Re: [rfc-i] deprecating relref

[Julian Reschke <[e-mail]>]

Am 09.04.2021 um 21:57 schrieb Carsten Bormann:
> On 2021-04-09, at 21:38, Peter Saint-Andre <[e-mail]> wrote:
>>
>> On 4/9/21 2:18 AM, Julian Reschke wrote:
>>
>>> FWIW, I believe the direction to insert *all* defaults is totally
>>> misguided; it leads to "canonical" XML that's full with completely
>>> useless information.
>>
>> Agreed. The whole point of a default is you don't need to include it (as
>> long as the schema doesn't change from under you, so let's not do that).
>
> XML was meant to be useful without a schema.
>
> Requiring canonical (“prepped”) RFCXML to be a PSVI makes a lot of sense.
>
> It makes less sense to allow attributes to be present when they don’t mean anything.
> ...

Absolutely.

We should have a close look at these defaults, and either exclude some,
or switch to a positive list.

For example:

   <section anchor="toc" numbered="false" removeInRFC="false"
toc="exclude" pn="section-toc.1">

Of the attributes inserted here, which ones would we ever change the
defaults for?

Or here:

   <t indent="0" pn="section-1-1">

Inserting things that at the preptool stage isn't bad per se, we just
need to keep in mind what it is for.

There are other things where we do *not* insert generated stuff. For
instance, the code for <u> and <address> consults external resources not
under IETF control for content, but then does not insert that
information into the canonoical XML.

On the other hand, stuff is inlined where it doesn't make any sense at
all, like the Toc or the Authors section are inserted, with very
interesting side effects on the author visible grammar.

Best regards, Julian
_______________________________________________
rfc-interest mailing list
[e-mail]
https://www.rfc-editor.org/mailman/listinfo/rfc-interest