Thanks for the heads up!
I think too much like a mathematician sometimes; to me it is not an
immediate instinct that solving fix-points
require recursive function applications.
I changed the code to
fun{a: vt0ype}
recursive_composition
{m1, n1, m2, n2: int| n1 >= m2; m1 > n2; m2 > 0}
(pr1: processor(a, m1, n1), pr2: processor(a, m2, n2)):
processor(a, m1-n2, n1) =
let
val ~PROC(p1) = pr1
val ~PROC(p2) = pr2
val m2 = p2.ins
val n2 = p2.outs
val f1 = p1.hom
val f2 = p2.hom
val f1 = $UN.castvwtp0{ptr}(f1)
val f2 = $UN.castvwtp0{ptr}(f2)
val f = fix
f(us: signals(a, m1-n2)): signals(a, n1) =<cloptr1> let
val f1 = $UN.castvwtp0{hom(signals(a, m1), signals(a,
n1))}(f1)
val f2 = $UN.castvwtp0{hom(signals(a, m2), signals(a,
n2))}(f2)
val us' = dataget(us)
val us' = $UN.castvwtp0{signals(a, m1-n2)}(us')
in
f1[signals_append<a>{n2, m1-n2}
( f2[signals_mem<a>{m2}(signals_take<a>{n1}(f(us),
m2))]
, us' )]
end
val phi = defn(llam(us) =<cloptr1> evcloptr(f, us))
in
PROC @{ hom = phi
, ins = p1.ins - p2.outs
, outs = p1.outs }
end
and it typechecks but gives me the same compiler error, verbatim. Did I
misunderstand your hint?
Den tisdag 18 april 2017 kl. 02:31:54 UTC+2 skrev gmhwxi:
>
> You are definitely using a feature in ATS that has not been tested :)
>
> Here is a problem I see:
>
> You have
>
> val f = fix f(...) =<lincloptr1> ...
>
> As 'f' is a lincloptr, it cannot really be defined recursively (because
> 'f' is supposed to be
> called only once). To put it in another way, f1 and f2 are to be called
> and freed once the
> first call to f happens; any subsequent call to 'f' are unable to use
> either 'f1' or 'f2'.
>
> I really need to insert some form of checking to stop a lincloptr function
> from being recursively
> defined (via the use of 'fix').
>
>
> On Mon, Apr 17, 2017 at 8:02 PM, August Alm <augu...@gmail.com
> <javascript:>> wrote:
>
>> I've begun to implement the very basics of the FAUST language in ATS.
>> FAUST is a very elegant
>> functional language for audio/signal processing. The syntax is based on a
>> certain "block diagram algebra",
>> where a block is like a black-box function with some number of input
>> signals and some number of outputs,
>> and the algebra is given by five "composition" operations for grafting
>> together inputs and outputs. The block
>> diagram algebra can be mathematically explained as a (wheeled) PROP or as
>> an form of arrow enrichment.
>> I've done a lot of mathematical research on PROPs, properads and the
>> like, so recreating this in a programming
>> language is something I'm very enthusiastic about.
>> In FAUST the in-/outputs are audio signals and the blocks are audio
>> processors. FAUST is written in C++
>> and is very performant, but it is also very limited in what type an audio
>> signal can be given. (Essentially, only
>> streams of ints or floats, I believe). My idea is that it should be
>> possible to give a very direct implementation of
>> the block diagram algebra in ATS, that works for signals of any
>> [viewt0ype].
>>
>> My code so far is here:
>> https://github.com/August-Alm/ATS-Experiments/blob/master/faust.dats
>>
>> Everything works just fine except the most elusive of the five
>> operations: the [recursive_composition] which grafts
>> in- and outputs into a feedback loop. The compiler says:
>>
>> $ patscc -DATS_MEMALLOC_LIBC -o faust faust.dats
>> faust_dats.c: In function ‘__patsfun_32__32__1’:
>> faust_dats.c:9229:1: warning: implicit declaration of function
>> ‘__patsfun_32__32’ [-Wimplicit-function-declaration]
>> ATSINSmove(tmp81__1, __patsfun_32__32(env0, env1, env2, arg0)) ;
>> ^
>> In file included from faust_dats.c:15:0:
>> /usr/lib/ATS2/ccomp/runtime/pats_ccomp_instrset.h:270:35: warning:
>> assignment makes pointer from integer without a cast
>> #define ATSINSmove(tmp, val) (tmp = val)
>> ^
>> faust_dats.c:9229:1: note: in expansion of macro ‘ATSINSmove’
>> ATSINSmove(tmp81__1, __patsfun_32__32(env0, env1, env2, arg0)) ;
>> ^
>> /tmp/ccF3c5gZ.o: In function `__patsfun_32__32__1':
>> faust_dats.c:(.text+0x21e1): undefined reference to `__patsfun_32__32'
>> collect2: error: ld returned 1 exit status
>>
>> I've tried everything that I can think of. In the github code I use
>> [int]-signals but I've tried boxed types as well, and
>> various other things. I've meticulously annotated all templates, and so
>> on. Always the same compiler error. Any suggestions?
>>
>> Best wishes,
>> August
>>
>> --
>> You received this message because you are subscribed to the Google Groups
>> "ats-lang-users" group.
>> To unsubscribe from this group and stop receiving emails from it, send an
>> email to ats-lang-user...@googlegroups.com <javascript:>.
>> To post to this group, send email to ats-lan...@googlegroups.com
>> <javascript:>.
>> Visit this group at https://groups.google.com/group/ats-lang-users.
>> To view this discussion on the web visit
>> https://groups.google.com/d/msgid/ats-lang-users/87d806f5-1ed8-4173-82d6-0b3ef7075084%40googlegroups.com
>>
>> <https://groups.google.com/d/msgid/ats-lang-users/87d806f5-1ed8-4173-82d6-0b3ef7075084%40googlegroups.com?utm_medium=email&utm_source=footer>
>> .
>>
>
>
--
You received this message because you are subscribed to the Google Groups
"ats-lang-users" group.
To unsubscribe from this group and stop receiving emails from it, send an email
to ats-lang-users+unsubscr...@googlegroups.com.
To post to this group, send email to ats-lang-users@googlegroups.com.
Visit this group at https://groups.google.com/group/ats-lang-users.
To view this discussion on the web visit
https://groups.google.com/d/msgid/ats-lang-users/c4f49335-34a4-424a-b90f-05a83ad82a13%40googlegroups.com.
