generates an ICE, I think it is a regression write (*,'(2A3)') 'X'//(/"1","2"/)//'Y' END
Yes, it is indeed a regression, which was introduced between 2007-05-25-r125057 and 2007-05-29-r125159. ==26453== at 0x474E05: gfc_add_loop_ss_code (trans-array.c:1638) ==26453== by 0x475495: gfc_conv_loop_setup (trans-array.c:3288) ==26453== by 0x494070: gfc_trans_transfer (trans-io.c:1969) ==26453== by 0x46D2C4: gfc_trans_code (trans.c:690)
(In reply to comment #1) > Yes, it is indeed a regression, which was introduced between > 2007-05-25-r125057 and 2007-05-29-r125159. ...and is almost certainly my doing - damn! I am on to it Paul
(In reply to comment #2) Yes - it's r125088 that is responsible for the regression. Paul
> Yes - it's r125088 that is responsible for the regression. http://gcc.gnu.org/ml/gcc-cvs/2007-05/msg00788.html
Subject: Bug 32156 Author: pault Date: Thu May 31 21:11:31 2007 New Revision: 125240 URL: http://gcc.gnu.org/viewcvs?root=gcc&view=rev&rev=125240 Log: 2007-05-31 Paul Thomas <pault@gcc.gnu.org> PR fortran/32156 * trans-array.c (gfc_trans_array_constructor): Treat the case where the ss expression charlen is missing. 2007-05-31 Paul Thomas <pault@gcc.gnu.org> PR fortran/32156 * gfortran.dg/char_array_constructor_3.f90: New test. Added: trunk/gcc/testsuite/gfortran.dg/char_array_constructor_3.f90 Modified: trunk/gcc/fortran/ChangeLog trunk/gcc/testsuite/ChangeLog
Subject: Bug 32156 Author: pault Date: Thu May 31 21:12:10 2007 New Revision: 125241 URL: http://gcc.gnu.org/viewcvs?root=gcc&view=rev&rev=125241 Log: 2007-05-31 Paul Thomas <pault@gcc.gnu.org> PR fortran/32156 * trans-array.c (gfc_trans_array_constructor): Treat the case where the ss expression charlen is missing. 2007-05-31 Paul Thomas <pault@gcc.gnu.org> PR fortran/32156 * gfortran.dg/char_array_constructor_3.f90: New test. Modified: trunk/gcc/fortran/trans-array.c
Fixed as 'obvious'. Paul