[m-dev.] Deprecating stuff in builtin module

Julien Fischer jfischer at opturion.com
Tue Dec 1 09:40:30 AEDT 2015

On Tue, 1 Dec 2015, Zoltan Somogyi wrote:

> On Tue, 1 Dec 2015 00:06:42 +1100 (AEDT), Julien Fischer <jfischer at opturion.com> wrote:
>>> No objection from me, but I don't think we have any facilities
>>> for marking a *mode*, as opposed to a *predicate*, as obsolete.
>> We don't, beyond documenting that the modes are deprecated (which is
>> what I intend to do).
> Yes, I figured that. The problem is that if someone has code
> that uses those modes, their code will continue to compile without
> incident, and they won't KNOW that they should read the new comment.
> The comment may therefore help people who look up the documentation
> before writing NEW code, but it won't help most people with existing code.
> For them. deprecating the mode now and deleting it later will be
> effectively indistinguishable from just deleting it later.

We should certainly look at extending obsolescencepragmas to allow
entities other than predicates or functions to be marked as obsolete,
however that's not a high-priority task at the moment IMO.   In this
case, the insts / modes in questions are so obscure that I don't think
it will be a significant problem.


More information about the developers mailing list