Avoid "Scheme implementation" prompt on export

Avoid "Scheme implementation" prompt on export

[Roi Martin]

Hello,

I have an org document with the following structure:

---
Racket code block:

#+begin_src scheme
  #lang racket
  (+ 1 2)
#+end_src

Chez Scheme code block:

#+begin_src scheme
  (+ 3 4)
#+end_src
---

It includes several code blocks with different scheme implementations
(e.g. Racket and Chez).

When I export the document to HTML, both from the Org Export Dispatcher
(C-c C-e) or calling (org-publish-all), I'm asked to choose a "Scheme
implementation" for every code block.

If I understand correctly, I can set a default Scheme implementation
with the `geiser-scheme-implementation' variable.  However, I would like
to specify an implementation per block without having to answer the
prompt every time.

Is it possible?

Just in case, it is useful, I'm using the following package versions:

- GNU Emacs 29.4
- org 9.6.15
- geiser 20240907.2235
- geiser-chez 20230707.1334
- geiser-guile 20240920.35
- geiser-racket 20210421.125
- htmlize 20240915.1657

Thanks for your help!

Best,
        Roi

Re: Avoid "Scheme implementation" prompt on export

[Bruno Barbier]

Hi Roi,

Roi Martin <jroi.martin@gmail.com> writes:

> If I understand correctly, I can set a default Scheme implementation
> with the `geiser-scheme-implementation' variable.  However, I would like
> to specify an implementation per block without having to answer the
> prompt every time.
>
> Is it possible?

According to the current source code, you should be able to pick the
scheme implementation using the :scheme argument (either per block, per
heading, or per buffer). It should take the same value as the variable
'geiser-scheme-implementation'.


I don't have any scheme installed: I couldn't test it.  Note that if
you're using sessions, that value will probably be ignored for already
opened sessions.

HTH,

Bruno

Re: Avoid "Scheme implementation" prompt on export

[Roi Martin]

Sorry, I forgot to reply to the list.

Roi Martin <jroi.martin@gmail.com> writes:

> Hi Bruno,
>
> Bruno Barbier <brubar.cs@gmail.com> writes:
>
>> According to the current source code, you should be able to pick the
>> scheme implementation using the :scheme argument (either per block, per
>> heading, or per buffer). It should take the same value as the variable
>> 'geiser-scheme-implementation'.
>
> I tried with:
>
> #+begin_src scheme :scheme chez
>   (+ 3 4)
> #+end_src
>
> and also with ':scheme racket'.
>
> Unfortunately, it does not seem to work when exporting the document.
>
> Thanks for your help!
>
>         Roi

Re: Avoid "Scheme implementation" prompt on export

[Bruno Barbier]

[-- Attachment #1: Type: text/plain, Size: 767 bytes --]

Roi Martin <jroi.martin@gmail.com> writes:

>> I tried with:
>>
>> #+begin_src scheme :scheme chez
>>   (+ 3 4)
>> #+end_src
>>
>> and also with ':scheme racket'.
>>
>> Unfortunately, it does not seem to work when exporting the document.
>>
>> Thanks for your help!
>>
>>         Roi

I've installed chez and racket.  Indeed, the argument :session
if half-ignored.

I've attached a quick ugly hack; with that change, it is now working
for me.

    #+begin_src scheme :scheme racket
      (banner)
    #+end_src

    #+RESULTS:
    : Welcome to Racket v8.14 [cs].\n

    (setq-default geiser-chez-binary "chezscheme")
    #+begin_src scheme :scheme chez 
      (scheme-version)
    #+end_src

    #+RESULTS:
    : Chez Scheme Version 10.0.0

Could you try it?

Bruno


[-- Warning: decoded text below may be mangled, UTF-8 assumed --]
[-- Attachment #2: ob-scheme: Hack to pass the implementation to 'geiser-mode' --]
[-- Type: text/x-patch, Size: 918 bytes --]

From dbb3068bd3e22ee8d38c99c69061b26d3a4064d6 Mon Sep 17 00:00:00 2001
From: Bruno BARBIER <brubar.cs@gmail.com>
Date: Tue, 15 Oct 2024 12:42:44 +0200
Subject: [PATCH] ob-scheme: Hack to pass the implementation to 'geiser-mode'

---
 lisp/ob-scheme.el | 4 +++-
 1 file changed, 3 insertions(+), 1 deletion(-)

diff --git a/lisp/ob-scheme.el b/lisp/ob-scheme.el
index 29fdcd1ec..7bf8acf6c 100644
--- a/lisp/ob-scheme.el
+++ b/lisp/ob-scheme.el
@@ -184,7 +184,9 @@ (defun org-babel-scheme-execute-with-geiser (code output impl repl &optional hos
       (newline)
       (let ((beg (point)))
         (insert code)
-        (geiser-mode)
+        ;; Hack to pass our 'impl' to 'geiser-mode'.
+        (let ((geiser-impl--implementation impl))
+          (geiser-mode))
         (let ((geiser-repl-window-allow-split nil)
 	      (geiser-repl-use-other-window nil))
 	  (let ((repl-buffer (save-current-buffer
-- 
2.45.2

Re: Avoid "Scheme implementation" prompt on export

[Roi Martin]

Bruno Barbier <brubar.cs@gmail.com> writes:

> I've installed chez and racket.  Indeed, the argument :session
> if half-ignored.
>
> I've attached a quick ugly hack; with that change, it is now working
> for me.
>
>     #+begin_src scheme :scheme racket
>       (banner)
>     #+end_src
>
>     #+RESULTS:
>     : Welcome to Racket v8.14 [cs].\n
>
>     (setq-default geiser-chez-binary "chezscheme")
>     #+begin_src scheme :scheme chez 
>       (scheme-version)
>     #+end_src
>
>     #+RESULTS:
>     : Chez Scheme Version 10.0.0
>
> Could you try it?

I tried the current main branch (commit da0f6eff75cd ("ox-texinfo: Fix
edge case with - Variable: nil:: definition") with and without the
patch.  Unless I'm doing something wrong, exporting the document to HTML
with "C-c C-e h H" asks for the Scheme implementation for every code
block.  Please, note that the htmlize package must be installed.

Also, note that executing the code block with "C-c C-c" never asked for
the Scheme implementation when the ":scheme" argument is specified.

Cheers,
Roi

Re: Avoid "Scheme implementation" prompt on export

[Bruno Barbier]

[-- Attachment #1: Type: text/plain, Size: 1586 bytes --]



Roi Martin <jroi.martin@gmail.com> writes:    
> I tried the current main branch (commit da0f6eff75cd ("ox-texinfo: Fix
> edge case with - Variable: nil:: definition") with and without the
> patch.  Unless I'm doing something wrong, exporting the document to HTML
> with "C-c C-e h H" asks for the Scheme implementation for every code
> block.  Please, note that the htmlize package must be installed.
>
> Also, note that executing the code block with "C-c C-c" never asked for
> the Scheme implementation when the ":scheme" argument is specified.

My patch fixes a real bug though.  Are you sure org is using the correct
scheme implementation when doing "C-c C-c" ?


I just checked exporting my example to html: it works for me.

I've tested using the same version as you:

     Org mode commit: da0f6eff75cd
     
with my patch applied on top, and using a clean bare bone emacs,
without any config.

I'm launching emacs like this:

  emacs -q test_ob-scheme-bug.org


Running the elisp code block, I'm getting the expected result in HTML.

In the HTML file, I see:

     #+begin_example
     <pre class="example">
     Welcome to Racket v8.14 [cs].\n
     </pre>
     #+end_example
and:
     #+begin_example
     <pre class="example">
     Chez Scheme Version 10.0.0
     </pre>
     #+end_example

I've attached the org file that I've used to test this.  The first elisp
block allows to configure Emacs from scratch (downloading and installing
the required packages), export the org file to HTML and open the HTML
result.

Maybe something is wrong in your configuration ?

Bruno


[-- Warning: decoded text below may be mangled, UTF-8 assumed --]
[-- Attachment #2: test_ob-scheme-bug.org --]
[-- Type: text/org, Size: 849 bytes --]

* Mandatory first heading
   :PROPERTIES:
   :header-args:scheme: :exports both :results value
   :END:

 #+begin_src elisp
   (setq-local org-confirm-babel-evaluate nil)
   (setq package-archives
      '(("gnu"          . "http://elpa.gnu.org/packages/")
        ("melpa-stable" . "http://stable.melpa.org/packages/")))
   (setq package-check-signature t) ; 'allow-unsigned)
   (setq package-unsigned-archives (list "melpa-stable"))
   (package-initialize)

   (load-library "ob-scheme")
   (package-install 'geiser)
   (package-install 'geiser-chez)
   (package-install 'geiser-racket)
   (setq-default geiser-chez-binary "chezscheme")
   (org-export-to-file 'html "test.html")
   (find-file "test.html")
 #+end_src

 
 #+begin_src scheme :scheme racket 
   (banner)
 #+end_src

 
 #+begin_src scheme :scheme chez
   (scheme-version)
 #+end_src

Re: Avoid "Scheme implementation" prompt on export

[Roi Martin]

[-- Attachment #1: Type: text/plain, Size: 1656 bytes --]

Bruno Barbier <brubar.cs@gmail.com> writes:

> My patch fixes a real bug though.  Are you sure org is using the correct
> scheme implementation when doing "C-c C-c" ?

I think it works fine if you set the `org-babel-load-languages' variable
in the elisp code block:

        (org-babel-do-load-languages 'org-babel-load-languages '((emacs-lisp . t)
                                                                 (scheme . t)))

> I just checked exporting my example to html: it works for me.

Have you tried with the htmlize package installed?

  (package-install 'htmlize)

> I'm launching emacs like this:
>
>   emacs -q test_ob-scheme-bug.org
>
> Maybe something is wrong in your configuration ?

I did my last tests launching emacs like this:

        emacs -Q --init-directory=initdir -L ~/src/org-mode/lisp test_ob-scheme-bug.org

I also simplified the elisp code block a bit:

        #+begin_src elisp
          (setq package-archives
                '(("gnu" . "https://elpa.gnu.org/packages/")
                  ("nongnu" . "https://elpa.nongnu.org/nongnu/")))
          (setq package-check-signature t)
          (package-initialize)
          (package-install 'geiser)
          (package-install 'geiser-chez)
          (package-install 'geiser-racket)
          (package-install 'htmlize)
        
          (setq-local org-confirm-babel-evaluate nil)
          (org-babel-do-load-languages 'org-babel-load-languages '((emacs-lisp . t)
                                                                   (scheme . t)))
          (org-export-to-file 'html "test.html")
        #+end_src

I attached my test file.  I hope it helps.

Best,
Roi


[-- Attachment #2: test_ob-scheme-bug.org --]
[-- Type: text/plain, Size: 744 bytes --]

* Mandatory first heading
  :PROPERTIES:
  :header-args:scheme: :exports both :results value
  :END:

#+begin_src elisp
  (setq package-archives
	'(("gnu" . "https://elpa.gnu.org/packages/")
	  ("nongnu" . "https://elpa.nongnu.org/nongnu/")))
  (setq package-check-signature t)
  (package-initialize)
  (package-install 'geiser)
  (package-install 'geiser-chez)
  (package-install 'geiser-racket)
  (package-install 'htmlize)

  (setq-local org-confirm-babel-evaluate nil)
  (org-babel-do-load-languages 'org-babel-load-languages '((emacs-lisp . t)
							   (scheme . t)))
  (org-export-to-file 'html "test.html")
#+end_src

#+begin_src scheme :scheme racket 
  (banner)
#+end_src

#+begin_src scheme :scheme chez
  (scheme-version)
#+end_src

Re: Avoid "Scheme implementation" prompt on export

[Bruno Barbier]

Roi Martin <jroi.martin@gmail.com> writes:

> Bruno Barbier <brubar.cs@gmail.com> writes:
>
>> My patch fixes a real bug though.  Are you sure org is using the correct
>> scheme implementation when doing "C-c C-c" ?

> I think it works fine if you set the `org-babel-load-languages' variable
> in the elisp code block:
>
>         (org-babel-do-load-languages 'org-babel-load-languages '((emacs-lisp . t)
>                                                                  (scheme . t)))
>

ok. I simplified:

    (load-library "ob-scheme")

into:

    (org-babel-do-load-languages 'org-babel-load-languages '((emacs-lisp . t)
                                                             (scheme . t))
:)



>> I just checked exporting my example to html: it works for me.
>
> Have you tried with the htmlize package installed?
>
>   (package-install 'htmlize)
>

Nope.  Everything was included in my minimal reproducible example and
htmlize was not there.  And I didn't use the GNU nongnu repo either.

Using htmlize *AND* the nongnu version, I'm indeed able to reproduce.

The HTML fontification is launching REPLs, and *may* prompts
for the implementation to use.  I'm able to work around this with this
setting:

  (setq geiser-mode-auto-p nil)

I'm not sure what the real solution should be though.

And thanks for your test file! (probably impossible to spot the melpa vs
nongnu difference without it).

Bruno

Re: Avoid "Scheme implementation" prompt on export

[Roi Martin]

Bruno Barbier <brubar.cs@gmail.com> writes:

> The HTML fontification is launching REPLs, and *may* prompts
> for the implementation to use.  I'm able to work around this with this
> setting:
>
>   (setq geiser-mode-auto-p nil)
>
> I'm not sure what the real solution should be though.

Thanks a lot for the workaround!  It works like a charm.

After pinpointing the issue to having htmlize installed, I've been
digging a bit and it seems that the `org-html-fontify-code' function in
`ox-html` triggers the prompt.  It creates a temporary buffer with the
code and enables the corresponding language mode:

 ...
  (with-temp-buffer
    ;; Switch to language-specific mode.
    (funcall lang-mode)
    (insert code)
 ...

So, it makes sense to set `geiser-mode-auto-p' to nil in order to avoid
being prompted for the specific Scheme implementation.

Maybe, we could have an alist that associates a language to some kind of
"setup" function.  For instance, in the case of Scheme, it would read
the `:scheme' argument and set the geiser implementation.  However, I'm
not sure if it is worth it.  Also, I'm not familiar with the code base
of org-mode, so bear with me. Probably, what I'm proposing does not make
sense or is not the optimal solution.

What do you think?

In any case, thanks a lot for your help!

Best,
Roi

Re: Avoid "Scheme implementation" prompt on export

[Ihor Radchenko]

Roi Martin <jroi.martin@gmail.com> writes:

> I have an org document with the following structure:
>
> ---
> Racket code block:
>
> #+begin_src scheme
>   #lang racket
>   (+ 1 2)
> #+end_src
> ...
> It includes several code blocks with different scheme implementations
> (e.g. Racket and Chez).
>
> When I export the document to HTML, both from the Org Export Dispatcher
> (C-c C-e) or calling (org-publish-all), I'm asked to choose a "Scheme
> implementation" for every code block.
>
> If I understand correctly, I can set a default Scheme implementation
> with the `geiser-scheme-implementation' variable.  However, I would like
> to specify an implementation per block without having to answer the
> prompt every time.
>
> Is it possible?

Org mode fontification routines do not support multiple different
variants of major mode in src blocks.
We also cannot handle major modes prompting user when they are
activated.

Actually, I have no idea how to handle such scenarios.

-- 
Ihor Radchenko // yantar92,
Org mode contributor,
Learn more about Org mode at <https://orgmode.org/>.
Support Org development at <https://liberapay.com/org-mode>,
or support my work at <https://liberapay.com/yantar92>

Re: Avoid "Scheme implementation" prompt on export

[Ihor Radchenko]

Bruno Barbier <brubar.cs@gmail.com> writes:

>>> #+begin_src scheme :scheme chez
>>>   (+ 3 4)
>>> #+end_src
>>>
>>> and also with ':scheme racket'.
>>>
>>> Unfortunately, it does not seem to work when exporting the document.

:scheme header argument does not affect export by default.
It only affects code evaluation and may only affect export when you also
have :eval yes and :exports both/results.

> @@ -184,7 +184,9 @@ (defun org-babel-scheme-execute-with-geiser (code output impl repl &optional hos
>        (newline)
>        (let ((beg (point)))
>          (insert code)
> -        (geiser-mode)
> +        ;; Hack to pass our 'impl' to 'geiser-mode'.
> +        (let ((geiser-impl--implementation impl))
> +          (geiser-mode))

Just a line before Org tells geiser to use IMPL value via
(insert (format ";; -*- geiser-scheme-implementation: %s -*-" impl))

I am not sure what you are trying to fix here.

-- 
Ihor Radchenko // yantar92,
Org mode contributor,
Learn more about Org mode at <https://orgmode.org/>.
Support Org development at <https://liberapay.com/org-mode>,
or support my work at <https://liberapay.com/yantar92>

Re: Avoid "Scheme implementation" prompt on export

[Ihor Radchenko]

Roi Martin <jroi.martin@gmail.com> writes:

> Maybe, we could have an alist that associates a language to some kind of
> "setup" function.  For instance, in the case of Scheme, it would read
> the `:scheme' argument and set the geiser implementation.  However, I'm
> not sure if it is worth it.  Also, I'm not familiar with the code base
> of org-mode, so bear with me. Probably, what I'm proposing does not make
> sense or is not the optimal solution.
>
> What do you think?

The problem is deeper.
Org mode does not even always call a given major mode for each src
block. For the purposes of fontification, a single major mode buffer may
be created and then reused to fontify src blocks with the same language.

If you want some kind of "setup" function, it is what scheme-mode-hook
(or <major-mode>-hook) is for.

-- 
Ihor Radchenko // yantar92,
Org mode contributor,
Learn more about Org mode at <https://orgmode.org/>.
Support Org development at <https://liberapay.com/org-mode>,
or support my work at <https://liberapay.com/yantar92>

Re: Avoid "Scheme implementation" prompt on export

[Bruno Barbier]

Ihor Radchenko <yantar92@posteo.net> writes:

> Bruno Barbier <brubar.cs@gmail.com> writes:
>
>>>> #+begin_src scheme :scheme chez
>>>>   (+ 3 4)
>>>> #+end_src
>>>>
>>>> and also with ':scheme racket'.
>>>>
>>>> Unfortunately, it does not seem to work when exporting the document.
>
> :scheme header argument does not affect export by default.
> It only affects code evaluation and may only affect export when you also
> have :eval yes and :exports both/results.

We all agree here I guess.


>> @@ -184,7 +184,9 @@ (defun org-babel-scheme-execute-with-geiser (code output impl repl &optional hos
>>        (newline)
>>        (let ((beg (point)))
>>          (insert code)
>> -        (geiser-mode)
>> +        ;; Hack to pass our 'impl' to 'geiser-mode'.
>> +        (let ((geiser-impl--implementation impl))
>> +          (geiser-mode))
>
> Just a line before Org tells geiser to use IMPL value via
> (insert (format ";; -*- geiser-scheme-implementation: %s -*-" impl))
>
> I am not sure what you are trying to fix here.

I'm not sure either.  Maybe, during my tests, the function
'geiser-impl--guess' somehow ignored the local variable
'geiser-scheme-implementation'.

Anyway, with a fresh org, the nongnu geiser and an empty config, that
hack seems useless indeed.

Good catch! Thanks!

>
> -- 
> Ihor Radchenko // yantar92,
> Org mode contributor,
> Learn more about Org mode at <https://orgmode.org/>.
> Support Org development at <https://liberapay.com/org-mode>,
> or support my work at <https://liberapay.com/yantar92>

Re: Avoid "Scheme implementation" prompt on export

[Bruno Barbier]

Hi Ihor,

Ihor Radchenko <yantar92@posteo.net> writes:

> Org mode fontification routines do not support multiple different
> variants of major mode in src blocks.
> We also cannot handle major modes prompting user when they are
> activated.
>

In case you missed it, the workaround that solved
the OP problem is:
  (setq geiser-mode-auto-p nil)

Without this, htmlfontify turn on scheme-mode, and geiser then asks for
which implementation to use.


Bruno


> -- 
> Ihor Radchenko // yantar92,
> Org mode contributor,
> Learn more about Org mode at <https://orgmode.org/>.
> Support Org development at <https://liberapay.com/org-mode>,
> or support my work at <https://liberapay.com/yantar92>

Re: Avoid "Scheme implementation" prompt on export

[Ihor Radchenko]

Bruno Barbier <brubar.cs@gmail.com> writes:

> Ihor Radchenko <yantar92@posteo.net> writes:
>
>> Org mode fontification routines do not support multiple different
>> variants of major mode in src blocks.
>> We also cannot handle major modes prompting user when they are
>> activated.
>>
>
> In case you missed it, the workaround that solved
> the OP problem is:
>   (setq geiser-mode-auto-p nil)
>
> Without this, htmlfontify turn on scheme-mode, and geiser then asks for
> which implementation to use.

I saw it.
I was commenting on the possibility for Org mode to solve this type of
problems in some general way.

-- 
Ihor Radchenko // yantar92,
Org mode contributor,
Learn more about Org mode at <https://orgmode.org/>.
Support Org development at <https://liberapay.com/org-mode>,
or support my work at <https://liberapay.com/yantar92>