[m-dev.] promoted ROTDs

[Zoltan Somogyi]

On Tue, 24 Jul 2018 11:56:02 +1000, Peter Wang <novalazy at gmail.com> wrote:

> On Tue, 24 Jul 2018 11:19:58 +1000, Peter Wang <novalazy at gmail.com> wrote:
> > 
> > Is there anything known to be wrong with rotd-2018-07-23?
> 
> There are some C compiler warnings in hlc grades that would be better to
> fix first, e.g.
> 
> bitmap.c: In function 'mercury__bitmap__to_string_chars_4_p_0':
> bitmap.c:2627:26: warning: initialization makes integer from pointer without a cast [-Wint-conversion]
>          MR_Word Var_25 = ((MR_Box) ((MR_String) "bitmap.to_string: internal error"));
> 
> lexer.c: In function 'mercury__lexer__string_get_source_line_number_7_p_0':
> lexer.c:9887:27: warning: assignment makes integer from pointer without a cast [-Wint-conversion]
>              *HaveToken_12 = ((MR_Box) ((MR_Integer) -1));

I will fix these, since they are almost certainly caused by my recent
changes to ml_unify_gen_*.m. I found them just before I packed up
my laptop for my trip back to Melbourne over the weekend. (I arrived
sunday evening.) I will get on it when my brain stops being mushy
with jet lag :-(

BTW, we need to decide whether we want to switch on the packing optimizations
I have completed so far for the rotd we select as the release, because if we do,
then we want to switch them on for a few rotds *before* that as well, for more
real-world testing.

I strongly suspect, but don't know for certain, that the tabled_typeclass
test case failures are caused by the .exp files not being updated after a change
to the library, rather than a true failure. I will have a look at that as well.

It would be nice if we had an automatic way of catching such problems again,
such as we used to have at UniMelb: servers running bootchecks in every grade
overnight. Does anyone have a spare machine we could use for this? If not,
how much it cost to rent one, e.g. from Amazon? Does anyone know?
aws.amazon.com does not seem to have pricing data available without an account.

Zoltan.