transfer of the record length for io from internal files

[Paul Thomas]

Feng Wang has opened the door to full use of internal files with his 
recent hollerith patches.  There remains, however a need to transfer the 
record length, aka the character length, and to use it it the library io 
routines.

The enclosed patch accomplishes the transfer.  It has been regtested on 
RH9/Athlon.

This

program test
  character*80   ::  a(4)
  write(a, *) "hello, welcome to Jamaica, have a nice day"
end program test

produces

MAIN__ ()
{
  char a[4][1:80];
 
  _gfortran_filename = "test.f90";
  _gfortran_line = 3;
  _gfortran_ioparm.internal_unit = (char *) &a;
  _gfortran_ioparm.internal_unit_len = 320;
  _gfortran_ioparm.recl_in = 80;
  _gfortran_ioparm.list_format = 1;
  _gfortran_st_write ();
  _gfortran_transfer_character ("hello, welcome to Jamaica, have a nice 
day", 42);
  _gfortran_st_write_done ();
}

I have checked that the record length really is correctly picked up in 
the library as ioparm.recl_in.

thus, we now need the modifications to the library, to recognise and 
signal eor's, \n, \r and all the other good stuff.

I am open to changing the order of the arguments of set_string to making 
it more pleasing to the eye.  It was convenient to do it this way.  Are 
there any other ioparm string  options that should be using the 
character length?  I have tested one or two and they seem to be 
protected by the parser.  For example, status will not accept an array 
as its value.  If only the one option uses ts->cl, should the few lines 
of code be removed from set_string and deposited in build_dt?

Anyhow, over to you Jerry!

Best regards

Paul T
-------------- next part --------------
An embedded and charset-unspecified text was scrubbed...
Name: trans-io.diff
URL: <http://gcc.gnu.org/pipermail/fortran/attachments/20050714/ce37766d/attachment.ksh>