Re: School Assignment--Can this be simplified?

[N J Chackowsky <chackowsky@brandonsd.mb.ca>]

Thanks for the advice/warnings/etc. I think that where I'm getting
myself confused is with which libraries are standard and which aren't. I
had *thought* that any ada.* were standard. The text I'm using (Feldman)
doesn't make this clear. I've started to plow through the reference
manual, looking for the functions available in
...generics_elementary_functions.

The reason I was using such a mix of "magic numbers", constants, and
variables, was to "code around" the multiple data types, as a result of
using ada.numerics.aux (based on double). As I get to know the libraries
better, I'll tidy up my code.

(Using gnat-3_13p-nt.exe on Win95/98.)

One problem that smote my eyes *just* after making the post was, of
course, that the calculation can be vastly simplified. Since it takes
the form
	 x
	a                (x-y)
       ----   which is a       .  Not really an Ada95-specific issue,
though.
	 y
	a

Again, thanks for the responses.

Nick J Chackowsky
Brandon, MB

Martin Dowie wrote:
> 
> A little more detail about what compile suite you are using might be
> helpful...
> 
> What is 'ada.numerics.aux' - it isn't a standard library unit?
> Why not use an instance of 'Ada.Numerics.Generic_Elementary_Functions',
> which is the standard library generic for maths functions? (some
> implementations provide 'ready-to-use' instances for float/long_float).
> 
> On the programming style front (feel free to ignore the rest of this
> response ;-)
> 
> 1. 'magic numbers' (e.g. 273.15) are repeated and would be more
>    readable as named numbers (e.g. c_to_k_factor : constant := 273.15;)
> 
> 2. Why are all the data objects 'variables' that are only ever assigned
>    once? This is a little pet-hate of mine, as by declaring them as
>    constants would help reduce target code size on some (most?)
>    compilers. I was always taught it is simply 'better programming'.
> 
>    It is a current loathing of mine simply because we are using
>    a non-optimizing compiler just now; 3 out of the original 5 s/w
>    engineers have left and their stuff is litter with this and, hey,
>    guess what, we're short on space on the target! If they had used
>    constants like this we would have an extra couple of percent -
>    not much, but for a little thought when programming, it is a couple
>    of percent we wouldn't have to be scrapping together now! As a side
>    effect, by reducing code size we increase throughput on the target
>    too :-)
> 
> 3. I best not even get into the whole 'why is everything a 'float' and
>    not user-defined float types' thing... This smacks of 'C' programs
>    not Ada programs.
> 
> N J Chackowsky wrote:
> >
> [snip]
> >
> --
> The views expressed here are personal and not those of BAE Systems.