IDL support

[halcyon-gh]

Add support for IDL notebooks
https://www.harrisgeospatial.com/docs/IDL_Kernel.html
https://github.com/gnudatalanguage/idl_kernel

[codecov]

Codecov Report

Merging #232 into master will increase coverage by <.01%.
The diff coverage is 100%.

@@            Coverage Diff             @@
##           master     #232      +/-   ##
==========================================
+ Coverage   99.23%   99.23%   +<.01%     
==========================================
  Files          65       65              
  Lines        6172     6176       +4     
==========================================
+ Hits         6125     6129       +4     
  Misses         47       47

Impacted Files	Coverage Δ
jupytext/languages.py	100% <ø> (ø)	⬆️
tests/test_mirror.py	100% <100%> (ø)	⬆️

---

Continue to review full report at Codecov.

Legend - Click here to learn more
Δ = absolute <relative> (impact), ø = not affected, ? = missing data
Powered by Codecov. Last update 6a870b2...cbae18d. Read the comment docs.

[mwouts]

Hello @halcyon-gh , thanks for your contribution. I will integrate this in the next release soon.

[mwouts]

This has been integrated into Jupytext 1.1.3. Thanks @halcyon-gh !

[mwouts]

Hello @halcyon-gh , I have a question for you... Yesterday I extended the round trip test for Jupyter notebooks as Markdown files to all languages, including IDL. But that did not work well as of 1.1.6, because the IDL language is sometimes encoded as IDL (in the kernel) or as idl (in our extension mapping). So we have to choose one... This is not a big issue, but I'd like to know your preference: should a code chunk, in Markdown, start with

```idl

or

```IDL

For now I have chosen the first one, see this file (go to the raw representation). Do you think it is OK?

[halcyon-gh]

Hi, not really sure what is most appropriate. In the metadata of the official Exelis kernel you get this:

{
  "kernelspec": {
    "name": "idl",
    "display_name": "IDL",
    "language": "IDL"
  },
  "language_info": {
    "name": "idl",
    "codemirror_mode": "idl",
    "mimetype": "text/x-idl",
    "file_extension": ".pro"
  }
}

You would think kernelspec,language and language_info,name would be the same. Is one of those specifically queried by jupytext? Otherwise going with the lowercase version is as valid a choice as any.

[mwouts]

I see. Well, Jupytext takes the language from the kernelspec dictionary, so that's why until version 1.1.6 the Markdown representation had ```IDL. But I do think lower case is more appropriate. Since you also agree (thanks!) I'll go for ```idl in 1.1.7. I will also add an additional test to make sure that Markdown documents with the uppercase version can still be read.