On Thu, Nov 01, 2018 at 12:01:28AM +0100, Anders Waldenborg wrote:
> Jeff King writes:
>
> > On the other hand, if the rule were not "this affects the next
> > placeholder" but had a true ending mark, then we could make a real
> > parse-tree out of it, and format chunks of placeholders. E.g.:
> >
> > %(format:lpad=30,filename)%(subject) %(authordate)%(end)
> >
> > would pad and format the whole string with two placeholders. I know that
> > going down this road eventually involves reinventing XML, but I think
> > having an actual tree structure may not be an unreasonable thing to
> > shoot for.
>
> Yes. I'm thinking that with [] for formatting specifiers and () for
> placeholders, {} would be available for nesting. E.g:
>
> %[lpad=30,mangle]{%(subject) %ad%}
Hmm. That's kind of ugly, but probably not really any uglier than any of
the things I showed. And it has the advantage that we could implement
%[] now, and later extend it (well, I guess we'd want to make sure that
"%[lpad=30]{foo}" does not treat the curly braces literally, since we'd
eventually make them syntactically significant).
> I'm planning to work on the initial "trailer:key=" part later this
> week. Maybe I can play around with different formatting options and see
> how it affects the parser.
Great! Thanks for working on this.
-Peff
