to write good code" project
And that is definitely a place to start
 
Jason B.
Jason B.
rdkit is great, and if you find things in that toolkit really useful let me know and I can propose them for WL functions
 
b3m2a1
b3m2a1
11:15 PM
Sounds good to me. Is there going to be any way to call an arbitrary RDKit function or will the LibraryLink need to be recompiled for every addition?
 
Jason B.
Jason B.
no, it needs recompiling. C++ just isn't structured enough like java is
 
b3m2a1
b3m2a1
@JasonB. would you say RDKit can do pretty much everything OB can?
 
Jason B.
Jason B.
even for their python wrappers, which are much more just straight wrappers, they need to recompile to expose things to python
 
b3m2a1
b3m2a1
Yeah I remember doing that for a python C++ extension I wrote
 
Jason B.
Jason B.
no - that's why I still wanna make a link to OB, it doesn't have nearly as much coverage for file formats
 
b3m2a1
b3m2a1
11:17 PM
You just can't release that as an official WRI link because of licensing, right?
If that existed I would 100% contribute/extend/write utilities on top of it
 
Jason B.
Jason B.
I am still planning to make the OB link a github project and farm a lot of it out to you
:-P
it is LGPL, so it isn't as easy to statically link it
gotta run
 
b3m2a1
b3m2a1
Sounds good. Let me know when it's up and going and I'll se what I can do.
 

« first day (2555 days earlier)      last day (1919 days later) »