Thank you Michele.
Thank you Derick.
I turn down "phase-less" now.

I considered what I want for the library system, and I decided to take
Derick's suggestion, since my intension is different from 'explicit
phasing' or 'implicit phasing'.

  On May 7, 3:50 pm, Derick Eddington <[email protected]>
wrote:
  > it should be called always instantiating every import regardless
of its phase.

Re: explicit phasing,
Personally, I am not comfortable with 'for' keyword.

Re: implicit phasing,
I thought about implement it to ypsilon today, but its characteristics
does not sit well with me.
For example, I expect following script prints '(100 101 102).

  $ cat var.sls
  #!r6rs
  (library (var)
    (export var-ref var-set!)
    (import (rnrs))
    (define var 'unknown)
    (define (var-ref) var)
    (define (var-set! n) (set! var n))
    )

  $ cat fetch.sls
  #!r6rs
  (library (fetch)
    (export var++ var-ref var-set!)
    (import (rnrs) (var))
    (define (var++) (let ((n (var-ref))) (var-set! (+ n 1)) n))
    (var-set! 0)
    )

  $ cat test.sps
  #!r6rs
  (import (rnrs) (fetch))
  (var-set! 100)
  (display (list (var++) (var++) (var++)))
  (newline)

Re: ypsilon,
Going my way.
But make a retreat as soon as I see no way out. :-)

Thank you again!
-- fujita

On May 7, 3:50 pm, Derick Eddington <[email protected]>
wrote:
> On Thu, 2009-05-07 at 07:30 +0200, Michele Simionato wrote:
> > On Thu, May 7, 2009 at 6:29 AM, Derick Eddington
> > <[email protected]> wrote:
> > > Michele, as you've demonstrated, you don't know what you're talking
> > > about.  You seem determined to not learn in order to continue some pet
> > > fantasy.
>
> > > In implicit phasing, the phases are *implied* by where identifiers occur. 
> > >  In
> > > explicit phasing, the phases are explicitly specified.  Implicit phasing
> > > does not mean only that the (for --- (meta ---)) import syntax is not
> > > required, it means that the phases when libraries are instantiated is
> > > implied.  If libraries are always instantiated regardless of the phase
> > > of identifiers, it is not implicit phasing because what is implied is
> > > ignored, therefore the term "implicit" is inappropriate.  I should never
> > > have used the term "on-demand" because what it actually is is the
> > > essence of implicit phasing: instantiation phases happening when
> > > identifiers imply they must.
>
> > I understand this is your definition of what implicit phasing means.
> > I am not sure if Aziz thinks this is the only interpretation or if
> > he is willing to admit Ypsilon interpretation as acceptable
> > (he said he has no copyright on the term),
>
>     On Wed, 2009-05-06 at 10:44 +0300, Abdulaziz Ghuloum wrote:
>     > On May 6, 2009, at 10:11 AM, Derick Eddington wrote:
>     > > I say Ypsilon is doing implicit phasing wrong.
>     >
>     > Or maybe just applying the label incorrectly for different semantics.
>     > Maybe Fujita took "implicit phasing" to mean: the "for" syntax is
>     > ignored and identifiers are the same at all levels.  This partially
>     > (but not full) characterizes implicit phasing.  But since I have not
>     > trademarked the term, I can't sue him for using it. :-)
>
> > I interpreted "implicit" in a weak
> > sense just meaning "there is no need for (for (meta))" whereas you
> > say "no, implicit also means the phases are instantiated
> > implicitly depending if the identifiers are used!", in a strong sense.
>
> "Implicit" is only appropriate if there is something implicit.  For
> implicit phasing, phasing is what is implicit.  Explicit phasing means
> phasing is explicit, and implicit is the opposite of explicit, and both
> explicit phasing and implicit phasing are about phasing, therefore what
> explicit phasing makes explicit is what implicit phasing makes implicit.
> It is not phasing to instantiate imports when they're not needed for a
> phase.  Therefore, it's neither implicit nor explicit phasing, it should
> be called always instantiating every import regardless of its phase.
>
> --
> : Derick
> ----------------------------------------------------------------

Reply via email to