[Bug fortran/60928] gfortran issue with allocatable components and OpenMP

[quantheory at gmail dot com]

https://gcc.gnu.org/bugzilla/show_bug.cgi?id=60928

--- Comment #5 from Sean Santos <quantheory at gmail dot com> ---
Well, I thought I understood this, but maybe not.

I was thinking that "subobject" in this context meant "component". A "list
item" here is just any variable or common block listed in a clause, in this
case the private clause. And "allocatable" would have to apply to the
component, not the variable, because otherwise it would make no sense to say
that the subobject has an allocation status. So in this interpretation, the
third and fourth examples you list would be perfectly valid cases.

The problem with that is that Fortran 2003 defines "subobject" in a way that
can include array sections, and OpenMP 4.0 also allows "list item" to refer to
an array section. So there's an alternative interpretation here, which is that
all this "subobject" stuff is just a very poorly worded way of referring to the
fact that you can have a section of an array be thread private. (But then it is
still kind of strange to talk about the "allocation status" of an array
section.)

Anyway, I retract what I said before. It's not clear to me what the standard is
trying to say, and just as plausible that this is trying to refer to array
sections as to derived type components.