Re: Links & images with different attributes in the same paragraph

[Max Nikulin <manikulin@gmail.com>]

On 16/12/2023 21:44, Ihor Radchenko wrote:
> Max Nikulin writes:
> 
>>> @wrap{{outer wrap; @wrap{{inner wrap allowing } as well}}; back}}.
>>
>> Are you assuming invisible zero-width space as a way to escape literal
>> {{ or }}? I would prefer some visible characters.
> 
> No, not zero-width space. Literally, {{...}}. The idea is to define
> delimiters as "[{]+" the matching number of "}". This way, we do not
> need to worry about escaping "}" inside and can get nested markup for
> free. It is more or less how Org parser works for special block:
> the opening delimiter is #+begin_whatever is matched against
> #+end_<same as opening delimiter>.

I am afraid, there is an issue if wrapped content is surrounded by 
braces. An ambiguity arises for

     @wrap{{{content}}}

it may be @wrap{{{ content }}}, @wrap{{ {content} }},
or @wrap{ {{content}} }. It seems, some escape character is unavoidable.

> Also, see https://list.orgmode.org/orgmode/87mtaez8do.fsf@localhost/
> with my original proposal and some discussion that followed up.

I completely forgot that you wrote about balanced parenthesis earlier.

>> Juan Manuel Macías. Re: About 'inline special blocks'
>>> Bringing that into the paragraph is
>>> unnecessarily overloading the paragraph and breaking the social contract
>>> of lightweight markup, where paragraphs should still look like
>>> paragraphs.

> I am pretty sure that I replied to that concern raised in a parallel
> thread. My idea was to allow macro replacement inside attributes:
> 
> #+macro: alt #+attr_html :alt $1
> What about @wrap[<<<alt(Text)>>>]{[[/path/to/image]]}

Certainly it allows to shorten in-text arguments. Just as noweb 
references, it has some disadvantages. I consider comma as macro 
argument separator as a kind of a pitfall. Macro definitions, unlike 
code blocks definitions for noweb references, can not be multiline
(unless defined in elisp code).

Unless something better will be proposed, I consider macro expansion 
inside arguments as a viable approach. (I can not recall if it was 
discussed earlier.)