Re: += in ada

[dmytrylavrov@fsmail.net (Dmytry Lavrov)]

Russ,

How many times you're mistyped :+ as := ? (because + is on same key as
=)
And,if i'm writing
c:=a+b;
(a,b,c is not a booleans)
and making mistake
c:=a=b;
i will see compiler error.
But in
a:+b;
and a:=b;
it's compilable typo.
I think(because had very bad typos) that compilable typo is a really
worstest bug you can have.
It's spending days thinking what's wrong!(especially if almost always
a=0,and after some improvement,a<>0).

>Oh, I see. So efficiency is no longer an issue. Well, excuse me, but
I
>think we'd be better off leaving that judgment to each application
>developer. Controlling a building temperature is one thing, but
>delivering a cruise missile to Saddam Hussein's front door is quite
>another.

About speed of a:=b+c;
Some CPU's simply can't "add to ram"  (don't have such instruction),
for many other CPU's,it's NOT faster to do
mov eax,b
add a,eax
than
mov eax,a
add eax,b
mov a,eax
because in both examples,processor have to read a and b into
registers,add 'em,and then put result into ram.It's basics about how
cpu works.Ram simply can't "add value from cpu".If course,first
example is smaller,and have lesser number of instructions to
execute,but it does not make so big difference,and some x86 CPUs does
last example FASTER than first (really,you should know it if you're
worry so much about speed).

And about more complex datatypes:
it's better to write procedure than to overload an operator,especially
in safety critical system.
If you're REALLY care about speed,you should manually  simplify
expressions with,for example,vectors,for each component of
vector(huh,do yo ever think about doing it for matrices(big
enought,say,>4x4)? It's like nightmare).
And especially about matrices ,vectors,and ":*"
1:math operation need temp itself.
2: a*b<>b*a, so we should have different operators for 
a:=a*b;
and
a:=b*a;

Dmytry Lavrov.
http://dmytrylavrov.narod.ru/voxfoge.htm