4 ms·

Sho: An interactive environment for data analysis and scientific computing

21 points by xpaulbettsx 16y ago

sparky 16y ago
Can someone familiar with Matplotlib comment on how Sho's Python plotting utilities[1] compare? This looks cool; still trying to wrap my head around what's fundamentally new vs. what is packaging together existing Python libraries and letting it all talk to .Net code (not that that isn't neat and useful in itself). [1] http://research.microsoft.com/en-us/um/redmond/projects/sho/documentation/sho2.0/TheBookofSho.html#_Toc282280659 http://research.microsoft.com/en-us/um/redmond/projects/sho/...
- conjectures 16y ago
  Sho probably needs to sell its differences a bit more (if it has them). I didn't see any promises that would make it worth my while branching out from python's scipy + matplotlib + rpy packages. Maybe they're aiming for .Net users more...
- adolgert 16y ago
  IPython, with Matplotlib and Numpy, has a ton of numerical libraries, especially LAPACK, which aren't in SHO. Matplotlib is also much more extensive in its capabilities. Some coder cried when they were told to write another Matrix class. SHO has all the advantages of .NET's ecosystem, so that it can call anything .NET and be called by anything .NET. Making an environment like this should not be about using a Python REPL. You could work on the hard parts - models for asynchronous interactive calculations, or provenance tracking, or helping translate data formats from one API to another - and let the user choose a REPL.
EugeneG 16y ago
Looks like "R.Net" - which is great, since I've never been a fan of the R language syntax and the available documentation
- hessenwolf 16y ago
  My immediate thought was 'why wouldn't I just use R?' R has the advantage that almost everything you need to do statistically already has at least one package built for it. The examples in the video section; a sticky sorter prototype? What is sho? I don't get it.
  - dasmoth 16y ago
    There's a lot of useful stuff built on R, but I think it's a shame to say there isn't any scope for innovation in this area. Incanter (Clojure) is worth a look, too.
    - hessenwolf 16y ago
      I tut-tutted and pish-toshed for the first couple of slides but Incanter actually looks quite interesting. Is Clojure here to stay or a fad?
dasmoth 16y ago
Any ideas how practical it might be to get this running under Mono?

21 points by xpaulbettsx 16y ago

Can someone familiar with Matplotlib comment on how Sho's Python plotting utilities[1] compare? This looks cool; still trying to wrap my head around what's fundamentally new vs. what is packaging together existing Python libraries and letting it all talk to .Net code (not that that isn't neat and useful in itself). [1] http://research.microsoft.com/en-us/um/redmond/projects/sho/documentation/sho2.0/TheBookofSho.html#_Toc282280659 http://research.microsoft.com/en-us/um/redmond/projects/sho/...

conjectures 16y ago
Sho probably needs to sell its differences a bit more (if it has them). I didn't see any promises that would make it worth my while branching out from python's scipy + matplotlib + rpy packages. Maybe they're aiming for .Net users more...
adolgert 16y ago
IPython, with Matplotlib and Numpy, has a ton of numerical libraries, especially LAPACK, which aren't in SHO. Matplotlib is also much more extensive in its capabilities. Some coder cried when they were told to write another Matrix class. SHO has all the advantages of .NET's ecosystem, so that it can call anything .NET and be called by anything .NET. Making an environment like this should not be about using a Python REPL. You could work on the hard parts - models for asynchronous interactive calculations, or provenance tracking, or helping translate data formats from one API to another - and let the user choose a REPL.

EugeneG 16y ago

Looks like "R.Net" - which is great, since I've never been a fan of the R language syntax and the available documentation

hessenwolf 16y ago
My immediate thought was 'why wouldn't I just use R?' R has the advantage that almost everything you need to do statistically already has at least one package built for it. The examples in the video section; a sticky sorter prototype? What is sho? I don't get it.
- dasmoth 16y ago
  There's a lot of useful stuff built on R, but I think it's a shame to say there isn't any scope for innovation in this area. Incanter (Clojure) is worth a look, too.
  - hessenwolf 16y ago
    I tut-tutted and pish-toshed for the first couple of slides but Incanter actually looks quite interesting. Is Clojure here to stay or a fad?

dasmoth 16y ago

Any ideas how practical it might be to get this running under Mono?