Re: Are there noticable differences in Ada acceptance by country?

[gpriv@axonx.com]

On Mar 22, 3:02 am, Simon Wright <simon.j.wri...@mac.com> wrote:
> Adam Beneschan <a...@irvine.com> writes:
> > On Mar 21, 12:21 pm, gp...@axonx.com wrote:
> >> On Mar 21, 12:39 pm, Simon Wright <simon.j.wri...@mac.com> wrote:
>
> >> > gp...@axonx.com writes:
> >> > > From my own C experience: using code from different sources X and Y
> >> > > coordinates in function calls were in different order: foo(x,y);
> >> > > boo(height, width), etc. Easy to overlook.
>
> >> > That was my point (I expect I failed to make it properly): if you use
> >> > a record for Coordinate, your chance of getting the components the
> >> > wrong way round is hugely reduced, and I don't think the added value
> >> > of making the H & V components different _types_ is worth the added
> >> > cost.
>
> >> Is there any runtime cost of using new type if no range is defined?
>
> > Very unlikely, I think.  I think the type of "cost" Simon was talking
> > about was the added burden (on both writer and reader) of additional
> > type conversions cluttering up the code, when (in this case) they
> > probably accomplish very little in the way of preventing programmer
> > errors.
>
> That was it, precisely: thanks, Adam!

Sorry for misunderstanding. My initial post had to do with 2-d pixel-
coordinate related issues.  These were integers there and no
transforms used.  For true N-d geometry it's certainly is "cleaner"
design to use vectors and matrices preferably hiding implementation
details.

G.