Re: [POLL] Should we accept breaking changes to get rid of Org libraries that perform side effects when loading?

[Bastien Guerry]

Ihor Radchenko <yantar92@posteo.net> writes:

> Yes, org-mouse modifies the behavior of certain key bindings. Not
> directly, but by advising `org-open-at-point'.

IIRC Emacs core libraries should not advise functions.
This is something we should fix.

Also, I'm not sure org-mouse.el has its place in Org's core nowadays.

> It changes the very notion of that is a headline - the syntax definition
> is altered. Very deeply nested headlines may become inlinetasks upon
> loading org-inlinetask, touching all aspects of Org, not just editing.

Same here, I'd be tempted to deny Org citizenship to inline tasks: it
always felt like a nice hack for a niche use-case, but a hack anyway.

If it modifies Org syntax in surprising ways, this is another argument
for removing org-inlinetask.el from Org's core.  Remember: this is not
to say that inline tasks are forbidden, it's just a message for users
that inline tasks are something not maintained by Org's core team.

> And it is not clear how to fix this. We did not make inlinetasks into
> standard Org syntax in the past and now it is in the weird state when we
> have (featurep 'org-inlinetask) sprinkled across the code just to
> accommodate for this conditional syntax.

Yes, this is ugly.

> Inlinetasks are too similar in syntax with headlines, so it is
> impossible to make the change backwards-compatible.
>
>>> With the current state of affairs, it is often enough to
>>> (require 'org-library) to get things work. If we get rid of all the
>>> possible side effects, users will have to adapt their configurations
>>> and we will thus violate "I won't force you to update your
>>> configuration."
>>
>> Defining new functions is a desirable "side-effect" of all Elisp
>> library, I don't think we should worry abou this.
>
> Defining new functions by itself is not a big deal. But there are parts
> of Org that alter their behavior depending on whether a feature is
> loaded (like org-inlinetask) or depending whether certain function
> symbol is defined (babel). Similarly, loading new link types re-defines
> Org syntax in all the documents, affecting editing of everything that
> looks like the loaded link type (org-ctags).

I feel like the stakes are not the same for features like org-mouse.el
and org-inlinetask.el and for core features like Babel libs and links.
For the former, a decision should be made relatively to the usefulness
of the feature; for the latter, loading libs (with side-effects on the
syntax) is required by the design of the core feature at hand (Babel
and links).

I'd focus on solving the problem with org-mouse and org-inlinetasks
first. Let's make a poll for org-mouse.el then for org-inlinetasks.el ?

-- 
 Bastien Guerry