Re: [O] python sessions

[Ista Zahn]

On Wed, Mar 20, 2013 at 2:44 AM, Andreas Röhler
<address@hidden> wrote:
> Am 20.03.2013 04:07, schrieb Nick Dokos:
>
>> John Hendy <address@hidden> wrote:
>>
>>> On Tue, Mar 19, 2013 at 4:40 PM, Gary Oberbrunner <address@hidden>
>>> wrote:
>>>>
>>>> I must be failing to understand something.  I'm running Emacs 24.3 on
>>>> Windows, with latest trunk org-mode.  I can't get python functions to
>>>> persist across blocks in session mode.  Here's my foo.org:
>>>>
>>>> ===============
>>>>
>>>> * My Document
>>>> #+BEGIN_SRC python :exports results :results output :session
>>>> def foo(x):
>>>>    return x+1
>>>> print "hi"
>>>> #+END_SRC
>>>>
>>>> #+RESULTS:
>>>> : hi
>>>>
>>>> #+BEGIN_SRC python :exports results :results output :session
>>>> print foo(100)
>>>> print "bye"
>>>> #+END_SRC
>>>>
>>>> ========================
>>>>
>>>> In session mode, shouldn't foo be defined in the second python block?
>>>> When
>>>> I export this, I get "NameError: name 'foo' is not defined"
>>>>
>>>> I may be doing something wrong, because if I name my python session, I
>>>> never
>>>> see a buffer of that name, and I expected to.  Any help?
>>>>
>>>> --
>>>> Gary
>>>
>>>
>>> You should probably post your babel configuration from .emacs. This
>>> works for me (mostly). I'm using python 3.3 and so the print function
>>> has changed to requiring parentheses. I can switch to a buffer called
>>> *Python*, however, and =print("hi")= works fine.
>>>
>>> If I change to =print(foo(100))=, I get 101 in the #+RESULTS block.
>>>
>>
>> I don't - I get the same error as Gary.
>>
>> And looking at the code of org-babel-python-initiate-session-by-key,
>> I don't understand how it's supposed to work: python-buffer is nil
>> to begin with; the cond takes the first branch and starts a python
>> session.
>> When we come to
>>
>>        (setq org-babel-python-buffers
>>             (cons (cons session python-buffer)
>>                   (assq-delete-all session org-babel-python-buffers)))
>>
>> python-buffer is still nil, so we are cooked.
>>
>> Maybe python-buffer was set as a side-effect of run-python in earlier
>> versions of emacs? If so, it does not seem to be the case now.
>>
>> Nick
>>
>> Org-mode version 8.0-pre (release_8.0-pre-144-g855dcf.dirty @
>> /home/nick/elisp/org-mode/lisp/)
>> GNU Emacs 24.3.50.2 (x86_64-unknown-linux-gnu, GTK+ Version 2.24.4) of
>> 2012-12-29 on alphaville
>>
>>
>
> AFAIU :session is broken, because Python shell as opened by run-python or
> py-shell isn't used by ob-babel.
> Seems ob-babel sends it's code w/ an own shell command, thus opening a new
> python shell internally every time.
>
> Did :session ever work? Then I might be wrong with this comment.

Not only did it work, it currently works for me. I seem to remember
bad interaction with python modes other than the one shipped with
emacs. In my .emacs I have

;; Python
;; there are several modes. Currently using
https://github.com/fgallina/python.el
;; hopefully oneday a glorious default python mode will be shipped
with emacs and we won't need this anymore.

;; (add-to-list 'load-path "~/.emacs.d/izahn")
;; (require 'python)

;; (setq
;;  python-shell-interpreter "ipython"
;;  python-shell-interpreter-args ""
;;  python-shell-prompt-regexp "In \\[[0-9]+\\]: "
;;  python-shell-prompt-output-regexp "Out\\[[0-9]+\\]: "
;;  python-shell-completion-setup-code
;;    "from IPython.core.completerlib import module_completion"
;;  python-shell-completion-module-string-code
;;    "';'.join(module_completion('''%s'''))\n"
;;  python-shell-completion-string-code
;;    "';'.join(get_ipython().Completer.all_completions('''%s'''))\n")

;; (require 'python-mode) ;;

;; python auto completion
(require 'ac-python)

All the python-mode and ipython mode stuff is commented out partly
because it did not play nice with org-mode. Using the default python
mode works for me. So, for those of you who are finding that this is
not working, what python mode are you using?

Best,
Ista

>
> Andreas
>
>
>