Re: [TASK] Allow customizeable ditaa executable in ob-ditaa.el

[Leo Butler]

On Sat, Nov 11 2023, Ihor Radchenko <yantar92@posteo.net> wrote:

> Leo Butler <Leo.Butler@umanitoba.ca> writes:
>
>>> And even if we do want to add certain features in future (like
>>> supporting ditaa executable herein), it does not mean that we have to
>>> rush them by any cost.
>>
>> Ihor, I don't understand that sentence.
>>
>> The *documentation* patch was intended to show that ob-ditaa did not
>> need to be changed. A user can already run ditaa from a script file by
>> setting the customization variables appropriately (or, inappropriately,
>> as Max said ;-) ).
>
> Let me elaborate.
> The below explanation in your patch relies upon the implementation
> detail in `org-babel-execute:ditaa' - how the ditaa command is called.
> however, it can easily happen that we change that detail in future.

Thanks for your clarifications here and in a related email. 

> In fact, your explanation is already not correct for
> :file foo.eps - org-ditaa-jar-path value is ignored in such scenario:

You say `not correct', I say `mutatus mutandis'.

>
> (cmd (concat org-babel-ditaa-java-cmd
>                     " " java " " org-ditaa-jar-option " "
>                     (shell-quote-argument
>                      (expand-file-name
>                       (if eps org-ditaa-eps-jar-path org-ditaa-jar-path)))
>                     " " cmdline
>                     " " (org-babel-process-file-name in-file)
>                     " " (if pdf-cmd
>                             eps-file
>                           (org-babel-process-file-name out-file))))
>
> Further, it won't help with the discussed problem -
> trying
> (setq org-ditaa-jar-path "flatpak-spawn --host toolbox run ditaa") will
> simply fail when passed through `shell-quote-argument'.

The patch says "Users may need to use a script to run ditaa." It does
not mention passing arbitrary command strings.

>
> And now imagine that we change how CMD is produced in future. (For
> example, there is a WIP branch that unifies escaping command arguments
> to avoid vulnerabilities). Your documentation patch may cease working
> any moment, causing damage to users who tried to follow it. Or we may
> have to constrain the ways we change the internal implementation details
> in order to not break the existing documentation. Either way is not good
> and that's why I am saying no to your proposed documentation change.

Ok, I have a clearer idea of how to proceed.

Leo