On Mon, 2004-04-26 at 23:15, Leopold Toetsch wrote:
> > Per my reading of the NCI PDD and the code, this isn't yet supported.
>
> Then the POD is missing some items:
>
> ,--[ src/call_list.txt ]
> | # 2 - pointer to short
> | # 3 - pointer to int
> | # 4 - pointer to long
> `-----------------------
Patch attached for those.
> > (width, height) = SizeText(font, text);
>
> This signature doesn't match the call's signature. Shuffling return
> values around like that isn't really pleasant and you don't get it that
> way from C.
That's true. Still, there has to be some extra support for out
parameters, or else NCI will walk all over the newly-updated contents of
the registers with the return results.
I patched pcf_i_pt33() manually to prevent it from assigning to I5 and
the following call does exactly what I want:
(width, height) = SizeText( font, text, width, height )
I don't particularly mind if I have to give it a signature of 'ipt3!3!'
or whatever to signify that I want the NCI sub to write into the
appropriate registers, but I do want to be able to fetch the values it
writes.
-- c
Index: docs/pdds/pdd16_native_call.pod
===================================================================
RCS file: /cvs/public/parrot/docs/pdds/pdd16_native_call.pod,v
retrieving revision 1.7
diff -u -u -r1.7 pdd16_native_call.pod
--- docs/pdds/pdd16_native_call.pod 2 Mar 2004 12:25:01 -0000 1.7
+++ docs/pdds/pdd16_native_call.pod 28 Apr 2004 03:42:10 -0000
@@ -75,6 +75,18 @@
long. An integer type. You know the drill.
+=item 2
+
+A pointer to a short, taken from an I register.
+
+=item 3
+
+A pointer to an int, taken from an I register.
+
+=item 4
+
+A pointer to a long, taken from an I register.
+
=item f
float. F register denizen.
