On Tue, May 29, 2018 at 07:02:03PM +0100, Thomas Gummerer wrote:
> On 05/28, Junio C Hamano wrote:
> > This is a tangent, but the use of footnote below looks a but
> > curious.  How would {1} reference pick which :1: to use?  The
> > closest preceding one?
> 
> Tbh I didn't look at the docs for doing this, but just used the same
> syntax as we're already using and tried it with both asciidoc and
> asciidoctor.  And yes it seems like it always picks the preceeding
> one.

Yes, I believe the attributes namespace is flat and substituted using
the current version that's defined.  I wouldn't rely extensively on
that, though, so unique names are probably better.

> > As this appears on a page that already has other footnotes attached
> > to an adjacent paragraph, I am wondering if they should be made into
> > a part of the same numbering sequence.
> 
> I have now actually looked at the docs, and this numbering has nothing
> to do with the footnote format, but rather is used to substitute the
> attribute that's specified in the curly braces with the text that's
> after :<attribute>: [1].  This initially confused me a bit.  Maybe it
> would be nicer to give the attributes names instead of just numbers?
> As we keep adding footnotes, that would be less likely to produce
> conflicts between the different attributes I think.
> I'm also adding brian to the cc list, as he first converted this to
> AsciiDoc for opinions.

In AsciiDoc, footnotes use the named macro syntax.  I thought it would
be difficult to read to have the footnotes inline, so I chose to use an
attribute to substitute them.  I used numbers because we had a small
number of them and the original footnotes were numbered.  I was trying
to make a minimal, faithful conversion.

I have no objection to named footnotes and I agree they're easier to use
if we have a large number of them.  I think whatever we use, we should
try to make them unique, as I mentioned above.
-- 
brian m. carlson: Houston, Texas, US
OpenPGP: https://keybase.io/bk2204