|
|
9805c9 |
From f3e3034684c7ac44a14c70d6a248d8acee303176 Mon Sep 17 00:00:00 2001
|
|
|
9805c9 |
From: law <law@138bc75d-0d04-0410-961f-82ee72b054a4>
|
|
|
9805c9 |
Date: Thu, 10 May 2018 11:48:34 +0100
|
|
|
9805c9 |
Subject: [PATCH 01/16] Default widths for i, f and g format specifiers in
|
|
|
9805c9 |
format strings.
|
|
|
9805c9 |
|
|
|
9805c9 |
Enabled using -fdec.
|
|
|
9805c9 |
|
|
|
9805c9 |
The behaviour is modelled on the Oracle Fortran compiler. At the time
|
|
|
9805c9 |
of writing, the details were available at this URL:
|
|
|
9805c9 |
|
|
|
9805c9 |
https://docs.oracle.com/cd/E19957-01/805-4939/6j4m0vnc3/index.html#z4000743746d
|
|
|
9805c9 |
|
|
|
9805c9 |
Addition by Mark Eggleston <mark.eggleston@codethink.com>:
|
|
|
9805c9 |
|
|
|
9805c9 |
Use -fdec-format-defaults to enable this feature. Also enabled using -fdec.
|
|
|
9805c9 |
---
|
|
|
9805c9 |
gcc/fortran/io.c | 31 +++++++++++--
|
|
|
9805c9 |
gcc/fortran/lang.opt | 4 ++
|
|
|
9805c9 |
gcc/fortran/options.c | 1 +
|
|
|
9805c9 |
.../gfortran.dg/fmt_f_default_field_width_1.f90 | 43 ++++++++++++++++++
|
|
|
9805c9 |
.../gfortran.dg/fmt_f_default_field_width_2.f90 | 46 +++++++++++++++++++
|
|
|
9805c9 |
.../gfortran.dg/fmt_f_default_field_width_3.f90 | 28 ++++++++++++
|
|
|
9805c9 |
.../gfortran.dg/fmt_g_default_field_width_1.f90 | 48 ++++++++++++++++++++
|
|
|
9805c9 |
.../gfortran.dg/fmt_g_default_field_width_2.f90 | 52 ++++++++++++++++++++++
|
|
|
9805c9 |
.../gfortran.dg/fmt_g_default_field_width_3.f90 | 31 +++++++++++++
|
|
|
9805c9 |
.../gfortran.dg/fmt_i_default_field_width_1.f90 | 38 ++++++++++++++++
|
|
|
9805c9 |
.../gfortran.dg/fmt_i_default_field_width_2.f90 | 42 +++++++++++++++++
|
|
|
9805c9 |
.../gfortran.dg/fmt_i_default_field_width_3.f90 | 35 +++++++++++++++
|
|
|
9805c9 |
libgfortran/io/format.c | 35 +++++++++++++++
|
|
|
9805c9 |
libgfortran/io/io.h | 50 +++++++++++++++++++++
|
|
|
9805c9 |
libgfortran/io/read.c | 6 +++
|
|
|
9805c9 |
libgfortran/io/write.c | 22 +++++----
|
|
|
9805c9 |
libgfortran/io/write_float.def | 37 ++++++++++++---
|
|
|
9805c9 |
17 files changed, 531 insertions(+), 18 deletions(-)
|
|
|
9805c9 |
create mode 100644 gcc/testsuite/gfortran.dg/fmt_f_default_field_width_1.f90
|
|
|
9805c9 |
create mode 100644 gcc/testsuite/gfortran.dg/fmt_f_default_field_width_2.f90
|
|
|
9805c9 |
create mode 100644 gcc/testsuite/gfortran.dg/fmt_f_default_field_width_3.f90
|
|
|
9805c9 |
create mode 100644 gcc/testsuite/gfortran.dg/fmt_g_default_field_width_1.f90
|
|
|
9805c9 |
create mode 100644 gcc/testsuite/gfortran.dg/fmt_g_default_field_width_2.f90
|
|
|
9805c9 |
create mode 100644 gcc/testsuite/gfortran.dg/fmt_g_default_field_width_3.f90
|
|
|
9805c9 |
create mode 100644 gcc/testsuite/gfortran.dg/fmt_i_default_field_width_1.f90
|
|
|
9805c9 |
create mode 100644 gcc/testsuite/gfortran.dg/fmt_i_default_field_width_2.f90
|
|
|
9805c9 |
create mode 100644 gcc/testsuite/gfortran.dg/fmt_i_default_field_width_3.f90
|
|
|
9805c9 |
|
|
|
9805c9 |
diff --git a/gcc/fortran/io.c b/gcc/fortran/io.c
|
|
|
9805c9 |
index 9828897852a..57117579627 100644
|
|
|
9805c9 |
--- a/gcc/fortran/io.c
|
|
|
9805c9 |
+++ b/gcc/fortran/io.c
|
|
|
9805c9 |
@@ -903,6 +903,13 @@ data_desc:
|
|
|
9805c9 |
|
|
|
9805c9 |
if (u != FMT_POSINT)
|
|
|
9805c9 |
{
|
|
|
9805c9 |
+ if (flag_dec_format_defaults)
|
|
|
9805c9 |
+ {
|
|
|
9805c9 |
+ /* Assume a default width based on the variable size. */
|
|
|
9805c9 |
+ saved_token = u;
|
|
|
9805c9 |
+ break;
|
|
|
9805c9 |
+ }
|
|
|
9805c9 |
+
|
|
|
9805c9 |
format_locus.nextc += format_string_pos;
|
|
|
9805c9 |
gfc_error ("Positive width required in format "
|
|
|
9805c9 |
"specifier %s at %L", token_to_string (t),
|
|
|
9805c9 |
@@ -1027,6 +1034,13 @@ data_desc:
|
|
|
9805c9 |
goto fail;
|
|
|
9805c9 |
if (t != FMT_ZERO && t != FMT_POSINT)
|
|
|
9805c9 |
{
|
|
|
9805c9 |
+ if (flag_dec_format_defaults)
|
|
|
9805c9 |
+ {
|
|
|
9805c9 |
+ /* Assume the default width is expected here and continue lexing. */
|
|
|
9805c9 |
+ value = 0; /* It doesn't matter what we set the value to here. */
|
|
|
9805c9 |
+ saved_token = t;
|
|
|
9805c9 |
+ break;
|
|
|
9805c9 |
+ }
|
|
|
9805c9 |
error = nonneg_required;
|
|
|
9805c9 |
goto syntax;
|
|
|
9805c9 |
}
|
|
|
9805c9 |
@@ -1096,8 +1110,17 @@ data_desc:
|
|
|
9805c9 |
goto fail;
|
|
|
9805c9 |
if (t != FMT_ZERO && t != FMT_POSINT)
|
|
|
9805c9 |
{
|
|
|
9805c9 |
- error = nonneg_required;
|
|
|
9805c9 |
- goto syntax;
|
|
|
9805c9 |
+ if (flag_dec_format_defaults)
|
|
|
9805c9 |
+ {
|
|
|
9805c9 |
+ /* Assume the default width is expected here and continue lexing. */
|
|
|
9805c9 |
+ value = 0; /* It doesn't matter what we set the value to here. */
|
|
|
9805c9 |
+ saved_token = t;
|
|
|
9805c9 |
+ }
|
|
|
9805c9 |
+ else
|
|
|
9805c9 |
+ {
|
|
|
9805c9 |
+ error = nonneg_required;
|
|
|
9805c9 |
+ goto syntax;
|
|
|
9805c9 |
+ }
|
|
|
9805c9 |
}
|
|
|
9805c9 |
else if (is_input && t == FMT_ZERO)
|
|
|
9805c9 |
{
|
|
|
9805c9 |
@@ -4368,8 +4391,8 @@ get_io_list:
|
|
|
9805c9 |
}
|
|
|
9805c9 |
|
|
|
9805c9 |
/* See if we want to use defaults for missing exponents in real transfers
|
|
|
9805c9 |
- and other DEC runtime extensions. */
|
|
|
9805c9 |
- if (flag_dec)
|
|
|
9805c9 |
+ and other DEC runtime extensions. */
|
|
|
9805c9 |
+ if (flag_dec_format_defaults)
|
|
|
9805c9 |
dt->dec_ext = 1;
|
|
|
9805c9 |
|
|
|
9805c9 |
/* A full IO statement has been matched. Check the constraints. spec_end is
|
|
|
9805c9 |
diff --git a/gcc/fortran/lang.opt b/gcc/fortran/lang.opt
|
|
|
9805c9 |
index 9151d02c491..26e82601b62 100644
|
|
|
9805c9 |
--- a/gcc/fortran/lang.opt
|
|
|
9805c9 |
+++ b/gcc/fortran/lang.opt
|
|
|
9805c9 |
@@ -444,6 +444,10 @@ fdec-include
|
|
|
9805c9 |
Fortran Var(flag_dec_include)
|
|
|
9805c9 |
Enable legacy parsing of INCLUDE as statement.
|
|
|
9805c9 |
|
|
|
9805c9 |
+fdec-format-defaults
|
|
|
9805c9 |
+Fortran Var(flag_dec_format_defaults)
|
|
|
9805c9 |
+Enable default widths for i, f and g format specifiers.
|
|
|
9805c9 |
+
|
|
|
9805c9 |
fdec-intrinsic-ints
|
|
|
9805c9 |
Fortran Var(flag_dec_intrinsic_ints)
|
|
|
9805c9 |
Enable kind-specific variants of integer intrinsic functions.
|
|
|
9805c9 |
diff --git a/gcc/fortran/options.c b/gcc/fortran/options.c
|
|
|
9805c9 |
index 02970d59066..4f91486e977 100644
|
|
|
9805c9 |
--- a/gcc/fortran/options.c
|
|
|
9805c9 |
+++ b/gcc/fortran/options.c
|
|
|
9805c9 |
@@ -74,6 +74,7 @@ set_dec_flags (int value)
|
|
|
9805c9 |
SET_BITFLAG (flag_dec_static, value, value);
|
|
|
9805c9 |
SET_BITFLAG (flag_dec_math, value, value);
|
|
|
9805c9 |
SET_BITFLAG (flag_dec_include, value, value);
|
|
|
9805c9 |
+ SET_BITFLAG (flag_dec_format_defaults, value, value);
|
|
|
9805c9 |
}
|
|
|
9805c9 |
|
|
|
9805c9 |
/* Finalize DEC flags. */
|
|
|
9805c9 |
diff --git a/gcc/testsuite/gfortran.dg/fmt_f_default_field_width_1.f90 b/gcc/testsuite/gfortran.dg/fmt_f_default_field_width_1.f90
|
|
|
9805c9 |
new file mode 100644
|
|
|
9805c9 |
index 00000000000..49c77155761
|
|
|
9805c9 |
--- /dev/null
|
|
|
9805c9 |
+++ b/gcc/testsuite/gfortran.dg/fmt_f_default_field_width_1.f90
|
|
|
9805c9 |
@@ -0,0 +1,43 @@
|
|
|
9805c9 |
+! { dg-do run }
|
|
|
9805c9 |
+! { dg-options -fdec }
|
|
|
9805c9 |
+!
|
|
|
9805c9 |
+! Test case for the default field widths enabled by the -fdec-format-defaults flag.
|
|
|
9805c9 |
+!
|
|
|
9805c9 |
+! This feature is not part of any Fortran standard, but it is supported by the
|
|
|
9805c9 |
+! Oracle Fortran compiler and others.
|
|
|
9805c9 |
+!
|
|
|
9805c9 |
+! libgfortran uses printf() internally to implement FORMAT. If you print float
|
|
|
9805c9 |
+! values to a higher precision than the type can actually store, the results
|
|
|
9805c9 |
+! are implementation dependent: some platforms print zeros, others print random
|
|
|
9805c9 |
+! numbers. Don't depend on this behaviour in tests because they will not be
|
|
|
9805c9 |
+! portable.
|
|
|
9805c9 |
+
|
|
|
9805c9 |
+ character(50) :: buffer
|
|
|
9805c9 |
+
|
|
|
9805c9 |
+ real*4 :: real_4
|
|
|
9805c9 |
+ real*8 :: real_8
|
|
|
9805c9 |
+ real*16 :: real_16
|
|
|
9805c9 |
+ integer :: len
|
|
|
9805c9 |
+
|
|
|
9805c9 |
+ real_4 = 4.18
|
|
|
9805c9 |
+ write(buffer, '(A, F, A)') ':',real_4,':'
|
|
|
9805c9 |
+ print *,buffer
|
|
|
9805c9 |
+ if (buffer.ne.": 4.1799998:") stop 1
|
|
|
9805c9 |
+
|
|
|
9805c9 |
+ real_4 = 0.00000018
|
|
|
9805c9 |
+ write(buffer, '(A, F, A)') ':',real_4,':'
|
|
|
9805c9 |
+ print *,buffer
|
|
|
9805c9 |
+ if (buffer.ne.": 0.0000002:") stop 2
|
|
|
9805c9 |
+
|
|
|
9805c9 |
+ real_8 = 4.18
|
|
|
9805c9 |
+ write(buffer, '(A, F, A)') ':',real_8,':'
|
|
|
9805c9 |
+ print *,buffer
|
|
|
9805c9 |
+ len = len_trim(buffer)
|
|
|
9805c9 |
+ if (len /= 27) stop 3
|
|
|
9805c9 |
+
|
|
|
9805c9 |
+ real_16 = 4.18
|
|
|
9805c9 |
+ write(buffer, '(A, F, A)') ':',real_16,':'
|
|
|
9805c9 |
+ print *,buffer
|
|
|
9805c9 |
+ len = len_trim(buffer)
|
|
|
9805c9 |
+ if (len /= 44) stop 4
|
|
|
9805c9 |
+end
|
|
|
9805c9 |
diff --git a/gcc/testsuite/gfortran.dg/fmt_f_default_field_width_2.f90 b/gcc/testsuite/gfortran.dg/fmt_f_default_field_width_2.f90
|
|
|
9805c9 |
new file mode 100644
|
|
|
9805c9 |
index 00000000000..1c2ec0413a7
|
|
|
9805c9 |
--- /dev/null
|
|
|
9805c9 |
+++ b/gcc/testsuite/gfortran.dg/fmt_f_default_field_width_2.f90
|
|
|
9805c9 |
@@ -0,0 +1,46 @@
|
|
|
9805c9 |
+! { dg-do run }
|
|
|
9805c9 |
+! { dg-options -fdec-format-defaults }
|
|
|
9805c9 |
+!
|
|
|
9805c9 |
+! Test case for the default field widths enabled by the -fdec-format-defaults flag.
|
|
|
9805c9 |
+!
|
|
|
9805c9 |
+! This feature is not part of any Fortran standard, but it is supported by the
|
|
|
9805c9 |
+! Oracle Fortran compiler and others.
|
|
|
9805c9 |
+!
|
|
|
9805c9 |
+! libgfortran uses printf() internally to implement FORMAT. If you print float
|
|
|
9805c9 |
+! values to a higher precision than the type can actually store, the results
|
|
|
9805c9 |
+! are implementation dependent: some platforms print zeros, others print random
|
|
|
9805c9 |
+! numbers. Don't depend on this behaviour in tests because they will not be
|
|
|
9805c9 |
+! portable.
|
|
|
9805c9 |
+!
|
|
|
9805c9 |
+! Test case added by Mark Eggleston <mark.eggleston@codethink.com> to check
|
|
|
9805c9 |
+! use of -fdec-format-defaults
|
|
|
9805c9 |
+!
|
|
|
9805c9 |
+ character(50) :: buffer
|
|
|
9805c9 |
+
|
|
|
9805c9 |
+ real*4 :: real_4
|
|
|
9805c9 |
+ real*8 :: real_8
|
|
|
9805c9 |
+ real*16 :: real_16
|
|
|
9805c9 |
+ integer :: len
|
|
|
9805c9 |
+
|
|
|
9805c9 |
+ real_4 = 4.18
|
|
|
9805c9 |
+ write(buffer, '(A, F, A)') ':',real_4,':'
|
|
|
9805c9 |
+ print *,buffer
|
|
|
9805c9 |
+ if (buffer.ne.": 4.1799998:") stop 1
|
|
|
9805c9 |
+
|
|
|
9805c9 |
+ real_4 = 0.00000018
|
|
|
9805c9 |
+ write(buffer, '(A, F, A)') ':',real_4,':'
|
|
|
9805c9 |
+ print *,buffer
|
|
|
9805c9 |
+ if (buffer.ne.": 0.0000002:") stop 2
|
|
|
9805c9 |
+
|
|
|
9805c9 |
+ real_8 = 4.18
|
|
|
9805c9 |
+ write(buffer, '(A, F, A)') ':',real_8,':'
|
|
|
9805c9 |
+ print *,buffer
|
|
|
9805c9 |
+ len = len_trim(buffer)
|
|
|
9805c9 |
+ if (len /= 27) stop 3
|
|
|
9805c9 |
+
|
|
|
9805c9 |
+ real_16 = 4.18
|
|
|
9805c9 |
+ write(buffer, '(A, F, A)') ':',real_16,':'
|
|
|
9805c9 |
+ print *,buffer
|
|
|
9805c9 |
+ len = len_trim(buffer)
|
|
|
9805c9 |
+ if (len /= 44) stop 4
|
|
|
9805c9 |
+end
|
|
|
9805c9 |
diff --git a/gcc/testsuite/gfortran.dg/fmt_f_default_field_width_3.f90 b/gcc/testsuite/gfortran.dg/fmt_f_default_field_width_3.f90
|
|
|
9805c9 |
new file mode 100644
|
|
|
9805c9 |
index 00000000000..e513063189b
|
|
|
9805c9 |
--- /dev/null
|
|
|
9805c9 |
+++ b/gcc/testsuite/gfortran.dg/fmt_f_default_field_width_3.f90
|
|
|
9805c9 |
@@ -0,0 +1,28 @@
|
|
|
9805c9 |
+! { dg-do compile }
|
|
|
9805c9 |
+! { dg-options "-fdec -fno-dec-format-defaults" }
|
|
|
9805c9 |
+!
|
|
|
9805c9 |
+! Test case for the default field widths not enabled.
|
|
|
9805c9 |
+!
|
|
|
9805c9 |
+! Test case added by Mark Eggleston <mark.eggleston@codethink.com> to check
|
|
|
9805c9 |
+! use of -fno-dec-format-defaults
|
|
|
9805c9 |
+!
|
|
|
9805c9 |
+
|
|
|
9805c9 |
+ character(50) :: buffer
|
|
|
9805c9 |
+
|
|
|
9805c9 |
+ real*4 :: real_4
|
|
|
9805c9 |
+ real*8 :: real_8
|
|
|
9805c9 |
+ real*16 :: real_16
|
|
|
9805c9 |
+ integer :: len
|
|
|
9805c9 |
+
|
|
|
9805c9 |
+ real_4 = 4.18
|
|
|
9805c9 |
+ write(buffer, '(A, F, A)') ':',real_4,':' ! { dg-error "Nonnegative width required" }
|
|
|
9805c9 |
+
|
|
|
9805c9 |
+ real_4 = 0.00000018
|
|
|
9805c9 |
+ write(buffer, '(A, F, A)') ':',real_4,':' ! { dg-error "Nonnegative width required" }
|
|
|
9805c9 |
+
|
|
|
9805c9 |
+ real_8 = 4.18
|
|
|
9805c9 |
+ write(buffer, '(A, F, A)') ':',real_8,':' ! { dg-error "Nonnegative width required" }
|
|
|
9805c9 |
+
|
|
|
9805c9 |
+ real_16 = 4.18
|
|
|
9805c9 |
+ write(buffer, '(A, F, A)') ':',real_16,':' ! { dg-error "Nonnegative width required" }
|
|
|
9805c9 |
+end
|
|
|
9805c9 |
diff --git a/gcc/testsuite/gfortran.dg/fmt_g_default_field_width_1.f90 b/gcc/testsuite/gfortran.dg/fmt_g_default_field_width_1.f90
|
|
|
9805c9 |
new file mode 100644
|
|
|
9805c9 |
index 00000000000..6e2ad141d4a
|
|
|
9805c9 |
--- /dev/null
|
|
|
9805c9 |
+++ b/gcc/testsuite/gfortran.dg/fmt_g_default_field_width_1.f90
|
|
|
9805c9 |
@@ -0,0 +1,48 @@
|
|
|
9805c9 |
+! { dg-do run }
|
|
|
9805c9 |
+! { dg-options -fdec }
|
|
|
9805c9 |
+!
|
|
|
9805c9 |
+! Test case for the default field widths enabled by the -fdec-format-defaults flag.
|
|
|
9805c9 |
+!
|
|
|
9805c9 |
+! This feature is not part of any Fortran standard, but it is supported by the
|
|
|
9805c9 |
+! Oracle Fortran compiler and others.
|
|
|
9805c9 |
+!
|
|
|
9805c9 |
+! libgfortran uses printf() internally to implement FORMAT. If you print float
|
|
|
9805c9 |
+! values to a higher precision than the type can actually store, the results
|
|
|
9805c9 |
+! are implementation dependent: some platforms print zeros, others print random
|
|
|
9805c9 |
+! numbers. Don't depend on this behaviour in tests because they will not be
|
|
|
9805c9 |
+! portable.
|
|
|
9805c9 |
+
|
|
|
9805c9 |
+ character(50) :: buffer
|
|
|
9805c9 |
+
|
|
|
9805c9 |
+ real*4 :: real_4
|
|
|
9805c9 |
+ real*8 :: real_8
|
|
|
9805c9 |
+ real*16 :: real_16
|
|
|
9805c9 |
+ integer :: len
|
|
|
9805c9 |
+
|
|
|
9805c9 |
+ real_4 = 4.18
|
|
|
9805c9 |
+ write(buffer, '(A, G, A)') ':',real_4,':'
|
|
|
9805c9 |
+ print *,buffer
|
|
|
9805c9 |
+ if (buffer.ne.": 4.180000 :") stop 1
|
|
|
9805c9 |
+
|
|
|
9805c9 |
+ real_4 = 0.00000018
|
|
|
9805c9 |
+ write(buffer, '(A, G, A)') ':',real_4,':'
|
|
|
9805c9 |
+ print *,buffer
|
|
|
9805c9 |
+ if (buffer.ne.": 0.1800000E-06:") stop 2
|
|
|
9805c9 |
+
|
|
|
9805c9 |
+ real_4 = 18000000.4
|
|
|
9805c9 |
+ write(buffer, '(A, G, A)') ':',real_4,':'
|
|
|
9805c9 |
+ print *,buffer
|
|
|
9805c9 |
+ if (buffer.ne.": 0.1800000E+08:") stop 3
|
|
|
9805c9 |
+
|
|
|
9805c9 |
+ real_8 = 4.18
|
|
|
9805c9 |
+ write(buffer, '(A, G, A)') ':',real_8,':'
|
|
|
9805c9 |
+ print *,buffer
|
|
|
9805c9 |
+ len = len_trim(buffer)
|
|
|
9805c9 |
+ if (len /= 27) stop 4
|
|
|
9805c9 |
+
|
|
|
9805c9 |
+ real_16 = 4.18
|
|
|
9805c9 |
+ write(buffer, '(A, G, A)') ':',real_16,':'
|
|
|
9805c9 |
+ print *,buffer
|
|
|
9805c9 |
+ len = len_trim(buffer)
|
|
|
9805c9 |
+ if (len /= 44) stop 5
|
|
|
9805c9 |
+end
|
|
|
9805c9 |
diff --git a/gcc/testsuite/gfortran.dg/fmt_g_default_field_width_2.f90 b/gcc/testsuite/gfortran.dg/fmt_g_default_field_width_2.f90
|
|
|
9805c9 |
new file mode 100644
|
|
|
9805c9 |
index 00000000000..7b218af8610
|
|
|
9805c9 |
--- /dev/null
|
|
|
9805c9 |
+++ b/gcc/testsuite/gfortran.dg/fmt_g_default_field_width_2.f90
|
|
|
9805c9 |
@@ -0,0 +1,52 @@
|
|
|
9805c9 |
+! { dg-do run }
|
|
|
9805c9 |
+! { dg-options -fdec-format-defaults }
|
|
|
9805c9 |
+!
|
|
|
9805c9 |
+! Test case for the default field widths enabled by the -fdec-format-defaults flag.
|
|
|
9805c9 |
+!
|
|
|
9805c9 |
+! This feature is not part of any Fortran standard, but it is supported by the
|
|
|
9805c9 |
+! Oracle Fortran compiler and others.
|
|
|
9805c9 |
+!
|
|
|
9805c9 |
+! libgfortran uses printf() internally to implement FORMAT. If you print float
|
|
|
9805c9 |
+! values to a higher precision than the type can actually store, the results
|
|
|
9805c9 |
+! are implementation dependent: some platforms print zeros, others print random
|
|
|
9805c9 |
+! numbers. Don't depend on this behaviour in tests because they will not be
|
|
|
9805c9 |
+! portable.
|
|
|
9805c9 |
+!
|
|
|
9805c9 |
+! Test case added by Mark Eggleston <mark.eggleston@codethink.com> to check
|
|
|
9805c9 |
+! use of -fdec-format-defaults
|
|
|
9805c9 |
+!
|
|
|
9805c9 |
+
|
|
|
9805c9 |
+ character(50) :: buffer
|
|
|
9805c9 |
+
|
|
|
9805c9 |
+ real*4 :: real_4
|
|
|
9805c9 |
+ real*8 :: real_8
|
|
|
9805c9 |
+ real*16 :: real_16
|
|
|
9805c9 |
+ integer :: len
|
|
|
9805c9 |
+
|
|
|
9805c9 |
+ real_4 = 4.18
|
|
|
9805c9 |
+ write(buffer, '(A, G, A)') ':',real_4,':'
|
|
|
9805c9 |
+ print *,buffer
|
|
|
9805c9 |
+ if (buffer.ne.": 4.180000 :") stop 1
|
|
|
9805c9 |
+
|
|
|
9805c9 |
+ real_4 = 0.00000018
|
|
|
9805c9 |
+ write(buffer, '(A, G, A)') ':',real_4,':'
|
|
|
9805c9 |
+ print *,buffer
|
|
|
9805c9 |
+ if (buffer.ne.": 0.1800000E-06:") stop 2
|
|
|
9805c9 |
+
|
|
|
9805c9 |
+ real_4 = 18000000.4
|
|
|
9805c9 |
+ write(buffer, '(A, G, A)') ':',real_4,':'
|
|
|
9805c9 |
+ print *,buffer
|
|
|
9805c9 |
+ if (buffer.ne.": 0.1800000E+08:") stop 3
|
|
|
9805c9 |
+
|
|
|
9805c9 |
+ real_8 = 4.18
|
|
|
9805c9 |
+ write(buffer, '(A, G, A)') ':',real_8,':'
|
|
|
9805c9 |
+ print *,buffer
|
|
|
9805c9 |
+ len = len_trim(buffer)
|
|
|
9805c9 |
+ if (len /= 27) stop 4
|
|
|
9805c9 |
+
|
|
|
9805c9 |
+ real_16 = 4.18
|
|
|
9805c9 |
+ write(buffer, '(A, G, A)') ':',real_16,':'
|
|
|
9805c9 |
+ print *,buffer
|
|
|
9805c9 |
+ len = len_trim(buffer)
|
|
|
9805c9 |
+ if (len /= 44) stop 5
|
|
|
9805c9 |
+end
|
|
|
9805c9 |
diff --git a/gcc/testsuite/gfortran.dg/fmt_g_default_field_width_3.f90 b/gcc/testsuite/gfortran.dg/fmt_g_default_field_width_3.f90
|
|
|
9805c9 |
new file mode 100644
|
|
|
9805c9 |
index 00000000000..e255c2f94a0
|
|
|
9805c9 |
--- /dev/null
|
|
|
9805c9 |
+++ b/gcc/testsuite/gfortran.dg/fmt_g_default_field_width_3.f90
|
|
|
9805c9 |
@@ -0,0 +1,31 @@
|
|
|
9805c9 |
+! { dg-do compile }
|
|
|
9805c9 |
+! { dg-options "-fdec -fno-dec-format-defaults" }
|
|
|
9805c9 |
+!
|
|
|
9805c9 |
+! Test case for the default field widths not enabled.
|
|
|
9805c9 |
+!
|
|
|
9805c9 |
+! Test case added by Mark Eggleston <mark.eggleston@codethink.com> to check
|
|
|
9805c9 |
+! use of -fno-dec-format-defaults
|
|
|
9805c9 |
+!
|
|
|
9805c9 |
+
|
|
|
9805c9 |
+ character(50) :: buffer
|
|
|
9805c9 |
+
|
|
|
9805c9 |
+ real*4 :: real_4
|
|
|
9805c9 |
+ real*8 :: real_8
|
|
|
9805c9 |
+ real*16 :: real_16
|
|
|
9805c9 |
+ integer :: len
|
|
|
9805c9 |
+
|
|
|
9805c9 |
+ real_4 = 4.18
|
|
|
9805c9 |
+ write(buffer, '(A, G, A)') ':',real_4,':' ! { dg-error "Positive width required" }
|
|
|
9805c9 |
+
|
|
|
9805c9 |
+ real_4 = 0.00000018
|
|
|
9805c9 |
+ write(buffer, '(A, G, A)') ':',real_4,':' ! { dg-error "Positive width required" }
|
|
|
9805c9 |
+
|
|
|
9805c9 |
+ real_4 = 18000000.4
|
|
|
9805c9 |
+ write(buffer, '(A, G, A)') ':',real_4,':' ! { dg-error "Positive width required" }
|
|
|
9805c9 |
+
|
|
|
9805c9 |
+ real_8 = 4.18
|
|
|
9805c9 |
+ write(buffer, '(A, G, A)') ':',real_8,':' ! { dg-error "Positive width required" }
|
|
|
9805c9 |
+
|
|
|
9805c9 |
+ real_16 = 4.18
|
|
|
9805c9 |
+ write(buffer, '(A, G, A)') ':',real_16,':' ! { dg-error "Positive width required" }
|
|
|
9805c9 |
+end
|
|
|
9805c9 |
diff --git a/gcc/testsuite/gfortran.dg/fmt_i_default_field_width_1.f90 b/gcc/testsuite/gfortran.dg/fmt_i_default_field_width_1.f90
|
|
|
9805c9 |
new file mode 100644
|
|
|
9805c9 |
index 00000000000..0d32d240394
|
|
|
9805c9 |
--- /dev/null
|
|
|
9805c9 |
+++ b/gcc/testsuite/gfortran.dg/fmt_i_default_field_width_1.f90
|
|
|
9805c9 |
@@ -0,0 +1,38 @@
|
|
|
9805c9 |
+! { dg-do run }
|
|
|
9805c9 |
+! { dg-options -fdec }
|
|
|
9805c9 |
+!
|
|
|
9805c9 |
+! Test case for the default field widths enabled by the -fdec-format-defaults flag.
|
|
|
9805c9 |
+!
|
|
|
9805c9 |
+! This feature is not part of any Fortran standard, but it is supported by the
|
|
|
9805c9 |
+! Oracle Fortran compiler and others.
|
|
|
9805c9 |
+
|
|
|
9805c9 |
+ character(50) :: buffer
|
|
|
9805c9 |
+ character(1) :: colon
|
|
|
9805c9 |
+
|
|
|
9805c9 |
+ integer*2 :: integer_2
|
|
|
9805c9 |
+ integer*4 :: integer_4
|
|
|
9805c9 |
+ integer*8 :: integer_8
|
|
|
9805c9 |
+
|
|
|
9805c9 |
+ write(buffer, '(A, I, A)') ':',12340,':'
|
|
|
9805c9 |
+ print *,buffer
|
|
|
9805c9 |
+ if (buffer.ne.": 12340:") stop 1
|
|
|
9805c9 |
+
|
|
|
9805c9 |
+ read(buffer, '(A1, I, A1)') colon, integer_4, colon
|
|
|
9805c9 |
+ if (integer_4.ne.12340) stop 2
|
|
|
9805c9 |
+
|
|
|
9805c9 |
+ integer_2 = -99
|
|
|
9805c9 |
+ write(buffer, '(A, I, A)') ':',integer_2,':'
|
|
|
9805c9 |
+ print *,buffer
|
|
|
9805c9 |
+ if (buffer.ne.": -99:") stop 3
|
|
|
9805c9 |
+
|
|
|
9805c9 |
+ integer_8 = -11112222
|
|
|
9805c9 |
+ write(buffer, '(A, I, A)') ':',integer_8,':'
|
|
|
9805c9 |
+ print *,buffer
|
|
|
9805c9 |
+ if (buffer.ne.": -11112222:") stop 4
|
|
|
9805c9 |
+
|
|
|
9805c9 |
+! If the width is 7 and there are 7 leading zeroes, the result should be zero.
|
|
|
9805c9 |
+ integer_2 = 789
|
|
|
9805c9 |
+ buffer = '0000000789'
|
|
|
9805c9 |
+ read(buffer, '(I)') integer_2
|
|
|
9805c9 |
+ if (integer_2.ne.0) stop 5
|
|
|
9805c9 |
+end
|
|
|
9805c9 |
diff --git a/gcc/testsuite/gfortran.dg/fmt_i_default_field_width_2.f90 b/gcc/testsuite/gfortran.dg/fmt_i_default_field_width_2.f90
|
|
|
9805c9 |
new file mode 100644
|
|
|
9805c9 |
index 00000000000..6cee3f86809
|
|
|
9805c9 |
--- /dev/null
|
|
|
9805c9 |
+++ b/gcc/testsuite/gfortran.dg/fmt_i_default_field_width_2.f90
|
|
|
9805c9 |
@@ -0,0 +1,42 @@
|
|
|
9805c9 |
+! { dg-do run }
|
|
|
9805c9 |
+! { dg-options -fdec-format-defaults }
|
|
|
9805c9 |
+!
|
|
|
9805c9 |
+! Test case for the default field widths enabled by the -fdec-format-defaults flag.
|
|
|
9805c9 |
+!
|
|
|
9805c9 |
+! This feature is not part of any Fortran standard, but it is supported by the
|
|
|
9805c9 |
+! Oracle Fortran compiler and others.
|
|
|
9805c9 |
+!
|
|
|
9805c9 |
+! Test case added by Mark Eggleston <mark.eggleston@codethink.com> to check
|
|
|
9805c9 |
+! use of -fdec-format-defaults
|
|
|
9805c9 |
+!
|
|
|
9805c9 |
+
|
|
|
9805c9 |
+ character(50) :: buffer
|
|
|
9805c9 |
+ character(1) :: colon
|
|
|
9805c9 |
+
|
|
|
9805c9 |
+ integer*2 :: integer_2
|
|
|
9805c9 |
+ integer*4 :: integer_4
|
|
|
9805c9 |
+ integer*8 :: integer_8
|
|
|
9805c9 |
+
|
|
|
9805c9 |
+ write(buffer, '(A, I, A)') ':',12340,':'
|
|
|
9805c9 |
+ print *,buffer
|
|
|
9805c9 |
+ if (buffer.ne.": 12340:") stop 1
|
|
|
9805c9 |
+
|
|
|
9805c9 |
+ read(buffer, '(A1, I, A1)') colon, integer_4, colon
|
|
|
9805c9 |
+ if (integer_4.ne.12340) stop 2
|
|
|
9805c9 |
+
|
|
|
9805c9 |
+ integer_2 = -99
|
|
|
9805c9 |
+ write(buffer, '(A, I, A)') ':',integer_2,':'
|
|
|
9805c9 |
+ print *,buffer
|
|
|
9805c9 |
+ if (buffer.ne.": -99:") stop 3
|
|
|
9805c9 |
+
|
|
|
9805c9 |
+ integer_8 = -11112222
|
|
|
9805c9 |
+ write(buffer, '(A, I, A)') ':',integer_8,':'
|
|
|
9805c9 |
+ print *,buffer
|
|
|
9805c9 |
+ if (buffer.ne.": -11112222:") stop 4
|
|
|
9805c9 |
+
|
|
|
9805c9 |
+! If the width is 7 and there are 7 leading zeroes, the result should be zero.
|
|
|
9805c9 |
+ integer_2 = 789
|
|
|
9805c9 |
+ buffer = '0000000789'
|
|
|
9805c9 |
+ read(buffer, '(I)') integer_2
|
|
|
9805c9 |
+ if (integer_2.ne.0) stop 5
|
|
|
9805c9 |
+end
|
|
|
9805c9 |
diff --git a/gcc/testsuite/gfortran.dg/fmt_i_default_field_width_3.f90 b/gcc/testsuite/gfortran.dg/fmt_i_default_field_width_3.f90
|
|
|
9805c9 |
new file mode 100644
|
|
|
9805c9 |
index 00000000000..3a6684b3c4d
|
|
|
9805c9 |
--- /dev/null
|
|
|
9805c9 |
+++ b/gcc/testsuite/gfortran.dg/fmt_i_default_field_width_3.f90
|
|
|
9805c9 |
@@ -0,0 +1,35 @@
|
|
|
9805c9 |
+! { dg-do compile }
|
|
|
9805c9 |
+! { dg-options "-fdec -fno-dec-format-defaults" }
|
|
|
9805c9 |
+!
|
|
|
9805c9 |
+! Test case for the default field widths enabled by the -fdec-format-defaults flag.
|
|
|
9805c9 |
+!
|
|
|
9805c9 |
+! This feature is not part of any Fortran standard, but it is supported by the
|
|
|
9805c9 |
+! Oracle Fortran compiler and others.
|
|
|
9805c9 |
+!
|
|
|
9805c9 |
+! Test case added by Mark Eggleston <mark.eggleston@codethink.com> to check
|
|
|
9805c9 |
+! use of -fdec-format-defaults
|
|
|
9805c9 |
+!
|
|
|
9805c9 |
+
|
|
|
9805c9 |
+ character(50) :: buffer
|
|
|
9805c9 |
+ character(1) :: colon
|
|
|
9805c9 |
+
|
|
|
9805c9 |
+ integer*2 :: integer_2
|
|
|
9805c9 |
+ integer*4 :: integer_4
|
|
|
9805c9 |
+ integer*8 :: integer_8
|
|
|
9805c9 |
+
|
|
|
9805c9 |
+ write(buffer, '(A, I, A)') ':',12340,':' ! { dg-error "Nonnegative width required" }
|
|
|
9805c9 |
+
|
|
|
9805c9 |
+ read(buffer, '(A1, I, A1)') colon, integer_4, colon ! { dg-error "Nonnegative width required" }
|
|
|
9805c9 |
+ if (integer_4.ne.12340) stop 2
|
|
|
9805c9 |
+
|
|
|
9805c9 |
+ integer_2 = -99
|
|
|
9805c9 |
+ write(buffer, '(A, I, A)') ':',integer_2,':' ! { dg-error "Nonnegative width required" }
|
|
|
9805c9 |
+
|
|
|
9805c9 |
+ integer_8 = -11112222
|
|
|
9805c9 |
+ write(buffer, '(A, I, A)') ':',integer_8,':' ! { dg-error "Nonnegative width required" }
|
|
|
9805c9 |
+
|
|
|
9805c9 |
+! If the width is 7 and there are 7 leading zeroes, the result should be zero.
|
|
|
9805c9 |
+ integer_2 = 789
|
|
|
9805c9 |
+ buffer = '0000000789'
|
|
|
9805c9 |
+ read(buffer, '(I)') integer_2 ! { dg-error "Nonnegative width required" }
|
|
|
9805c9 |
+end
|
|
|
9805c9 |
diff --git a/libgfortran/io/format.c b/libgfortran/io/format.c
|
|
|
9805c9 |
index 688764785da..e798d9bda87 100644
|
|
|
9805c9 |
--- a/libgfortran/io/format.c
|
|
|
9805c9 |
+++ b/libgfortran/io/format.c
|
|
|
9805c9 |
@@ -956,12 +956,33 @@ parse_format_list (st_parameter_dt *dtp, bool *seen_dd)
|
|
|
9805c9 |
*seen_dd = true;
|
|
|
9805c9 |
if (u != FMT_POSINT && u != FMT_ZERO)
|
|
|
9805c9 |
{
|
|
|
9805c9 |
+ if (dtp->common.flags & IOPARM_DT_DEC_EXT)
|
|
|
9805c9 |
+ {
|
|
|
9805c9 |
+ tail->u.real.w = DEFAULT_WIDTH;
|
|
|
9805c9 |
+ tail->u.real.d = 0;
|
|
|
9805c9 |
+ tail->u.real.e = -1;
|
|
|
9805c9 |
+ fmt->saved_token = u;
|
|
|
9805c9 |
+ break;
|
|
|
9805c9 |
+ }
|
|
|
9805c9 |
fmt->error = nonneg_required;
|
|
|
9805c9 |
goto finished;
|
|
|
9805c9 |
}
|
|
|
9805c9 |
}
|
|
|
9805c9 |
+ else if (u == FMT_ZERO)
|
|
|
9805c9 |
+ {
|
|
|
9805c9 |
+ fmt->error = posint_required;
|
|
|
9805c9 |
+ goto finished;
|
|
|
9805c9 |
+ }
|
|
|
9805c9 |
else if (u != FMT_POSINT)
|
|
|
9805c9 |
{
|
|
|
9805c9 |
+ if (dtp->common.flags & IOPARM_DT_DEC_EXT)
|
|
|
9805c9 |
+ {
|
|
|
9805c9 |
+ tail->u.real.w = DEFAULT_WIDTH;
|
|
|
9805c9 |
+ tail->u.real.d = 0;
|
|
|
9805c9 |
+ tail->u.real.e = -1;
|
|
|
9805c9 |
+ fmt->saved_token = u;
|
|
|
9805c9 |
+ break;
|
|
|
9805c9 |
+ }
|
|
|
9805c9 |
fmt->error = posint_required;
|
|
|
9805c9 |
goto finished;
|
|
|
9805c9 |
}
|
|
|
9805c9 |
@@ -1100,6 +1121,13 @@ parse_format_list (st_parameter_dt *dtp, bool *seen_dd)
|
|
|
9805c9 |
{
|
|
|
9805c9 |
if (t != FMT_POSINT)
|
|
|
9805c9 |
{
|
|
|
9805c9 |
+ if (dtp->common.flags & IOPARM_DT_DEC_EXT)
|
|
|
9805c9 |
+ {
|
|
|
9805c9 |
+ tail->u.integer.w = DEFAULT_WIDTH;
|
|
|
9805c9 |
+ tail->u.integer.m = -1;
|
|
|
9805c9 |
+ fmt->saved_token = t;
|
|
|
9805c9 |
+ break;
|
|
|
9805c9 |
+ }
|
|
|
9805c9 |
fmt->error = posint_required;
|
|
|
9805c9 |
goto finished;
|
|
|
9805c9 |
}
|
|
|
9805c9 |
@@ -1108,6 +1136,13 @@ parse_format_list (st_parameter_dt *dtp, bool *seen_dd)
|
|
|
9805c9 |
{
|
|
|
9805c9 |
if (t != FMT_ZERO && t != FMT_POSINT)
|
|
|
9805c9 |
{
|
|
|
9805c9 |
+ if (dtp->common.flags & IOPARM_DT_DEC_EXT)
|
|
|
9805c9 |
+ {
|
|
|
9805c9 |
+ tail->u.integer.w = DEFAULT_WIDTH;
|
|
|
9805c9 |
+ tail->u.integer.m = -1;
|
|
|
9805c9 |
+ fmt->saved_token = t;
|
|
|
9805c9 |
+ break;
|
|
|
9805c9 |
+ }
|
|
|
9805c9 |
fmt->error = nonneg_required;
|
|
|
9805c9 |
goto finished;
|
|
|
9805c9 |
}
|
|
|
9805c9 |
diff --git a/libgfortran/io/io.h b/libgfortran/io/io.h
|
|
|
9805c9 |
index 5caaea280f0..f5e63797ba1 100644
|
|
|
9805c9 |
--- a/libgfortran/io/io.h
|
|
|
9805c9 |
+++ b/libgfortran/io/io.h
|
|
|
9805c9 |
@@ -1011,6 +1011,56 @@ memset4 (gfc_char4_t *p, gfc_char4_t c, int k)
|
|
|
9805c9 |
*p++ = c;
|
|
|
9805c9 |
}
|
|
|
9805c9 |
|
|
|
9805c9 |
+/* Used in width fields to indicate that the default should be used */
|
|
|
9805c9 |
+#define DEFAULT_WIDTH -1
|
|
|
9805c9 |
+
|
|
|
9805c9 |
+/* Defaults for certain format field descriptors. These are decided based on
|
|
|
9805c9 |
+ * the type of the value being formatted.
|
|
|
9805c9 |
+ *
|
|
|
9805c9 |
+ * The behaviour here is modelled on the Oracle Fortran compiler. At the time
|
|
|
9805c9 |
+ * of writing, the details were available at this URL:
|
|
|
9805c9 |
+ *
|
|
|
9805c9 |
+ * https://docs.oracle.com/cd/E19957-01/805-4939/6j4m0vnc3/index.html#z4000743746d
|
|
|
9805c9 |
+ */
|
|
|
9805c9 |
+
|
|
|
9805c9 |
+static inline int
|
|
|
9805c9 |
+default_width_for_integer (int kind)
|
|
|
9805c9 |
+{
|
|
|
9805c9 |
+ switch (kind)
|
|
|
9805c9 |
+ {
|
|
|
9805c9 |
+ case 1:
|
|
|
9805c9 |
+ case 2: return 7;
|
|
|
9805c9 |
+ case 4: return 12;
|
|
|
9805c9 |
+ case 8: return 23;
|
|
|
9805c9 |
+ case 16: return 44;
|
|
|
9805c9 |
+ default: return 0;
|
|
|
9805c9 |
+ }
|
|
|
9805c9 |
+}
|
|
|
9805c9 |
+
|
|
|
9805c9 |
+static inline int
|
|
|
9805c9 |
+default_width_for_float (int kind)
|
|
|
9805c9 |
+{
|
|
|
9805c9 |
+ switch (kind)
|
|
|
9805c9 |
+ {
|
|
|
9805c9 |
+ case 4: return 15;
|
|
|
9805c9 |
+ case 8: return 25;
|
|
|
9805c9 |
+ case 16: return 42;
|
|
|
9805c9 |
+ default: return 0;
|
|
|
9805c9 |
+ }
|
|
|
9805c9 |
+}
|
|
|
9805c9 |
+
|
|
|
9805c9 |
+static inline int
|
|
|
9805c9 |
+default_precision_for_float (int kind)
|
|
|
9805c9 |
+{
|
|
|
9805c9 |
+ switch (kind)
|
|
|
9805c9 |
+ {
|
|
|
9805c9 |
+ case 4: return 7;
|
|
|
9805c9 |
+ case 8: return 16;
|
|
|
9805c9 |
+ case 16: return 33;
|
|
|
9805c9 |
+ default: return 0;
|
|
|
9805c9 |
+ }
|
|
|
9805c9 |
+}
|
|
|
9805c9 |
+
|
|
|
9805c9 |
#endif
|
|
|
9805c9 |
|
|
|
9805c9 |
extern void
|
|
|
9805c9 |
diff --git a/libgfortran/io/read.c b/libgfortran/io/read.c
|
|
|
9805c9 |
index 52ffb4639ac..be9f6cb6f76 100644
|
|
|
9805c9 |
--- a/libgfortran/io/read.c
|
|
|
9805c9 |
+++ b/libgfortran/io/read.c
|
|
|
9805c9 |
@@ -635,6 +635,12 @@ read_decimal (st_parameter_dt *dtp, const fnode *f, char *dest, int length)
|
|
|
9805c9 |
|
|
|
9805c9 |
w = f->u.w;
|
|
|
9805c9 |
|
|
|
9805c9 |
+ /* This is a legacy extension, and the frontend will only allow such cases
|
|
|
9805c9 |
+ * through when -fdec-format-defaults is passed.
|
|
|
9805c9 |
+ */
|
|
|
9805c9 |
+ if (w == DEFAULT_WIDTH)
|
|
|
9805c9 |
+ w = default_width_for_integer (length);
|
|
|
9805c9 |
+
|
|
|
9805c9 |
p = read_block_form (dtp, &w);
|
|
|
9805c9 |
|
|
|
9805c9 |
if (p == NULL)
|
|
|
9805c9 |
diff --git a/libgfortran/io/write.c b/libgfortran/io/write.c
|
|
|
9805c9 |
index c8811e200e0..4ef35561fdd 100644
|
|
|
9805c9 |
--- a/libgfortran/io/write.c
|
|
|
9805c9 |
+++ b/libgfortran/io/write.c
|
|
|
9805c9 |
@@ -685,9 +685,8 @@ write_l (st_parameter_dt *dtp, const fnode *f, char *source, int len)
|
|
|
9805c9 |
p[wlen - 1] = (n) ? 'T' : 'F';
|
|
|
9805c9 |
}
|
|
|
9805c9 |
|
|
|
9805c9 |
-
|
|
|
9805c9 |
static void
|
|
|
9805c9 |
-write_boz (st_parameter_dt *dtp, const fnode *f, const char *q, int n)
|
|
|
9805c9 |
+write_boz (st_parameter_dt *dtp, const fnode *f, const char *q, int n, int len)
|
|
|
9805c9 |
{
|
|
|
9805c9 |
int w, m, digits, nzero, nblank;
|
|
|
9805c9 |
char *p;
|
|
|
9805c9 |
@@ -720,6 +719,9 @@ write_boz (st_parameter_dt *dtp, const fnode *f, const char *q, int n)
|
|
|
9805c9 |
/* Select a width if none was specified. The idea here is to always
|
|
|
9805c9 |
print something. */
|
|
|
9805c9 |
|
|
|
9805c9 |
+ if (w == DEFAULT_WIDTH)
|
|
|
9805c9 |
+ w = default_width_for_integer (len);
|
|
|
9805c9 |
+
|
|
|
9805c9 |
if (w == 0)
|
|
|
9805c9 |
w = ((digits < m) ? m : digits);
|
|
|
9805c9 |
|
|
|
9805c9 |
@@ -846,6 +848,8 @@ write_decimal (st_parameter_dt *dtp, const fnode *f, const char *source,
|
|
|
9805c9 |
|
|
|
9805c9 |
/* Select a width if none was specified. The idea here is to always
|
|
|
9805c9 |
print something. */
|
|
|
9805c9 |
+ if (w == DEFAULT_WIDTH)
|
|
|
9805c9 |
+ w = default_width_for_integer (len);
|
|
|
9805c9 |
|
|
|
9805c9 |
if (w == 0)
|
|
|
9805c9 |
w = ((digits < m) ? m : digits) + nsign;
|
|
|
9805c9 |
@@ -1206,13 +1210,13 @@ write_b (st_parameter_dt *dtp, const fnode *f, const char *source, int len)
|
|
|
9805c9 |
if (len > (int) sizeof (GFC_UINTEGER_LARGEST))
|
|
|
9805c9 |
{
|
|
|
9805c9 |
p = btoa_big (source, itoa_buf, len, &n);
|
|
|
9805c9 |
- write_boz (dtp, f, p, n);
|
|
|
9805c9 |
+ write_boz (dtp, f, p, n, len);
|
|
|
9805c9 |
}
|
|
|
9805c9 |
else
|
|
|
9805c9 |
{
|
|
|
9805c9 |
n = extract_uint (source, len);
|
|
|
9805c9 |
p = btoa (n, itoa_buf, sizeof (itoa_buf));
|
|
|
9805c9 |
- write_boz (dtp, f, p, n);
|
|
|
9805c9 |
+ write_boz (dtp, f, p, n, len);
|
|
|
9805c9 |
}
|
|
|
9805c9 |
}
|
|
|
9805c9 |
|
|
|
9805c9 |
@@ -1227,13 +1231,13 @@ write_o (st_parameter_dt *dtp, const fnode *f, const char *source, int len)
|
|
|
9805c9 |
if (len > (int) sizeof (GFC_UINTEGER_LARGEST))
|
|
|
9805c9 |
{
|
|
|
9805c9 |
p = otoa_big (source, itoa_buf, len, &n);
|
|
|
9805c9 |
- write_boz (dtp, f, p, n);
|
|
|
9805c9 |
+ write_boz (dtp, f, p, n, len);
|
|
|
9805c9 |
}
|
|
|
9805c9 |
else
|
|
|
9805c9 |
{
|
|
|
9805c9 |
n = extract_uint (source, len);
|
|
|
9805c9 |
p = otoa (n, itoa_buf, sizeof (itoa_buf));
|
|
|
9805c9 |
- write_boz (dtp, f, p, n);
|
|
|
9805c9 |
+ write_boz (dtp, f, p, n, len);
|
|
|
9805c9 |
}
|
|
|
9805c9 |
}
|
|
|
9805c9 |
|
|
|
9805c9 |
@@ -1247,13 +1251,13 @@ write_z (st_parameter_dt *dtp, const fnode *f, const char *source, int len)
|
|
|
9805c9 |
if (len > (int) sizeof (GFC_UINTEGER_LARGEST))
|
|
|
9805c9 |
{
|
|
|
9805c9 |
p = ztoa_big (source, itoa_buf, len, &n);
|
|
|
9805c9 |
- write_boz (dtp, f, p, n);
|
|
|
9805c9 |
+ write_boz (dtp, f, p, n, len);
|
|
|
9805c9 |
}
|
|
|
9805c9 |
else
|
|
|
9805c9 |
{
|
|
|
9805c9 |
n = extract_uint (source, len);
|
|
|
9805c9 |
p = gfc_xtoa (n, itoa_buf, sizeof (itoa_buf));
|
|
|
9805c9 |
- write_boz (dtp, f, p, n);
|
|
|
9805c9 |
+ write_boz (dtp, f, p, n, len);
|
|
|
9805c9 |
}
|
|
|
9805c9 |
}
|
|
|
9805c9 |
|
|
|
9805c9 |
@@ -1491,7 +1495,7 @@ size_from_kind (st_parameter_dt *dtp, const fnode *f, int kind)
|
|
|
9805c9 |
{
|
|
|
9805c9 |
int size;
|
|
|
9805c9 |
|
|
|
9805c9 |
- if (f->format == FMT_F && f->u.real.w == 0)
|
|
|
9805c9 |
+ if ((f->format == FMT_F && f->u.real.w == 0) || f->u.real.w == DEFAULT_WIDTH)
|
|
|
9805c9 |
{
|
|
|
9805c9 |
switch (kind)
|
|
|
9805c9 |
{
|
|
|
9805c9 |
diff --git a/libgfortran/io/write_float.def b/libgfortran/io/write_float.def
|
|
|
9805c9 |
index c63db4e77ef..daa16679f53 100644
|
|
|
9805c9 |
--- a/libgfortran/io/write_float.def
|
|
|
9805c9 |
+++ b/libgfortran/io/write_float.def
|
|
|
9805c9 |
@@ -113,7 +113,8 @@ determine_precision (st_parameter_dt * dtp, const fnode * f, int len)
|
|
|
9805c9 |
static void
|
|
|
9805c9 |
build_float_string (st_parameter_dt *dtp, const fnode *f, char *buffer,
|
|
|
9805c9 |
size_t size, int nprinted, int precision, int sign_bit,
|
|
|
9805c9 |
- bool zero_flag, int npad, char *result, size_t *len)
|
|
|
9805c9 |
+ bool zero_flag, int npad, int default_width, char *result,
|
|
|
9805c9 |
+ size_t *len)
|
|
|
9805c9 |
{
|
|
|
9805c9 |
char *put;
|
|
|
9805c9 |
char *digits;
|
|
|
9805c9 |
@@ -132,8 +133,17 @@ build_float_string (st_parameter_dt *dtp, const fnode *f, char *buffer,
|
|
|
9805c9 |
sign_t sign;
|
|
|
9805c9 |
|
|
|
9805c9 |
ft = f->format;
|
|
|
9805c9 |
- w = f->u.real.w;
|
|
|
9805c9 |
- d = f->u.real.d;
|
|
|
9805c9 |
+ if (f->u.real.w == DEFAULT_WIDTH)
|
|
|
9805c9 |
+ /* This codepath can only be reached with -fdec-format-defaults. */
|
|
|
9805c9 |
+ {
|
|
|
9805c9 |
+ w = default_width;
|
|
|
9805c9 |
+ d = precision;
|
|
|
9805c9 |
+ }
|
|
|
9805c9 |
+ else
|
|
|
9805c9 |
+ {
|
|
|
9805c9 |
+ w = f->u.real.w;
|
|
|
9805c9 |
+ d = f->u.real.d;
|
|
|
9805c9 |
+ }
|
|
|
9805c9 |
p = dtp->u.p.scale_factor;
|
|
|
9805c9 |
*len = 0;
|
|
|
9805c9 |
|
|
|
9805c9 |
@@ -960,6 +970,11 @@ determine_en_precision (st_parameter_dt *dtp, const fnode *f,
|
|
|
9805c9 |
int save_scale_factor;\
|
|
|
9805c9 |
volatile GFC_REAL_ ## x temp;\
|
|
|
9805c9 |
save_scale_factor = dtp->u.p.scale_factor;\
|
|
|
9805c9 |
+ if (w == DEFAULT_WIDTH)\
|
|
|
9805c9 |
+ {\
|
|
|
9805c9 |
+ w = default_width;\
|
|
|
9805c9 |
+ d = precision;\
|
|
|
9805c9 |
+ }\
|
|
|
9805c9 |
switch (dtp->u.p.current_unit->round_status)\
|
|
|
9805c9 |
{\
|
|
|
9805c9 |
case ROUND_ZERO:\
|
|
|
9805c9 |
@@ -1035,7 +1050,8 @@ determine_en_precision (st_parameter_dt *dtp, const fnode *f,
|
|
|
9805c9 |
nprinted = FDTOA(y,precision,m);\
|
|
|
9805c9 |
}\
|
|
|
9805c9 |
build_float_string (dtp, &newf, buffer, size, nprinted, precision,\
|
|
|
9805c9 |
- sign_bit, zero_flag, npad, result, res_len);\
|
|
|
9805c9 |
+ sign_bit, zero_flag, npad, default_width,\
|
|
|
9805c9 |
+ result, res_len);\
|
|
|
9805c9 |
dtp->u.p.scale_factor = save_scale_factor;\
|
|
|
9805c9 |
}\
|
|
|
9805c9 |
else\
|
|
|
9805c9 |
@@ -1045,7 +1061,8 @@ determine_en_precision (st_parameter_dt *dtp, const fnode *f,
|
|
|
9805c9 |
else\
|
|
|
9805c9 |
nprinted = DTOA(y,precision,m);\
|
|
|
9805c9 |
build_float_string (dtp, f, buffer, size, nprinted, precision,\
|
|
|
9805c9 |
- sign_bit, zero_flag, npad, result, res_len);\
|
|
|
9805c9 |
+ sign_bit, zero_flag, npad, default_width,\
|
|
|
9805c9 |
+ result, res_len);\
|
|
|
9805c9 |
}\
|
|
|
9805c9 |
}\
|
|
|
9805c9 |
|
|
|
9805c9 |
@@ -1059,6 +1076,16 @@ get_float_string (st_parameter_dt *dtp, const fnode *f, const char *source,
|
|
|
9805c9 |
{
|
|
|
9805c9 |
int sign_bit, nprinted;
|
|
|
9805c9 |
bool zero_flag;
|
|
|
9805c9 |
+ int default_width = 0;
|
|
|
9805c9 |
+
|
|
|
9805c9 |
+ if (f->u.real.w == DEFAULT_WIDTH)
|
|
|
9805c9 |
+ /* This codepath can only be reached with -fdec-format-defaults. The default
|
|
|
9805c9 |
+ * values are based on those used in the Oracle Fortran compiler.
|
|
|
9805c9 |
+ */
|
|
|
9805c9 |
+ {
|
|
|
9805c9 |
+ default_width = default_width_for_float (kind);
|
|
|
9805c9 |
+ precision = default_precision_for_float (kind);
|
|
|
9805c9 |
+ }
|
|
|
9805c9 |
|
|
|
9805c9 |
switch (kind)
|
|
|
9805c9 |
{
|
|
|
9805c9 |
--
|
|
|
9805c9 |
2.11.0
|
|
|
9805c9 |
|