forums.silverfrost.com

[dgurok]

I know that the assigned GO TO statement has been deleted in Fortran 95 but we try to port your Fortran 77 programs to Fortran 95 now.

You write that the feature is not deleted from ftn95, so I can compile and link my program but I get an access violation at this point when I run it!

My simple program:

[LitusSaxonicum]

When the error message says you can only ASSIGN to integer (KIND=3), that means to INTEGER* 4!

INTEGER*4 LABEL1

Eddie

[dgurok]

there are no errors while compiling

ftn95 test.for /debug

[FTN95/Win32 Ver. 6.10.0 Copyright (c) Silverfrost Ltd 1993-2011]
NO ERRORS [<LTEST> FTN95/Win32 v6.10.0]


When I execute the program I get the error:
---------------------------------------------------
Access Violation

main - in file test.for at line 6 [+0021]
---------------------------------------------------

line 6 is the "ASSIGN 50 TO LABEL1 " statement...

[LitusSaxonicum]

I'm using an older version. Have you tried it with INTEGER*4? All labels must be 32-bit.

Eddie

[dgurok]

When I use INTEGER*4 then I get this error:

[davidb]

Your code works for me if I use INTEGER*4 or INTEGER(Kind=3) or INTEGER. All of these are same 32 bit integer type.

[dgurok]

Thats strang...

Which version of ftn95 do you use??

Do you have some special compilation settings??

I use these:

[LitusSaxonicum]

I use 4.90 (Academic licence). For some bizarre and unfathomable reason that I do not understand, this machine (at my University) appears to be licensed for 5.20.1, but nevertheless it has my name on it !

I only use the Win32 / Clearwin options, and as far as I know, don't have anything set as standard in a config file.

/SINGLE_THREADED looks to be a hangover from the DBOS version of FTN77 - is it still valid?

Eddie

[PaulLaidler]

/ALT_KINDS will mess things up for you.

You need INTEGER(kind = 4) when using /ALT_KINDS.

That is, the kind values are 1, 2, 4 and 8 when you use ALT_KINDS.

[LitusSaxonicum]

Paul,

ALT_KINDS will upset the apple cart if you use KIND. It shouldn't for INTEGER*2, surely?

What puzzles me is that he doesn't get an error message. There is a very clear message with my licensed version, and also PE version 6.10.

Eddie

[PaulLaidler]

With the current version of FTN95 you should be able to use


INTEGER if the default KIND has not been changed

INTEGER*4

INTEGER(KIND=3) if the default KIND values have not been changed

INTEGER(KIND=4) if /ALT_KINDS has been used

Anything else should produce a compile time error.