Thanks for the offer Jim. I would be glad to send the entire project but I
would need your personal email as I can't send attachments through this
group. My personal email is
[hidden email] so you can reply there
is you fear making your personal email public. I would compress the project
with WinRAR and perhaps you could let me know if this is a problem. I do
know that 7-Zip can extract it. I'm using a 64-bit machine and thereby it
is possible that you could have some problems with the project that, I
think, you could fix in the project settings. Otherwise, you should not
have a problem as all classes that have "_64" at the end of their name
relate to REAL_64. Thanks again for the offer.
Regards
Chris Saunders
----- Original Message -----
From: boxer41a
To:
[hidden email]
Sent: Wednesday, September 02, 2009 4:49 PM
Subject: [eiffel_software] Re: Probably a dumb mistake, but I just can't see
it
Chris,
I am by no means the expert that the ISE folks are but I am not busy right
now so maybe I can help.
I tried to recreate your problem from the files you sent but I do not get
"0"; I get the epsilon you expected. However, the class
ALGEGRAIC_REAL_NUMBER was missing so I made some assumptions. Also, I am on
a 32 bit machine; not sure how Eiffel handles entities declared as 64 bit.
Probably doesn't work.
If you would like, zip up your files and send them to my email and I will
take a look:
jimmy.johnson at windstream.net.
jjj
--- In
[hidden email], "Chris Saunders" <evas@...> wrote:
>
> I'm going to include perhaps more code than is necessary here. I've been
> working fairly hard trying to find bugs in EiffelStudio 6.5. I've created
> two classes, one is an expanded version of the other. I'm having a probem,
> the source of which is, I think, in my code. I have been inspecting an
> testing this for a few days and just cannot find the source of the
> problem.
> First the classes and then I'll describe the problem:
>
> class
> REAL_NUMBER_32_REF
>
> inherit
> ALGEBRAIC_REAL_NUMBER [REAL_32]
> rename
> to_eiffel_type as to_real_32
> redefine
> default_create,
> is_equal,
> make
> end
>
> create
> default_create, make
>
> feature {NONE} -- Initialization
>
> default_create
> -- Initialization for `Current'.
> -- Create `item'.
> do
> epsilon := Flt_epsilon
> ensure then
> epsilon_set: epsilon = Flt_epsilon
> end
>
> make (v: like item)
> -- Initialization for `Current'.
> -- Set `item' to `v'.
> do
> epsilon := Flt_epsilon
> item := v
> ensure then
> epsilon_set: epsilon = Flt_epsilon
> end
>
> make_from_integer (v: INTEGER)
> -- Initialization for current object.
> -- Set `item' to `v'.
> do
> item := v.to_real
> epsilon := Flt_epsilon
> ensure
> item_set: item = v.to_real
> end
>
> make_from_real_64 (v: REAL_64)
> -- Initialization for current object.
> -- Set `item' to `v'.
> do
> item := v.truncated_to_real
> epsilon := Flt_epsilon
> ensure
> item_set: item = v.truncated_to_real
> end
>
> make_from_reference (v: REAL_NUMBER_32_REF)
> -- Initialization for current object.
> -- Set `item' to `v.item'.
> do
> item := v.item
> epsilon := v.epsilon
> ensure
> item_set: item = v.item
> epsilon_set: epsilon = v.epsilon
> end
>
> feature -- Access
>
> epsilon: REAL_32 assign set_epsilon
> -- Used for redifinition of `is_equal'.
>
> feature -- Comparison
>
> is_equal (other: like Current): BOOLEAN
> -- Is `other' attached to an object of the same type
> -- as current object and identical to it?
> do
> Result := (item - other.item).abs <= epsilon
> end
>
> feature -- Element change
>
> set_epsilon (v: like epsilon)
> -- Set `epsilon' to `v'.
> do
> epsilon := v
> ensure
> epsilon_set: epsilon = v
> end
>
> feature -- Conversion
>
> to_real_64: REAL_64
> -- Convert current object to a REAL_64.
> do
> Result := item.to_double
> ensure
> good_result: Result = item.to_double
> end
>
> feature {NONE} -- Externals
>
> Flt_epsilon: REAL_32
> --
> --
> external
> "C inline use <float.h>"
> alias
> "[
> return FLT_EPSILON;
> ]"
> end
>
> invariant
> epsilon_large_enough: epsilon >= Flt_epsilon
>
> end
>
> expanded class
> REAL_NUMBER_32
>
> inherit
> REAL_NUMBER_32_REF
>
> create
> default_create,
> make,
> make_from_integer,
> make_from_real_64,
> make_from_reference
>
> convert
> make ({REAL_32}),
> make_from_integer ({INTEGER}),
> make_from_real_64 ({REAL_64}),
> make_from_reference ({REAL_NUMBER_32_REF}),
> to_real_32: {REAL_32},
> to_real_64: {REAL_64}
>
> end
>
> In my test code I have this declaration "a: REAL_NUMBER_32" and what
> follows
> is part of my test code:
>
> make
> -- Run application.
> do
> a := 1.2
> print ("a.epsilon = "); print (a.epsilon); cr
> end
>
> Now, if I run the code as seen above the output is "a.epsilon = 0" and if
> I
> comment out the line "a := 1.2" the output is "a.epsilon = 1.19209e-007".
> Both `default_create' and `make_from_real_64' and all the other
> initialization procedures set `epsilon' so I can't see how `a' can ever be
> created with `epsilon set to 0. Sorry for showing this much code but I am
> stumped and thought that reducing it might cause me to not include
> something
> important. Any help truely appreciated.
>
> Regards
> Chris Saunders
>
