| Summary: | [4.3 Regression] .mod file misses renamed, USEd symbol | ||
|---|---|---|---|
| Product: | gcc | Reporter: | Jos de Kloe <kloedej> |
| Component: | fortran | Assignee: | Paul Thomas <pault> |
| Status: | RESOLVED FIXED | ||
| Severity: | normal | CC: | burnus, dfranke, gcc-bugs |
| Priority: | P3 | Keywords: | rejects-valid |
| Version: | 4.3.0 | ||
| Target Milestone: | --- | ||
| URL: | http://gcc.gnu.org/ml/fortran/2008-01/msg00077.html | ||
| Host: | Target: | ||
| Build: | Known to work: | 4.2.1 4.1.3 | |
| Known to fail: | 4.3.0 | Last reconfirmed: | 2008-01-07 13:06:22 |
| Bug Depends on: | |||
| Bug Blocks: | 32834 | ||
If one changes the order of the USE statements or remove the second USE statement, it works:
module MyMod3
USE MyMod2, only: write_MyInt => write_int
USE MyMod1, only: i2_
end module MyMod3
Looking at the MOD file,
(('write_myint' 'mymod2' 2))
is missing for the original example, but present if one one does either change mentioned above.
Another way of looking as it is that the interface is not getting renamed when MyMod3 is written. This works:
MODULE MyMod1
integer, parameter :: i2_ = Selected_Int_Kind(4)
END Module MyMod1
module MyMod2
INTERFACE write_int
module procedure write_int_local
END INTERFACE
contains
subroutine write_int_local(value)
integer, intent(in) :: value
print *,value
end subroutine write_int_local
end module MyMod2
module MyMod3
USE MyMod2, only: write_MyInt => write_int
USE MyMod1, only: i2_
end module MyMod3
module MyMod4
USE MyMod3, write_MyInt => write_int ! Use the original name
end module MYMOD4
I believe, although I have not yet checked, that when I reformed the interface handling in modules I did not change the writing part. I'll bet it uses the symbol name, rather than the symtree. This would have worked previously.
Paul
Since I have submitted the patch, I should take it on! Paul Subject: Bug 34672 Author: pault Date: Mon Jan 7 16:44:43 2008 New Revision: 131377 URL: http://gcc.gnu.org/viewcvs?root=gcc&view=rev&rev=131377 Log: 2008-01-07 Paul Thomas <pault@gcc.gnu.org> PR fortran/34672 * module.c (write_generic): Rewrite completely. (write_module): Change call to write_generic. 2008-01-07 Paul Thomas <pault@gcc.gnu.org> PR fortran/34672 * gfortran.dg/use_only_2.f90: New test. Added: trunk/gcc/testsuite/gfortran.dg/use_only_2.f90 Modified: trunk/gcc/fortran/ChangeLog trunk/gcc/fortran/module.c trunk/gcc/testsuite/ChangeLog Fixed on trunk Paul Dear people, this comment is just to let you know that the problem is also solved on my side now. My software compiles and runs again as expected. thanks a lot for your effort and fast response. Jos de Kloe |
The attached piece of valid Fortran90 code generates a symbol error: >gfortran -c moduletest.F90 moduletest.F90:22.19: USE MyMod3, only: write_MyInt 1 Error: Symbol 'write_myint' referenced at (1) not found in module 'mymod3' > Note that when the line marked with "mark" is commented out the error disappears! This seems a bug to me in gfortran. The code works just fine for the 3 other fortran compilers that I have installed (ifort, pgf90, g95), and also works fine with my previous test version of gfortran (version of 20071024), but the current version 20080103 shows this bug. Best regards, Jos de kloe ----- The following code demonstrates this behaviour: MODULE MyMod1 integer, parameter :: i2_ = Selected_Int_Kind(4) END Module MyMod1 module MyMod2 INTERFACE write_int module procedure write_int_local END INTERFACE contains subroutine write_int_local(value) integer, intent(in) :: value print *,value end subroutine write_int_local end module MyMod2 module MyMod3 USE MyMod2, only: write_MyInt => write_int USE MyMod1, only: i2_ ! mark end module MyMod3 module MyMod4 USE MyMod3, only: write_MyInt !contains ! subroutine MyWrite(a) ! integer, intent(in) :: a ! call write_MyInt(a) ! end subroutine MyWrite end module MYMOD4