To: J3                                                     J3/19-189
From: Paul Thomas
Subject: GNU view on templates
Date: 2019-July-30
References: J3/18-281r1

I have provoked some email discussions with other gfortran (Tobias
Burnus & Thomas Koenig) and gcc maintainers (Richard Biener and Jakub
Jelnik).  I now believe that my assertion that cost over benefit of the
Templates proposal (18-281r1) would go to infinity was over hasty [1].
We would like the following to be inserted into discussion at the
Tokyo meeting:

(i) Gfortran maintainers have started thinking about both template
functions and iterators but have not, as yet, any concrete proposals
for either or reactions, for official purposes, to any of the existing
proposals;

(ii) That said, we do not see any insurmountable difficulties with
either templates or iterators;

(iii) The g++ "technology" is of little help, other than perhaps
offering some guidance on name mangling schemes; and

(iv) One caveat is that the syntax for the new features should, if
possible, echo existing syntax.  For example, my suggestion of GENERIC
TYPE was intended to provide a syntax that has a familiar feel.
I would also suggest s/TEMPLATE/GENERIC/ for the same reason
(it also is the probable implementation route).

Best regards

Paul


[1] ukfortran -- BSI (UK) Fortran Mailing List,
    https://lists.accu.org/mailman/listinfo/ukfortran.