[Tools-discuss] Why do we even have text formats any more?

[Martin Thomson <[e-mail]>]

I realize that this might be a little inflammatory as far as subjects go, but bear with me.

There are probably a few narrow cases where rendering plain text is better than HTML.  But what we've been doing for years (thanks to Henrik's great tool) is take text and turn it into HTML using the power of regular expressions.  That's been good, but it's not always reliable (how many errata mention that "Section X of [FOO]" links to Section X of this document?).  It's also been lagging as the text format changed (case in point: lack of a table of contents).

Here's an alternative: style the HTML so that it looks like the text.  I tried this and it worked shockingly well.

Repo: https://github.com/martinthomson/rfc-txt-html
Demo: https://martinthomson.github.io/rfc-txt-html/diff.html

This isn't perfect, but it seems pretty good to me.  Keep in mind that this took only a little bit of time to sketch out. No doubt it can be improved.  The readme has a bunch of things I found, all minor.

I don't think that this is the end of text, but a possible way to limit our use of the htmlizer[1].  People who need to automate access to content might still use text, though I will argue that XML is superior in that regard.   The other thing that comes to mind is diffs: HTML-native diff tools are somewhat less than ideal.  Either way, serving HTML is just better.

Enjoy,
Martin


[1] Though I still see a shocking number of people authoring in XML (or XML-capable input formats) and submitting in text.  But I think we have plans to limit that.