Re: Allow <info> as root element?

Hi Dave,

On 30.11.22 12:00, Dave Pawson wrote:
> [...]
>
> Or a burden on writers to take care what markup they use?
> Sounds reasonable?

To some degree, yes. Especially if you have to deal with incorporating par=
ts of
a XML file, referencing etc.


> Your info 'root', I don't see how reasonable use would invalidate
> the use case (multiple) imports into n documents?

Sure it can work. However, if I load a file with <info> as root, my XML ed=
itor
_will_ complain. To avoid it, I'm forced to either a) make a DocBook
customization and allow <info>, or b) wrap artificial structural DocBook
elements around. Both are problematic in their own ways as I was trying to=
 explain.


> [...]
>> Yes, technically possible, but still a restriction. Additionally, I'm f=
orced to
>> add an XPointer expression into the <xi:include> element just to select=
 the
>> info. Sounds like a detour to me. :)
>
> No, sounds like you making use of xinclude and docbook for your own purp=
oses?
> Something many of us do and have done for a long time.

XIncludes aren't the problem per se. Maybe it works for your documents. I'=
m
glad it works for you. But not all documents are the same. If we can't use
<info> as root, we are forced to do all sorts of workarounds. Perhaps it w=
as
always the case.

However, in terms of topic-oriented writing, it becomes annoying when you =
are
restricted by the schema.


> I'm not sure psychological 'barriers' are other than local use constrain=
ts?

What I'm trying to say is, if we sell validation as sacred and as a good t=
hing,
we can't undermine this by allowing errors in another part, can we? Call i=
t
constraints, psychological barriers, or whatever you like. :-)

Probably it boils down to that DocBook allows many ways to process it. Som=
e
produce valid results other invalid intermediate results or something in
between. DocBook doesn't define one way to do it, for the better or worse.


=2D-
Gru=C3=9F/Regards
   Thomas Schraitle