1. Re: math.e and misc.e
Pete Lomax wrote:
>
> Juergen Luethje wrote:
> >
> > Pete Lomax wrote:
> >
> > > }}}
<eucode>
> > > -- file getelement.e
> > > constant get_element_NIL="NI"&'L' -- avoid constant pooling
> > >
> > > global function get_element(sequence s, integer idx)
> > > if idx<1 or idx>length(s) then
> > > return get_element_NIL
> > > end if
> > > return s[idx]
> > > end function
> > >
> > > global function is_my_NIL(object x)
> > > return x===get_element_NIL
> > > end function
> > > </eucode>
{{{
> > Unfortunately, I did not understand how is_my_NIL() can distinguish a "NIL"
> > that should signal an error from a string "NIL" that is by chance part of
> > the data.
> >
> In the above I renamed NIL as get_element_NIL to make things clearer.
>
> Technically the 'distinguish a "NIL" that should signal an error from some
> other
> "NIL"' is instead 'distinguish get_element_NIL from other "NIL"'.
>
> I assumed the above would be in its own file, so no other code could ever
> reference
> the constant, or as above rename it to something that won't be.
> The function is_my_NIL(x) determines whether x is a reference pointer to the
> local constant and should return False if x points to some other "NIL".
> The key point being that the only way to get a reference pointer to that local
> constant should be that return statement in get_element().
>
> While Eu normally detects sequences as (from execute.h):
>
> SEQUENCE: 100ppppp pppppppp pppppppp pppppppp (29-bit pointer)
>
> If you avoid that (eg by using the optimised EQUAL_IFW_I which normally only
> gets integer operands) and just compare the pointers, as 32-bit dwords, then
> it will not match any other "NIL". A case of obtaining the desired feature by
> using less code.
>
> An alternative to "===" is "identity(a,b)".
>
> Regards,
> Pete
NIL wouldn't be unique?????? I hope I misunderstood you.
I have an application which uses two libraries. Some function in the main code
may takes a an argument something that comes from either library. That something
is usually not NIL, but might be if something went wrong. Does your proposal
imply that I have to care for two different NILs from the libraries, plus the one
in my app, because of the second argument of the function?
If so, it looks like a very bad idea to me. The absence is unique, because a
difference is a difference between two things and 0 < 2. Hence NIL must be unique
across any source file set, as many files as it may have.
CChris
