Re: [Tools-discuss] Let's have better tools

[Carsten Bormann <[e-mail]>]

On 28. Jun 2021, at 02:51, John Levine <[e-mail]> wrote:
> 
>> https://codimd.ietf.org/mcr-kramdown-test
>> 
>> This is 90% of the way there.
> 
> That last 10% is a killer.

+1.

Note that CodiMD (which these days is called HedgeDoc, we are just running an older version) is already quite useful for editing rfc-markdown.
As is the markdown support in GitHub.

But to get those last 10 % of fidelity (“WYSIWYG”), we need xml2rfc in the loop.
For now, this means server-based approaches, as I alluded to.

As soon as references etc. are in cache, my laptop translates the source of RFC 8949 (approximately 80 pages (*)) to HTML in about 2.6 s (0.8 s kramdown-rfc + 1.8 s xml2rfc).
Adding a few 100 ms for browser processing, this means that the latency from an editor save to a redisplay could be ≈ 3 s.
Maybe not good enough for editing NFSv4.1, but for everything else, this should be quite usable.
(And I have a few ideas how to speed this up by a factor of three, but crawling comes before walking.)

Grüße, Carsten

(*) Yeah, I know.  We don’t get page numbers any more, except in the PDF, and that is slightly more dense, at 66 pages here.  Hence “approximately”.  But most people will have a feel for the size of their I-Ds in plaintext pages, so I’m using the 80 pages the I-D was.