On Mon, 26 Jul 2010 17:41:58 +0200, Raphael Proust <raphlalou@gmail.com> wrote:
> On Mon, Jul 26, 2010 at 5:13 PM, Nicolas Pouillard
> <nicolas.pouillard@gmail.com> wrote:
> > On Mon, 26 Jul 2010 16:41:46 +0200, Raphael Proust <raphlalou@gmail.com> wrote:
> > > Hi all,
> > Hi,
> >
> > > [...]
> > >
> > > (* Pre-parsed code: *)
> > > let start = <:on< f $y$ >> in
> > > let html_node =
> > >  span ~a:[onclick start] "some text" (* a is used for (html) attributes *)
> > >
> > > (* Server side post-parsed code: *)
> > > let start _arg1 =
> > >  "call_closure(some_unique_name," ^ mymarshall _arg1 ")"
> > > in
> > > let html_node = span ~a:[onclick (start y)] "some text"
> > >
> > > (* Client side post-parsed code: *)
> > > let _ = register_closure some_unique_name (fun _arg1 -> f _arg1)
> > >
> > >
> > > I'm unsure of what is the standard way of doing such a thing in Camlp4. What
> > > I
> > > have in mind is to use the original Ocaml syntax for the quotation expander.
> > > This would (IIUC) allow me to filter the AST to transform every
> > > antiquotation
> > > found inside the quotation itself.
> > >
> > > [...]
> >
> > If the <:on<...>> contents is valid OCaml syntax I would suggest you to not go
> > in the quotations direction. You will avoid a ton of syntactic issues.
> >
> > What I suggest you is to directly use camlp4 filters, for instance by changing
> > the meaning of some constructor application and labels for antiquotations:
> >
> > (* Pre-parsed code: *)
> > let start = On (f ~y) in
> > let html_node =
> >  span ~a:[onclick start] "some text" (* a is used for (html) attributes *)
> 
> I'd rather not use a constructor because it imposes a strong limitation on the
> user (it basically reserves a keyword) whereas the <:on< ... >> solution does
> not (<:smthg< ... >> is already "reserved").

You still qualify it if you want to avoid clashes: MyModule.On

> And labeling antiquotations is far from ideal since it prevents complex
> expressions from being inlined. You basically need to write:
>   let y = h a in
>   let z = g x z in
>   On (f ~y ~z)
> instead of:
>   <:on< f $h a$ $g x z$ >>

Nothing prevents you to reserve another syntax for going back to plain OCaml code.

> What I basically need is to get an AST with antiquotations and quotations being
> special nodes. How is this achievable w/o reimplementing a whole grammar?

Doing this amounts to making the almost the same thing than
Camlp4OCamlOriginalQuotationExpander but with some filtering code.

> The alternative solution is to use raw strings, to find antiquotation marks, to
> split the string and to reinject it in the different files. Is there a way to
> keep precise _loc information this way?

Yes _loc would be hard to preserve this way.

-- 
Nicolas Pouillard
http://nicolaspouillard.fr