[m-rev.] Refactoring make_temp

[Paul Bone]

On Thu, May 26, 2016 at 02:28:40PM +0800, Sebastian Godelet wrote:
> Hi Paul,
> 
> 
> From an initial screening I think you preserved all the additions I made, but I've yet to build this.

You probably want to check line 10682.  Note that this code dosn't use the
solution you had to create the file with the correct permissions.  Maybe it
should?

mkdir_temp does use the solution you created, but I hope I havn't broken it.

> 
> Couple of things though:
> 
> In line 10935 (https://github.com/Mercury-Language/mercury/compare/master...PaulBone:make_temp#diff-c8358543868f79aa0b536505b8138fbaR10935) you comment out a huge bunch of code,
> I'd rather rely on version management instead (so git blame and stuff continue to work); also I think io.m is already large enough.

Yes.  It's commented out to make it easy for me to read the old code while I
write new code.  I intend to delete it, and rebase my changes to condese it
into a single patch.

> In line 11226, you changed have_make_temp_directory to always return MR_TRUE, maybe we should get rid off the procedure altogether (if I remember that was a new, so unreleased, addition).

I beleive it returns false on the Erlang backend as that wasn't implemented.
What I changed was that it now always returns true on the C backend.

Oh that reminds me, what I havn't done is go through the C code, look at the
functions and macros/values I use and check that they exist in configure.ac,
and protect them with #ifdef if necessary.

> I'll try to build this changes later
>

Cool.

> P.S: I hope I didn't send emails twice or more times my Outlook was giving up on me. 

Not that I noticed.  Cheers.


-- 
Paul Bone