[J3] dumb question about module use not inheriting into interfaces

[Van Snyder]

On Fri, 2022-08-26 at 10:04 +0000, Jeff Hammond via J3 wrote:
> It is mildly annoying to have to add a use line to every single
> interface, when it is used in the module already, but I assume there
> is some good reason for this.
In 1986, during development of Fortran 90, a decision was made that
interface bodies do not access their host by host association.

At the time, I pointed out that this made it impossible to get at a
meaningful type definition, except by putting it into a different
module.

I argued that host association ought to work into interface bodies.

There was an inscrutable argument that blocking host association was
necessary so that one could put what used to be external procedures (or
maybe only their specification parts) into interface bodies, and expect
them to work the same way as they had as external procedures. I was
unable to convince X3J3 that this was nonsense.

Loren Meissner has my X3J3 correspondence from that era at the Computer
History Museum. I should have scanned it all before I sent it.

When we added IMPORT, I argued that J3 ought simply to turn on host
association into interface bodies. I don't understand the argument
against it.

-------------- next part --------------
An HTML attachment was scrubbed...
URL: <https://mailman.j3-fortran.org/pipermail/j3/attachments/20220826/520b796b/attachment.htm>