On Mon Mar 10, 2025 at 1:41 PM GMT, Jeremy Stanley wrote:
My groff/troff reply to your earlier suggestion of this was only
somewhat tongue-in-cheek. Markdown isn't a singular clearly-specified
syntax, but a family of them with several popular flavors in
widespread use (as Timo's parser option challenges indicate).
Personally I think markdown is a bust at the moment due to the lack of
client support (and MUA behaviour with an unknown text/* type seems
remarkably poor: falling back to plain presentation would seem the smart
choice there).
But that aside, I skimmed the RFCs for adding text/markdown and it
supports a "variant" attribute, with a handful of dialects pre-defined
by a sister RFC, including CommonMark, GitHub-flavoured and Pandoc.
registration of text/markdown:
https://www.rfc-editor.org/rfc/rfc7763.html
definition of various variants, inc. CommonMark:
https://www.rfc-editor.org/rfc/rfc7764#section-3.5
That would address the ambiguity.
It seems straightforward enough to me that Markdown's code blocks should
not be reflowed whereas most of the rest of a Markdown document probably
could be. (Perhaps the specs actually spell this out; I haven't
checked).
It's a shame that Markdown's blockquote ('> ') doesn't match how people
do quoted emails in practice. This won't work (requires more empty lines
between the tiers)
Foo said:
Bar said:
Baz said:
You're wrong
No you're wrong
You're both wrong
…and when it does work, Pandoc's rendering (to 'plain') at least makes
the tiers hard to distinguish. (assuming most other renderers will be
the same)
There's also no handling of signatures.
--
Please do not CC me for listmail.
👱🏻 Jonathan Dowland
✎ j...@debian.org
🔗 https://jmtd.net
