Add session async for ruby

[suonlight]

No description provided.

[nico202]

There are references to R

[jackkamm]

This looks great! Sorry for the delay -- I'm just seeing this now, my notifications settings must have gotten messed up.

I'll need to take a little time to look at this (I've never used Ruby) but wanted you to know it's on my radar now.

If you get a chance, it would be very helpful if you could also add a simple unit test in test/ob-session-async-test.el.

[suonlight]

@jackkamm The current tests are not passed on my local before this PR. I don't know why:

executing R code block...
ess-tracebug mode enabled
Starting evaluation...
Eval buffer
Code block evaluation complete.
Code block evaluation complete.

executing R code block...
Starting evaluation...
Eval buffer
Starting evaluation...
Eval buffer
Code block evaluation complete.
.Test ob-session-async-R-named-value backtrace:


Test ob-session-async-R-named-value condition:

    (ert-test-failed
     ((should
       (progn
         (org-babel-execute-src-block)
         (sleep-for 0 200)
         (string= ... ...)))
      :form
      (progn
        (org-babel-execute-src-block)
        (sleep-for 0 200)
        (string=
         (concat src-block results-after)
         (buffer-string)))
      :value nil))


executing R code block...
Starting evaluation...
Eval buffer
Code block evaluation complete.
Code block evaluation complete.

executing R code block...
Starting evaluation...
Eval buffer
Code block evaluation complete.
Code block evaluation complete.

executing R code block...
Starting evaluation...
Eval buffer
Starting evaluation...
Eval buffer
Code block evaluation complete.
F..Test ob-session-async-R-simple-session-async-value backtrace:


Test ob-session-async-R-simple-session-async-value condition:

    (ert-test-failed
     ((should
       (let
           (...)
         (and ... ...)))
      :form
      (let
          ((expected "Yep!"))
        (and
         (not ...)
         (string= expected ...)))
      :value nil))


executing R code block...
Starting evaluation...
Eval buffer
Starting evaluation...
Eval buffer
Code block evaluation complete.
FTest ob-session-async-R-value-drawer backtrace:


Test ob-session-async-R-value-drawer condition:

    (ert-test-failed
     ((should
       (progn
         (org-babel-execute-src-block)
         (sleep-for 0 200)
         (string= ... ...)))
      :form
      (progn
        (org-babel-execute-src-block)
        (sleep-for 0 200)
        (string=
         (concat src-block result)
         (buffer-string)))
      :value nil))

F

Ran 6 tests in 2.908 seconds
3 unexpected results:
   FAILED  ob-session-async-R-named-value
   FAILED  ob-session-async-R-simple-session-async-value
   FAILED  ob-session-async-R-value-drawer

[jackkamm]

Interesting -- it looks like the tests with :results value are the ones failing, whereas the tests with :results output are fine.

Could you try adding (print (org-babel-read-result)) to the test ob-session-async-R-simple-session-async-value to see what might be going on?

Also -- what OS and emacs version are you using?

Thanks in advance, and sorry for the tests failing -- I think you're the first person besides myself to run them, so this is useful info for me.

[suonlight]

I'm using:

OS: Mac OSX Mojave 10.14.5
Emacs: 27.0.50
Org-Mode: 9.2.5

The code always returns temp file even when I increases sleep-for
(print (org-babel-read-result));; /var/folders/_7/jvqjkfkj1p77dwlg23tl7d200000gn/T/R-kZawq4

[jackkamm]

Thanks for the info, I'm running on emacs 26 on Linux which explains the difference. I'll try to reproduce this on emacs 27/an old macbook I have lying around.

[suonlight]

If you use emacs 27, then you should add

(when (> emacs-major-version 26)
  (defalias 'ert--print-backtrace 'backtrace-to-string))

to test-helper. See rejeep/ert-runner.el#49 for more details.

[suonlight]

@jackkamm I have added units for Ruby and fix failed unit tests for R. The problem is: in each test case, we need to add (org-babel-temporary-directory "/tmp") to generate temp file like /tmp/ruby-fRIHNK. Otherwise, it will generate the temp file like ob_comint_async_ruby_file_/var/folders/_7/jvqjkfkj1p77dwlg23tl7d200000gn/T/ruby-fRIHNK which failed the indicator because of _7.

Now, all tests for ruby and R work well.

Pls help me review again.

[jackkamm]

This is great, thanks. The code looks good, and I can confirm all the unit tests pass on my system.

Also, I very much appreciate you fixing the R unit tests for emacs27/macOS -- thanks!

As soon as you fix my 1 minor comment above (about authorship) I will merge this in.

I should also note, this PR achieves a major milestone for this project, which is to add support for a second language!

I am thinking I would like to add 1 more language, just to make sure the core architecture in ob-session-async.el is general enough. Then I will start considering whether to submit this functionality upstream into org-mode. If you any thoughts/comments on this I would be interested to hear your opinions.

[suonlight]

@jackkamm Contributing to org-mode is great. I just have a concern about ob-async. This package and ob-async should merge into one before submitting to org-mode. But, we should add 1 more language first to see how the core architecture is good enough. In the meantime, we need a post on Reddit about this package again :) to get more users and insight

[jackkamm]

@suonlight

Session async has been added upstream to Org since 9.5. In particular, the functionality in ob-session-async.el was moved into ob-comint.el, though slightly renamed (e.g., ob-session-async-register is now org-babel-comint-async-register). Implementations for Python and R were also added to Org mode.

Therefore, I am going to archive this repo.

I encourage you to submit your async Ruby implementation into Org mode, or to make a standalone package for it.

Thanks again for implementing async Ruby session evaluation!