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

[Julian Reschke <[e-mail]>]

Am 29.06.2021 um 18:05 schrieb Carsten Bormann:
>>> As a format for authoring, making the input vastly more complicated in the name of some lofty goal (here: standards conformance) is exactly the leading symptom of the v3 development process.  Offering a bespoke, efficient(*) way of keyboarding IETF document references is exactly the right way to handle this, at least for people still authoring in XML.
>>
>> It is not "vastly more complicated". It's a different syntax (with the
>> benefit of it not being PI-based which is yet another XML complication
>> for most authors). (*)
>
> Yes.  Not using PIs means that this feature now comes under the strict eyes of the chaperone, the RNG validation.
> So we all have to fix up the RFCXML RNC to make this feature available in various environments.

Concrete example?

If this really is a problem in practice, we can of course publish a
schema that takes this into account (a pre-include schema).

FWIW, with the old style PIs, documents by definition never were valid
because of dangling anchors.

> This is not progress.
>
> Also, the fact that there now need to be full URIs instead of more reasonable document identifiers is confusing people.
> (Not that the old ones were particularly good, but they should have been simplified, not pessimized.)
>
> Yes, PIs are unclean in some universe.  They are also useful for adding features where being clean gets in the way.

These are orthogonal issues. As I said above, a v2 document using PIs
for includes by definition can not be valid, so I'm not sure how the
switch to x:include makes things worse. There's even a tiny chance of
things being better if the validator applies x:include before validation
(which it should).

Yes, we could have added an inclusion mechanism which is not generic,
and special-cases reference inclusion. We still can do that if people
feel that's important.

>> IFIUC, you miss the feature of not having to specify the complete URL of
>> the reference? (Or is there something else I'm not aware of?).
>
> Typing full URLs is not a good use of author time.

You don't type them - you paste them.

>> (*) And, BTW the *other* way to include stuff as explained in
>> <https://greenbytes.de/tech/webdav/rfc7749.html#including.files>
>> continues to work, as it's just a basic XML feature.
>
> Which is, of course the one I used during the first 10 years of using RFCXML; I wasn’t even aware about the PI form before I moved away from keyboarding XML.
> But others were (and some still use XML), and I’m trying to be their advocate here.

Can we hear from them, please?

Best regards, Julian