[m-rev.] for post-commit review: improvements in determinism error handling

[Julien Fischer]

Hi Zoltan,

On Fri, 8 Apr 2011, Zoltan Somogyi wrote:

> For post-commit review by anyone. Mostly I am seeking feedback on the
> name of the new pragma; the implementation is relatively straightforward.

I think the name is too generic; it doesn't give any indication of what
determinism warning is being disabled.  Based on the name of the
corresponding command line option, I would call it

    :- pragma no_warn_lax_determinism(foo/1).

However, I think it would be preferable to introduce a generic
supresss_warnings/2 pragma that takes a list of warning to supress for a
given procedure, e.g. so we could do something like

    :- pragma suppress_warnings(foo/1,
 	[lax_determinism, inferred_erroneous]).

(Being able to shut up the warning about something being inferred as
erroneous would also be useful.)

> The new pragma should also be announced in the NEWS file, but WHERE in the
> NEWS file depends on whether Julien wants to put this into 11.01 (which
> should really be 11.04 now).

It should go into the post 11.01 NEWS section, since this won't go onto
the branch.  (If we weren't using CVS I would rename the release, but as
we are.)

Julien.
--------------------------------------------------------------------------
mercury-reviews mailing list
Post messages to:       mercury-reviews at csse.unimelb.edu.au
Administrative Queries: owner-mercury-reviews at csse.unimelb.edu.au
Subscriptions:          mercury-reviews-request at csse.unimelb.edu.au
--------------------------------------------------------------------------