| Message ID | 549C29FD.7060508@netcologne.de |
|---|---|
| State | New |
| Headers | show |
> OK for trunk? What about the other open branches?
OK for trunk.
After it’s been in for a bit of time, probably OK for all active branches, unless someone (or you) think it’s unwise.
FX
Index: trans-array.c =================================================================== --- trans-array.c (Revision 219011) +++ trans-array.c (Arbeitskopie) @@ -4355,6 +4355,13 @@ gfc_conv_resolve_dependencies (gfc_loopinfo * loop && ss_expr->rank) nDepend = gfc_check_dependency (dest_expr, ss_expr, true); + /* Check for cases like c(:)(1:2) = c(2)(2:3) */ + if (!nDepend && dest_expr->rank > 0 + && dest_expr->ts.type == BT_CHARACTER + && ss_expr->expr_type == EXPR_VARIABLE) + + nDepend = gfc_check_dependency (dest_expr, ss_expr, false); + continue; }
Hello world, Merry Christmas everybody! This patch resolves one of the remaining F95 bugs by introducing a temporary for the case c(:)(1:2) = c(2)(2:3) It does have the disadvantage that c(:)(1:2) = c(2)(3:4) now also creates an unnecessary temporary because gfc_check_dependency is not smart about substrings. However, I think this is OK for fixing a wrong-code bug. I would downgrade the PR to an enhancement request after commiting this fix, and look at fixing the substring handling. OK for trunk? What about the other open branches? Thomas 2014-12-25 Thomas Koenig <tkoenig@gcc.gnu.org> PR fortran/56867 * trans-array.c (gfc_conv_resolve_dependencies): Also check dependencies when there may be substrings of character arrays. 2014-12-25 Thomas Koenig <tkoenig@gcc.gnu.org> PR fortran/56867 * gfortran.dg/dependency_45.f90: New test.