forums.silverfrost.com

[Anonymous]

Hello everybody,

As a novice Salfort FTN95 user I just discovered the following strange phenomenon.
Consider the following piece of code:

MODULE INTEGER_KINDS
INTEGER, PARAMETER :: bytes_1 = SELECTED_INT_KIND(2)
INTEGER, PARAMETER :: bytes_2 = SELECTED_INT_KIND(4)
INTEGER, PARAMETER :: bytes_4 = SELECTED_INT_KIND(9)
INTEGER, PARAMETER :: bytes_8 = SELECTED_INT_KIND(1
END MODULE INTEGER_KINDS

PROGRAM MAIN
USE INTEGER_KINDS
INTEGER (KIND=bytes_1) :: n=127
INTEGER (KIND=bytes_2) :: nn=32767
INTEGER (KIND=bytes_4) :: nnn=2147483647
INTEGER (KIND=bytes_8) :: nnnn=9223372036854775807_bytes_8

PRINT *, n, BIT_SIZE(n)
PRINT *, nn, BIT_SIZE(nn)
PRINT *, nnn, BIT_SIZE(nnn)
PRINT *, nnnn, BIT_SIZE(nnnn)
STOP
END

This produces the result:
127 8
32767 16
2147483647 32
9223372036854775807 32

I expected the last value to be 64 in stead of 32!!!
What is happening here?
Am I doing something wrong?

Hoping for a quick feedback!
Kind regards,
Lucas.

[PaulLaidler]

Lucas

FTN95 is not yet fully functional for INTEGER*8 so you may find some intrinsics that do not support 64 bit integers.

However, BIT_SIZE should work OK from version 4.91.

[Anonymous]

Ok Paul,

Thanks for the info.
Any idea when version 4.91 should be available?

[PaulLaidler]

Lucas

It is already available.
See the Announcements.

[Anonymous]

Absolutely right, Paul.

I must have overlooked it, due to my enthousiasm, or the heat...
Sorry.
But thanks for correcting me.

Lucas.

[Anonymous]

Well Paul, you were right again.

I have now installed version 4.91 of FTN95PE, and the result of the BIT_SIZE intrinsic function is now indeed 64 for INTEGER*8 variables.
However I had to install the Microsoft .NET v.1.1.4423 first, otherwise I couldn't do the upgrade from 4.90 to 4.91.

Thanks again for pointing me in the right direction!
Lucas.