[m-rev.] for review: Improve Mercury's temporary files and directories predicates.

[Paul Bone]

On Wed, Apr 13, 2016 at 03:37:57PM +1000, Julien Fischer wrote:
> 
> Hi Paul,
> 
> On Wed, 13 Apr 2016, Paul Bone wrote:
> 
> >On Wed, Apr 13, 2016 at 02:15:32PM +1000, Julien Fischer wrote:
> >>
> >>Hi Paul,
> >>
> >>>+
> >>>+:- pred system_temp_dir(string::out, int::out, io::di, io::uo) is det.
> >>
> >>Use a bool rather than an int for the second argument.
> >>(bool.YES and bool.NO in Java; mr_bool.YES and mr_bool.NO in C#)
> >>
> >
> >The other suggestions are fine, however I cannot follow through on this one.
> >Your recent change to allow this for the Java backend needs a couple of
> >weeks to bootstrap (which I can live with).
> 
> This is nothing to do with my upcoming changes to the Java backend.
> The export of the Mercury constructors yes/0 and no/0 to Java under
> the above names has been supported for years.   See the section
> ``Java data passing conventions'' in the reference manual.

Ah, What is changing then?

> As for Erlang, I think you just need to return {yes} or {no} as
> appropriate.  (Peter can correct me on that, if I'm wrong!)
> 
> >And the Erlang backend doesn't seem to support this.
> 
> ``Erlang data passing conventions'' in the reference manual.

I opened this section and searched for "bool" there were no matches.


-- 
Paul Bone