Re: [boost] De Bruijn Bind (alternate bind syntax) Interest?

Subject: Re: [boost] De Bruijn Bind (alternate bind syntax) Interest?
From: David Sankel (camior_at_[hidden])
Date: 2010-09-07 11:41:09

On Sun, Sep 5, 2010 at 7:56 PM, Larry Evans <cppljevans_at_[hidden]>wrote:

> On 09/04/10 13:54, David Sankel wrote:
> > On Sat, Sep 4, 2010 at 10:55 AM, Larry Evans <cppljevans_at_[hidden]
> >wrote:
> >
> >> On 09/04/10 08:24, Larry Evans wrote:
> >> [snip]
> >>> The attached illustrates the difficulty I was having with
> >>> mpl::apply.  Maybe the apply_apply typedef would be a good
> >>> use-case for De Bruijn's method.
> >> [snip]
> >> Attached is much simplified code that still illustrates problem.
> >> Compile errors included:
> >>
> >
> > Thanks for the example. If a De Bruijn indices were used, this would look
> > like
> >
> > typedef< app< lam< app< op, _1_1 > >
> >             , int_<1>
> >             >
> >        >::type apply_apply;
> >
> Typo:
>
>  typedef<
>
>
> >
> > I think this is a valid example of the needlessly cryptic bind/lambda
> > semantics that are widespread.
> >
> I tried translating the above code back to the, let's call at the
> type 'universe' as opposed to the value 'universe'.  I'm using
> universe because that's a term used by nuprl:
>
>
>
> http://www.cs.cornell.edu/Info/People/sfa/Nuprl/NuprlPrimitives/Xuniverse_doc.html
>
> to describe hierarchy of types and also because the only other name I
> could think of was domain, and that's used by proto and for other
> reasons.  Another candidate was 'kind':
>
>  http://www.haskell.org/haskellwiki/Kind
>
> but I thought of 'universe' first and it seemed more general.
>
> However, compiling my first try( the attached) gives:
> [COMPILATION]
> > make -f my-make.mk
> > /home/evansl/download/gcc/4.5.1-release/install/bin/g++ -Iinclude
> -I/home/evansl/prog_dev/boost-svn/ro/boost_1_44_0  -std=gnu++0x
> src/apply_apply.cpp src/test.cpp -o bin/apply_apply.exe
> > src/apply_apply.cpp: In function 'int main()':
> > src/apply_apply.cpp:63:28: error: cannot convert 'App<int (*)(int),
> boost::fusion::vector1<int> >' to 'int' in initialization
> > make: *** [bin/apply_apply.exe] Error 1
> [/COMPILATION]
>
> What am I doing wrong?
>

int result = app( next, 1 );

With the reference implementation, app should always be used within a lam.
In this case you don't need a lam:

int result = next( 1 ).

But if you really wanted to make a nullary function (note the extra '()' at
the end):

int result = lam<0>( app( next, 1 ) )();

Looking at the subsequent expression:

int result = app(lam_app,1);

Again we can apply the function directly here:

int result = lam_app(1);

Does that help?

-David

-- 
David Sankel
Sankel Software
www.sankelsoftware.com
585 617 4748 (Office)