Re: [O] [PATCH] Second proposal for ":session" doc

[Aaron Ecay]

Hi Thierry,

2014ko abenudak 7an, Thierry Pellé-ek idatzi zuen:
> 
> Hi, here is a proposal for some change on the ":session" header doc.
> 
> Comments are welcome!
> Thierry
> 
> 
> diff --git a/doc/org.texi b/doc/org.texi
> index c1e84d4..45e177e 100644
> --- a/doc/org.texi
> +++ b/doc/org.texi
> @@ -15566,7 +15566,9 @@ execution.
>  @cindex @code{:session}, src header argument
>  
>  The @code{:session} header argument starts a session for an interpreted
> -language where state is preserved.  By default, a session is not started.
> +language.  All code block of this langage is executed into the same
                            ^                ^^          ^^^^
                            s                are         by
> +interpreter process, the state of the session being preserved as long as 
> Emacs
                      ^                          ^^^^^
                      Start a new sentence here, and change to is
> +runs.  By default, a session is not started.

On a substantive note, “as long as Emacs runs” is not quite accurate –
it’s until the interpreter process exits, which could happen because the
user kills it, it crashes, a piece of code which is executed tells it to
exit, ...

>  
>  @itemize @bullet
>  @item @code{none}
> @@ -15574,9 +15576,9 @@ The default.  Each block is evaluated in its own 
> session.  The session is
>  not preserved after the evaluation.
>  @item @code{other}
>  Any other string passed to the @code{:session} header argument will give the
> -session a name.  All blocks with the same session name share the same
> -session.  Using different session name enables concurrent sessions (even for
> -the same interpreted language).  E.g., @code{:session mysession}.
> +session a name (possibly empty).  All blocks with the same session name share

I think something should be said like “If the session name is empty, a
default name based on the block’s language will be used.”  The session
name also becomes the interpreter process’s buffer name, and leaving
:session empty does not generate a buffer named “” (the empty string).

(It’s also a perverse but true fact that if you specify the same session
name for two blocks of different languages, they will both send their
code to the same interpreter process.  But for two blocks in different
languages with an empty :session, they will each use their language’s
default name.)

> +the same session.  Using different session name enables concurrent sessions
> +(for the same interpreted language).  E.g., @code{:session mysession} or 
> @code{:session}.
>  @end itemize
>  
>  @node noweb

Thanks,

-- 
Aaron Ecay
PhD candidate, Linguistics
University of Pennsylvania