Re: [Tools-discuss] .txt? [I-D Action: draft-xxx.txt]

[Carsten Bormann <[e-mail]>]

On 2021-06-30, at 15:02, Julian Reschke <[e-mail]> wrote:
> 
> What scared me was "This would make it easier for them to make drafts
> consistent when not created by software that already does that". So
> consistency was considered more important than less markup.

For those who haven’t fallen asleep yet:

The underlying problem is that SGML created a “mixed content” approach that works well when (both!) tags can be optional; the parser simply fills in the <t>/</t> and the processor only seems the normalized form.
SGML’s optional tags allowed truly acrobatic forms of schema evolution when the DTD changed for an unchanged instance.
XML has inherited “mixed content”, but not the optional tags.
(This is still visible in the <tbody> issue that xhtml has and that v3 normalized at the expense of more noise.)

So I think that wanting to normalize mixed content at the semantic level is a well-recognized tendency with grammars that started out in SGML-land (which is true for anything derived from HTML).  The prep-tool provides a weird form of PSVI where one can indulge in that vice.  We must resist...

Grüße, Carsten