Joseph Wright wrote:
> Hello everyone,
> 
> If you consult the LaTeX3 SVN, you'll see that I've now committed a lot
> of what has been discussed here to the real xparse module, removing
> xparse-alt and doing a little tidying of other xpackages to fit the
> updated model. At present, the update will break the "xcontent" package,
> which I'll try to sort out later today if I get time.
> 
> At present, the implementation covers all of the argument types
> discussed (including Will's optional-braced-argument, which I've called
> g = "grouped"), with "+" representing a long argument. I've gone with
> the idea of ">{\some_function:n}" for processing arguments, for example
> 
> \DeclareDocumentCommand \foo { >{\xparse_arg_split_on_comma:n}D(){0,0} }
>   { code }
> 
> is currently correct for finding an optional pair of co-ordinates and
> splitting at the comma.
> 
> I've not yet implemented \DeclareExpandableDocumentCommand. I'll
> probably try to do that over the weekend, with the very restricted set
> of valid arguments I suggested in my e-mail from yesterday.

Further to this, an implementation for \DeclareExpandableDocumentCommand
is now included in xparse in the SVN. In the end, it turned out that
with a "daisy-chain" implementation multiple optional arguments are
still okay in fully expandable commands. I've done my best to describe
the limitations of this approach in the documentation: basically, it is
never going to be as good as one that can perform assignments!

If most people are broadly happy with the resulting interface of xparse,
I'll look at sending it (plus an update to expl3: \exp_args:NnV was
missing) to CTAN in a few days time.
-- 
Joseph Wright