emacs-orgmode
[Top][All Lists]
Advanced

[Date Prev][Date Next][Thread Prev][Thread Next][Date Index][Thread Index]

Re: [Orgmode] [Announcement] Org-babel initial release


From: Eric Schulte
Subject: Re: [Orgmode] [Announcement] Org-babel initial release
Date: Tue, 15 Sep 2009 11:54:36 -0600
User-agent: Gnus/5.13 (Gnus v5.13) Emacs/23.1.50 (darwin)

Hi Sebastian,

Sebastian Rose <address@hidden> writes:

> "Eric Schulte" <address@hidden> writes:
>> Dan Davison and I (Eric Schulte) are happy to announce that Org-babel
>> has now been released as a contributed package in Org-mode with
>> corresponding documentation on worg [1].
>
>
> What else should I say - THIS IS GREAT NEWS!!
>

Thanks, I hope fellow Orgers find it useful

>
> I wonder how complicated it would be to add more languages. Especially
> PHP, JavaScript (e.g. per rhino) and Perl.
>

Yes, I must admit the current set of implemented language is very
specific to Dan and my personal needs.  I really hope that other people
add the languages they need/use, and I tried to design the structure of
org-babel to make the addition of new languages as painless as possible.

>
> Is there some documentation or advice on the net in that concern? I'd
> like to add a language I understand besides Bash... :)
>
> Hmmm - maybe `org-babel-sh.el' is a good starting point.
>

Yes, currently the best way to get a feel for how to add languages would
be to start with an existing language file (I'd suggest
org-babel-python.el or org-babel-ruby.el, or for simpler less
comprehensive language support look at org-babel-ditaa or
org-babel-haskell) and make changes from there.  I agree that a brief
tutorial for adding language support would be helpful.

Basically Org-babel expects any new language file to define two
functions...

- org-babel-execute:lang-name (body params) :: which executes the code
     in body according to the header arguments in params, and

- org-babel-prep-session:ruby (session params) :: which starts an
     interactive session in session setting any variables from params

>
> * Some thoughts
>
>
> I actually wonder, if all those interpreted languages are different at
> all. Why not add an generic call to interpreters. Executing Shell
> scripts or Perl, Php, JavaScript... makes no big difference here. On
> Linux at least, they all work with either shebang or called with OPTION
> FILENAME.
>

There are two key language specific features which keep us from treating
all interpreted languages identically.
1) Org-babel collects the last value of a source-code block to be
   returned (see [1]) and this value needs to be collected and
   potentially converted into elisp in a language specific manner
2) Org-babel has support for evaluation in a session allowing
   persistence of state between different blocks which use the same
   session.  I now notice that the :session header argument is not
   currently documented on the Worg page.  I'll try to add this
   documentation soon.  The sessions are handled through Emacs comint
   buffers which are very language specific.

>
> To execute a temporary file without shebang, all interpreters take a
> filename:
>
> rhino -f FILE  # -f is optional
> php   -f FILE  # -f is optional
> perl     FILE
>
> Or execute code directly (which is useless for us, since we would need
> to quote the code correctly...):
>
> rhino -e  CODE...
> perl  -e  CODE...
> php   -r  CODE...
>

I agree that it would be possible to implement a much more general
code-evaluation mechanism based on execution of temporary files, but we
would lose the nice features mentioned above.

>
>
> So how about:
>
>   #+srcname: generic-circumference(a)
>   #+begin_src javascript :interpreter rhino -f
>
>   print ( "Write me to temp file and call `rhino -f TMPFILE'" )
>   java.lang.System.out.println ( 2 * a * java.lang.Math.PI )
>
>   #+end_src
>
> This way, a source block written on my Linux-System would execute on her
> MAC and his Windows machine without change (provided the interpreter is
> installed and in $PATH... `org-program-exists' ... to use interpreters
> without having them in $PATH, a customizable map could be used).
>

Under the current setup, all source-code blocks should be executable on
*any* system which supports the required language and Emacs modes
mentioned in the commented elisp block at [2]

Best -- Eric

>
>
>
>
>
>
> BTW: I just discovered rhino - looks interesting, seems you can
> (de-)serialize Java(Script) objects... and thus keep track of things
> between sessions ;) start programs, call Java Methods...  See
> https://developer.mozilla.org/en/Rhino_Shell
>
>
>
> Best wishes,
>
>   Sebastian

Footnotes: 
[1]  http://orgmode.org/worg/org-contrib/babel/org-babel.php#results

[2]  http://orgmode.org/worg/org-contrib/babel/org-babel.php#languages





reply via email to

[Prev in Thread] Current Thread [Next in Thread]