Re: [xml2rfc] include processing directive

[Nico Williams <[e-mail]>]

On Tue, Nov 12, 2013 at 12:40 AM, Erik Wilde <[e-mail]> wrote:
> On 2013-11-11, 22:26 , Nico Williams wrote:
>> Agreed.  One use case I've seen a few times is where you have a large
>> ASN.1/XDR/whatever module that needs to go in an appendix but select
>> pieces of which need to be interspersed with the prose for explanatory
>> purposes.  [...]
>
> not really. unless you plan on making <artwork> itself inclusion-enabled
> (which i think would not be such a great way to go), you simply use XInclude
> and you're done. this is a working snippet from
> https://github.com/dret/I-D/blob/master/xml-patch/draft-wilde-xml-patch-08.xml,
> which requires XInclude processing (&files; is defined as
> "draft-wilde-xml-patch-08-files" in the XML):
>
> <artwork type="xml"><xi:include parse="text"
> href="&files;/patch-document.xml"/></artwork>

I guess I could live with that.  I'd rather have something somewhat
better integrated, but I suppose we've been hacking our way around
this so far reasonably well -- why stop?

> the current downside of XIPr (the XInclude processor i am using,
> https://github.com/dret/XIPr) is that you can only include complete text
> files, and not fragments. XInclude 1.1 fixes that, but i haven't implemented
> it yet. i have no idea how widely 1.1 is supported, but i would guess that
> most current XInclude implementations are 1.0 only. 1.1 support would allow
> something like this:
>
> <artwork type="xml"><xi:include parse="text"
> href="&files;/patch-document.xml#line=10,12"/></artwork>
>
> one problem is that if you support 1.1 (i.e., fragments for textual
> inclusion), once you insert one line at the start of your included file, you
> have to adjust all fragment identifiers or they'll "get out of sync"
> (integrity checks http://tools.ietf.org/html/rfc5147#section-3.1 can be used
> to detect this situation, but they won't fix it). there's no easy way around
> this. but apart from that, it's certainly much better than what i was doing
> before, i.e. copy/pasting files into the XML source.

Yeah.  No, that's awful.  I want fragment IDs.  In general all these
code modules are written in languages that permit comments, so the
obvious thing to do is to put the whole module in one file (well,
where it makes sense) and then mark it up with comment directives, and
extract fragments to Xinclude (or whatever) into the prose.  That
kinda means hacking it up every time, or at least once per-language.

Nico
--