[Bug fortran/49110] Deferred-length character result triggers (false positive) error for pure procedures
http://gcc.gnu.org/bugzilla/show_bug.cgi?id=49110 Tobias Burnus burnus at gcc dot gnu.org changed: What|Removed |Added Status|UNCONFIRMED |RESOLVED Resolution||FIXED --- Comment #17 from Tobias Burnus burnus at gcc dot gnu.org 2012-05-14 16:46:20 UTC --- The REPEAT issue has now also been FIXED on the trunk (4.8). Thanks for the patience.
[Bug fortran/49110] Deferred-length character result triggers (false positive) error for pure procedures
http://gcc.gnu.org/bugzilla/show_bug.cgi?id=49110 --- Comment #16 from Tobias Burnus burnus at gcc dot gnu.org 2012-05-14 16:45:31 UTC --- Author: burnus Date: Mon May 14 16:45:16 2012 New Revision: 187472 URL: http://gcc.gnu.org/viewcvs?root=gccview=revrev=187472 Log: 2012-05-14 Tobias Burnus bur...@net-b.de PR fortran/49110 PR fortran/51055 PR fortran/53329 * trans-expr.c (gfc_trans_assignment_1): Fix allocation handling for assignment of function results to allocatable deferred-length strings. * trans-decl.c (gfc_create_string_length): For deferred-length module variables, include module name in the assembler name. (gfc_get_symbol_decl): Don't override the assembler name. 2012-05-14 Tobias Burnus bur...@net-b.de PR fortran/49110 PR fortran/51055 PR fortran/53329 * gfortran.dg/deferred_type_param_4.f90: New. * gfortran.dg/deferred_type_param_6.f90: New. Added: trunk/gcc/testsuite/gfortran.dg/deferred_type_param_4.f90 trunk/gcc/testsuite/gfortran.dg/deferred_type_param_6.f90 Modified: trunk/gcc/fortran/ChangeLog trunk/gcc/fortran/trans-decl.c trunk/gcc/fortran/trans-expr.c trunk/gcc/testsuite/ChangeLog
[Bug fortran/49110] Deferred-length character result triggers (false positive) error for pure procedures
http://gcc.gnu.org/bugzilla/show_bug.cgi?id=49110 --- Comment #14 from Tobias Burnus burnus at gcc dot gnu.org 2012-05-12 09:54:00 UTC --- Author: burnus Date: Sat May 12 09:53:53 2012 New Revision: 187427 URL: http://gcc.gnu.org/viewcvs?root=gccview=revrev=187427 Log: 2012-05-12 Tobias Burnus bur...@net-b.de PR fortran/49110 PR fortran/52843 * resolve.c (resolve_fl_procedure): Don't regard character(len=:) as character(*) in the diagnostic. 2012-05-12 Tobias Burnus bur...@net-b.de PR fortran/49110 PR fortran/52843 * gfortran.dg/deferred_type_param_5.f90: New. Added: trunk/gcc/testsuite/gfortran.dg/deferred_type_param_5.f90 Modified: trunk/gcc/fortran/ChangeLog trunk/gcc/fortran/resolve.c trunk/gcc/testsuite/ChangeLog
[Bug fortran/49110] Deferred-length character result triggers (false positive) error for pure procedures
http://gcc.gnu.org/bugzilla/show_bug.cgi?id=49110 --- Comment #15 from Tobias Burnus burnus at gcc dot gnu.org 2012-05-12 09:59:10 UTC --- The commit of comment 13 fixes the issue with the bogus PURE diagnostic. Sorry that it took one year to get it fixed - and thanks for the bug report. * * * Regarding the REPEAT issue (cf. comment 9 and PR 51055): A patch solving that issue has been posted at http://gcc.gnu.org/ml/fortran/2012-05/msg00054.html
[Bug fortran/49110] Deferred-length character result triggers (false positive) error for pure procedures
http://gcc.gnu.org/bugzilla/show_bug.cgi?id=49110
Tobias Burnus burnus at gcc dot gnu.org changed:
What|Removed |Added
CC||burnus at gcc dot gnu.org
--- Comment #13 from Tobias Burnus burnus at gcc dot gnu.org 2012-01-02
18:53:47 UTC ---
(In reply to comment #9)
! character, intent(in) :: s(5) ! Compiles and runs
character, intent(in) :: s(:)! Compiles and segfaults
character(len=:), allocatable :: a
a = repeat('a', size(s))
For that example, try:
a = ( repeat('a', size(s)) )
Cf. PR 51055.
[Bug fortran/49110] Deferred-length character result triggers (false positive) error for pure procedures
http://gcc.gnu.org/bugzilla/show_bug.cgi?id=49110 --- Comment #1 from John jwmwalrus at gmail dot com 2011-05-22 16:13:49 UTC --- Created attachment 24327 -- http://gcc.gnu.org/bugzilla/attachment.cgi?id=24327 Test case
[Bug fortran/49110] Deferred-length character result triggers (false positive) error for pure procedures
http://gcc.gnu.org/bugzilla/show_bug.cgi?id=49110
kargl at gcc dot gnu.org changed:
What|Removed |Added
CC||kargl at gcc dot gnu.org
--- Comment #2 from kargl at gcc dot gnu.org 2011-05-22 16:32:51 UTC ---
This patch allows your code to compile, but I
don't know if it works correctly. Do you have
a complete self-contained small testcase?
Index: resolve.c
===
--- resolve.c (revision 173757)
+++ resolve.c (working copy)
@@ -10305,7 +10305,7 @@ resolve_fl_procedure (gfc_symbol *sym, i
actual length; (ii) To declare a named constant; or (iii) External
function - but length must be declared in calling scoping unit. */
if (sym-attr.function
- sym-ts.type == BT_CHARACTER
+ sym-ts.type == BT_CHARACTER !sym-ts.deferred
sym-ts.u.cl sym-ts.u.cl-length == NULL)
{
if ((sym-as sym-as-rank) || (sym-attr.pointer)
[Bug fortran/49110] Deferred-length character result triggers (false positive) error for pure procedures
http://gcc.gnu.org/bugzilla/show_bug.cgi?id=49110
--- Comment #4 from John jwmwalrus at gmail dot com 2011-05-22 16:46:32 UTC
---
(In reply to comment #2)
This patch allows your code to compile, but I
don't know if it works correctly. Do you have
a complete self-contained small testcase?
Index: resolve.c
===
--- resolve.c (revision 173757)
+++ resolve.c (working copy)
@@ -10305,7 +10305,7 @@ resolve_fl_procedure (gfc_symbol *sym, i
actual length; (ii) To declare a named constant; or (iii) External
function - but length must be declared in calling scoping unit. */
if (sym-attr.function
- sym-ts.type == BT_CHARACTER
+ sym-ts.type == BT_CHARACTER !sym-ts.deferred
sym-ts.u.cl sym-ts.u.cl-length == NULL)
{
if ((sym-as sym-as-rank) || (sym-attr.pointer)
The module shown is self-contained (it only depends on iso_c_binding, which is
an intrinsic module). I've added the corresponding source file to this bug as
an attachment.
[Bug fortran/49110] Deferred-length character result triggers (false positive) error for pure procedures
http://gcc.gnu.org/bugzilla/show_bug.cgi?id=49110
--- Comment #3 from John jwmwalrus at gmail dot com 2011-05-22 16:45:09 UTC
---
(In reply to comment #2)
This patch allows your code to compile, but I
don't know if it works correctly. Do you have
a complete self-contained small testcase?
Index: resolve.c
===
--- resolve.c (revision 173757)
+++ resolve.c (working copy)
@@ -10305,7 +10305,7 @@ resolve_fl_procedure (gfc_symbol *sym, i
actual length; (ii) To declare a named constant; or (iii) External
function - but length must be declared in calling scoping unit. */
if (sym-attr.function
- sym-ts.type == BT_CHARACTER
+ sym-ts.type == BT_CHARACTER !sym-ts.deferred
sym-ts.u.cl sym-ts.u.cl-length == NULL)
{
if ((sym-as sym-as-rank) || (sym-attr.pointer)
The module shown is self-contained (it only depends on iso_c_binding, which is
an intrinsic module). I've added the corresponding source file to this bug as
an attachment.
[Bug fortran/49110] Deferred-length character result triggers (false positive) error for pure procedures
http://gcc.gnu.org/bugzilla/show_bug.cgi?id=49110 --- Comment #5 from John jwmwalrus at gmail dot com 2011-05-22 16:57:02 UTC --- Created attachment 24331 -- http://gcc.gnu.org/bugzilla/attachment.cgi?id=24331 Test case with main unit
[Bug fortran/49110] Deferred-length character result triggers (false positive) error for pure procedures
http://gcc.gnu.org/bugzilla/show_bug.cgi?id=49110 --- Comment #6 from Steve Kargl sgk at troutmask dot apl.washington.edu 2011-05-22 17:11:03 UTC --- On Sun, May 22, 2011 at 04:57:44PM +, jwmwalrus at gmail dot com wrote: This patch allows your code to compile, but I don't know if it works correctly. Do you have a complete self-contained small testcase? The module shown is self-contained (it only depends on iso_c_binding, which is an intrinsic module). I've added the corresponding source file to this bug as an attachment. That's no what I meant. Obviously, I already stated the patch allowed me compile your code. Do you have a COMPLETE self-contained testcase? As in program foo use mod1 DO SOMETHING HERE TO USE c2fstring() end program foo
[Bug fortran/49110] Deferred-length character result triggers (false positive) error for pure procedures
http://gcc.gnu.org/bugzilla/show_bug.cgi?id=49110 --- Comment #7 from John jwmwalrus at gmail dot com 2011-05-22 17:24:07 UTC --- (In reply to comment #6) On Sun, May 22, 2011 at 04:57:44PM +, jwmwalrus at gmail dot com wrote: This patch allows your code to compile, but I don't know if it works correctly. Do you have a complete self-contained small testcase? The module shown is self-contained (it only depends on iso_c_binding, which is an intrinsic module). I've added the corresponding source file to this bug as an attachment. That's no what I meant. Obviously, I already stated the patch allowed me compile your code. Do you have a COMPLETE self-contained testcase? As in program foo use mod1 DO SOMETHING HERE TO USE c2fstring() end program foo Sorry for the double post (I now know the effect of the refresh button on Bugzilla). As I said, the source code is attached (the label is Test case with main unit, the name of the file is test_gfortran_pure_main.f90).
[Bug fortran/49110] Deferred-length character result triggers (false positive) error for pure procedures
http://gcc.gnu.org/bugzilla/show_bug.cgi?id=49110
--- Comment #8 from kargl at gcc dot gnu.org 2011-05-22 18:35:37 UTC ---
I think the subject line is incorrect. The following testcase
that gets rid of the C interop garbage, works for an input
array of characters with an explicit-shape, but fails with
an assumed-shape array.
program foo
implicit none
character s(5)
s = ['a', 'b', 'c', 'd', 'e']
print *, bar(s)
contains
pure function bar(s) result(a)
! character, intent(in) :: s(5) ! Compiles and runs
character, intent(in) :: s(:)! Compiles and segfaults
character(len=:), allocatable :: a
integer i
a = repeat('a', size(s))
do i = 1, size(s)
a(i:i) = s(i)
end do
end function bar
end program foo
The following workaround allows the original code to compile and
run.
function c2fstring(cbuffer) result(string)
character(:), allocatable :: string
character(KIND = C_CHAR), intent(IN) :: cbuffer(255)
character(len=255) s
integer :: i
s = ''
do i = 1, SIZE(cbuffer)
if (cbuffer(i) == C_NULL_CHAR) exit
s(i:i) = cbuffer(i)
end do
string = TRIM(s)
end function
[Bug fortran/49110] Deferred-length character result triggers (false positive) error for pure procedures
http://gcc.gnu.org/bugzilla/show_bug.cgi?id=49110
--- Comment #9 from kargl at gcc dot gnu.org 2011-05-22 19:00:06 UTC ---
Note the routine need not be pure to invoke the segfault. Here's
an even shorter test case.
program foo
implicit none
character s(5)
s = ['a', 'b', 'c', 'd', 'e']
print *, bar(s)
contains
function bar(s) result(a)
! character, intent(in) :: s(5) ! Compiles and runs
character, intent(in) :: s(:)! Compiles and segfaults
character(len=:), allocatable :: a
a = repeat('a', size(s))
end function bar
end program foo
[Bug fortran/49110] Deferred-length character result triggers (false positive) error for pure procedures
http://gcc.gnu.org/bugzilla/show_bug.cgi?id=49110
--- Comment #10 from John jwmwalrus at gmail dot com 2011-05-22 19:36:33 UTC
---
(In reply to comment #9)
Note the routine need not be pure to invoke the segfault. Here's
an even shorter test case.
program foo
implicit none
character s(5)
s = ['a', 'b', 'c', 'd', 'e']
print *, bar(s)
contains
function bar(s) result(a)
! character, intent(in) :: s(5) ! Compiles and runs
character, intent(in) :: s(:)! Compiles and segfaults
character(len=:), allocatable :: a
a = repeat('a', size(s))
end function bar
end program foo
My bug report is actually about the fact that the pure attribute triggers an
(incorrect) error, and that error is associated to the deferred-length
character result. The code compiles just fine by removing the pure attribute
from the original code... Or by changing character(:), allocatable :: string
to something like character(2047) :: string.
So I guess the subject line is fine.
The reason why I used interoperability with C, is because the module is just a
reduced version of the actual, practical case that triggered the error ---sorry
if it's garbage to you.
[Bug fortran/49110] Deferred-length character result triggers (false positive) error for pure procedures
http://gcc.gnu.org/bugzilla/show_bug.cgi?id=49110 --- Comment #11 from Dominique d'Humieres dominiq at lps dot ens.fr 2011-05-22 20:46:56 UTC --- This PR seems related to http://gcc.gnu.org/bugzilla/show_bug.cgi?id=45170#c9 .
[Bug fortran/49110] Deferred-length character result triggers (false positive) error for pure procedures
http://gcc.gnu.org/bugzilla/show_bug.cgi?id=49110
--- Comment #12 from Steve Kargl sgk at troutmask dot apl.washington.edu
2011-05-22 21:30:07 UTC ---
On Sun, May 22, 2011 at 08:03:32PM +, jwmwalrus at gmail dot com wrote:
http://gcc.gnu.org/bugzilla/show_bug.cgi?id=49110
--- Comment #10 from John jwmwalrus at gmail dot com 2011-05-22 19:36:33
UTC ---
(In reply to comment #9)
Note the routine need not be pure to invoke the segfault. Here's
an even shorter test case.
program foo
implicit none
character s(5)
s = ['a', 'b', 'c', 'd', 'e']
print *, bar(s)
contains
function bar(s) result(a)
! character, intent(in) :: s(5) ! Compiles and runs
character, intent(in) :: s(:)! Compiles and segfaults
character(len=:), allocatable :: a
a = repeat('a', size(s))
end function bar
end program foo
My bug report is actually about the fact that the pure attribute
triggers an (incorrect) error, and that error is associated to the
deferred-length character result.
Sigh. I already told you that my patch in Comment #2
fixes the errant error message. That does not fix the
more important bug that is exposed when the errant
error message is suppressed. With my patch in Comment #2,
and without touching your source code, I get
laptop:kargl[228] gfc4x -o z -fno-backtrace test_gfortran_pure_main.f90 ./z
Segmentation fault (core dumped)
The reason why I used interoperability with C, is because the
module is just a reduced version of the actual, practical case
that triggered the error ---sorry if it's garbage to you.
The C interop stuff is garbage in that it makes it much
more difficult to diagnosis and fix the real bug.
Your free to apply my patch gcc, build new gfortran, and use it
to compile and run your code.
I shalt spend anymore time on this PR.
