| Summary: | [Regression 4.3] Error without comp. initialization: Module variable with a comp. initialization must have SAVE | ||
|---|---|---|---|
| Product: | gcc | Reporter: | Tobias Burnus <burnus> |
| Component: | fortran | Assignee: | Not yet assigned to anyone <unassigned> |
| Status: | RESOLVED WORKSFORME | ||
| Severity: | normal | CC: | gcc-bugs |
| Priority: | P3 | Keywords: | diagnostic, rejects-valid |
| Version: | 4.3.0 | ||
| Target Milestone: | --- | ||
| Host: | Target: | ||
| Build: | Known to work: | ||
| Known to fail: | Last reconfirmed: | ||
Bootstapped & re-tested => Works now. |
The following code does not contain a component initialization, nonetheless, gfortran emits the following error: type(myT) :: t 1 Error: Module variable 't' at (1) with a default initialization must have the SAVE attribute This constrain is checked in result.c, excerpt from that file: /* 4th constraint in section 11.3: "If an object of a type for which component-initialization is specified (R429) appears in the specification-part of a module and does not have the ALLOCATABLE or POINTER attribute, the object shall have the SAVE attribute." */ module good2 implicit none type myT integer :: j ! = 5.5 integer, pointer :: p ! => null() end type myT type(myT) :: t end module good2 program main use good2 implicit none t%j = 15 nullify(t%p) end program main (Interestingly, only g95 (and gfortran) give an error if one uses component initialization, NAG f95, sunf95 and ifort don't.)