Re: [wsjt-devel] wsjtx-1.9.0-rc2 build failure with gcc-8.0.1

2018-03-21 Thread Jaroslav Skarvada
- Original Message - > > > - Original Message - > > > > > > - Original Message - > > > The gcc reponse to this problem is "not a bug" since the standard > > > requires > > > len=1 as the only valid thing for type c_char in the iso_c_binding > > > The FORTRAN group respon

Re: [wsjt-devel] wsjtx-1.9.0-rc2 build failure with gcc-8.0.1

2018-03-16 Thread Jaroslav Skarvada
- Original Message - > > > - Original Message - > > The gcc reponse to this problem is "not a bug" since the standard requires > > len=1 as the only valid thing for type c_char in the iso_c_binding > > The FORTRAN group response is that gcc has had a bug allowing len>1 since > >

Re: [wsjt-devel] wsjtx-1.9.0-rc2 build failure with gcc-8.0.1

2018-03-16 Thread Jaroslav Skarvada
- Original Message - > The gcc reponse to this problem is "not a bug" since the standard requires > len=1 as the only valid thing for type c_char in the iso_c_binding > The FORTRAN group response is that gcc has had a bug allowing len>1 since > this has been true since FORTRAN 2003. > >

Re: [wsjt-devel] wsjtx-1.9.0-rc2 build failure with gcc-8.0.1

2018-03-15 Thread Black Michael via wsjt-devel
The gcc reponse to this problem is "not a bug" since the standard requires len=1 as the only valid thing for type c_char in the iso_c_bindingThe FORTRAN group response is that gcc has had a bug allowing len>1 since this has been true since FORTRAN 2003. So...unless we put pressure on gcc to main

Re: [wsjt-devel] wsjtx-1.9.0-rc2 build failure with gcc-8.0.1

2018-03-15 Thread Black Michael via wsjt-devel
Nope...not the way it works.    c(10) would be an array of char * Mike On Thursday, March 15, 2018, 11:21:16 AM CDT, Jaroslav Skarvada wrote: The report for reference: https://gcc.gnu.org/bugzilla/show_bug.cgi?id=84885 But shouldn't it be: character(kind=c_char) :: c(10) to map 1:1 to

Re: [wsjt-devel] wsjtx-1.9.0-rc2 build failure with gcc-8.0.1

2018-03-15 Thread Jaroslav Skarvada
The report for reference: https://gcc.gnu.org/bugzilla/show_bug.cgi?id=84885 But shouldn't it be: character(kind=c_char) :: c(10) to map 1:1 to C language definition: char c[10]; ? Aren't we exploiting here some non-documented GNU Fortran feature? Jaroslav, OK2JRQ - Original Message -

Re: [wsjt-devel] wsjtx-1.9.0-rc2 build failure with gcc-8.0.1

2018-03-15 Thread Black Michael via wsjt-devel
I've been able to reproduce this on a small example with 8.0.1 and have submitted a bug report.This fails to compile with 8.0.1 but does compile with pre 8.0 compilersIt only shows the error when the character declaration is inside a type block. subroutine foo(i,c)   use, intrinsic :: iso_c_bin

[wsjt-devel] wsjtx-1.9.0-rc2 build failure with gcc-8.0.1

2018-03-15 Thread Jaroslav Skarvada
Hi, with gfortran 8.0.1 the compilation of wsjtx fails with the following error: make[2]: Leaving directory '/builddir/build/BUILD/wsjtx-1.9.0-rc2/wsjtx/build' jt9com.f90:44:47: Error: Component 'datetime' of BIND(C) type at (1) must have length one jt9com.f90:45:45: Error: Component 'mycall' o