[xml2rfc] xml:space in v3.rnc (Re: Insufficiency of txt format)

[Carsten Bormann <[e-mail]>]

On 2021-02-01, at 18:21, Carsten Bormann <[e-mail]> wrote:
> 
> white-space handling

And while we are talking about this:

What the [insert Unicode characters here] is the story behind the two occurrences of xml:space in v3.rnc.

   spanx =
     element spanx { […]
       [ a:defaultValue = "preserve" ]
       attribute xml:space { "default" | "preserve" }?,

So, for a (legacy) <spanx> specifically, xml:space defaults to “preserve”.

For artwork:

   artwork =
     element artwork { […]
       attribute xml:space { text }?,

… it is allowed to set xml:space to any value (invalid in XML), but there is no default.  Also, xml:space is not allowed for <sourcecode>.

Contrast this to v2.rnc:

   spanx =
     element spanx {
       [ a:defaultValue = "preserve" ]
       attribute xml:space { "default" | "preserve" }?,

So this is unchanged in v3 from v2.

   artwork =
     element artwork {
       [ a:defaultValue = "preserve" ]
       attribute xml:space { "default" | "preserve" }?,

… makes sense if you think that xml:space is going to be useful at all.

This was discussed in <https://mailarchive.ietf.org/arch/msg/xml2rfc-dev/U7ounlQdYz95U3I-PofVgheYmPI>, but apart from agreeing that xml:space is useless, not much light has been generated.

There are a few more mentions further in history that I can find, but none of them really discussed how the situation of just <spanx> and <artwork> using xml:space and the v3 weirdness around this occurred.

RFC 7991 says:

   o  Deprecate the "xml:space" attribute in <artwork> because there was
      only one useful value.  […]

… and dutifully notes the above change in the RNC diff.  
Er, since only one value is useful, we allow any value and then let the default lapse to the non-useful one?

Most likely, there is no reason (because nothing actually is using this attribute), so this can be cleaned up, but I sure would like to know if there is.  Should an authoring tool generate any of these?

(There is also the SVG issue, which I’ll pass now.)

Grüße, Carsten