Re: [xml2rfc-dev] xml2rfc: use of <boilerplate> in preptool step

[Julian Reschke <[e-mail]>]

On 11.10.2019 08:37, Martin Thomson wrote:
> On Thu, Oct 10, 2019, at 03:44, Heather Flanagan wrote:
>> I feel fairly strongly that the final XML must be self-contained. While
>> someone can derive the TOC based in section headers, deriving details
>> is not as good as just have the thing available in one place.
>
> I'm struggling to understand this argument.  In what way does having a TOC in the XML make the document more self-contained?  It only seems to add redundancy, which only invites contradiction as far as I can see.  In my view, informed by long experience with DRY (https://en.wikipedia.org/wiki/Don%27t_repeat_yourself), derivations are always far more robust.

Yes.

> Removing the redundancy, all you have is a signal about the desired *location* of the TOC, but I don't see any value in encoding that information.

Oh, that isn't supposed to be variable anyway.

>> You don’t agree. I get that. But I am insisting on this one.
>
> I get the appeal to authority bit (or assertion thereof really), but I don't think we should operate that way, and I hope that you don't expect us to accede to that either.

In particular, I would hope that the principle in question would be
applied everywhere, not just here. (Reminder: there is a least one far
more serious case where the canonical XML is not self-contained, and the
rendering depends on a concrete implementation (parts of which
undocumented), and a third-party database - Unicode - which is not
immutable).

Best regards, Julian