Hi all !
As I see it when you write a complex system it is not easy to know what
values different variables can take. A language with automatic type
conversions makes this problem even more complex. Maybe when you
designed and tested the system the variable had the type you intended.
In production the user enters a value which causes the variable to be of
a different type in which the loss of precision might occur, in which
you can get memory full, in which the calculation is suddenly extremely
slow, in which you get an exception or in which you get an unexpected
NaN or infinite result. Since I did not investigate all these
possibilities I just posed an open question about if this is relevant to
the discussion.
If the programmer controls the type he can be held accountable, when he
does not he obviously still is held accountable, but the responsibility
is harder to take and the resulting code might be more error prone.
Errors in production might well be serious and can cause things like
loss of life.
I'll be back about the strange ones J invented.
Cheers,
Erling
On 2016-08-20 19:38, Raul Miller wrote:
If loss of precision is a problem, you should be using extended precision
numbers (or rationals) for your values.
And, of course, you should be careful to use operations whose results are
extended precision numbers (or rationals) for the values you are working
with.
For example:
<.1r2+10^13x
10000000000000
Thanks,
----------------------------------------------------------------------
For information about J forums see http://www.jsoftware.com/forums.htm
