[m-rev.] for review: representations of complex types in .int files

Julien Fischer jfischer at opturion.com
Wed May 27 21:56:34 AEST 2020


On Wed, 27 May 2020, Zoltan Somogyi wrote:

> On Wed, 27 May 2020 21:17:40 +1000 (AEST), Julien Fischer <jfischer at opturion.com> wrote:
>> There are a few spots in this diff where you should be able to use
>> the support for uints with io.format in order to avoid casts.
>
> Ok, but that will require updating configure.ac to require a very recent
> compiler. I think I will leave that for a followup change, after I install
> a compiler that has your recent changes :-)

I will be adding a change on Friday that make the necessary update to
the configure script and removes various workarounds I put in place for
that change.

>> I assume that at the moment packing does not consider single-precision
>> floats on 64-bit platforms?
>
> While spf works differently than non-spf grades on 64 bit platforms,
> the current implementation reserves full word for a float on 64 bit
> machines with or without spf. This diff preserves that behavior.
> So this diff "considers" spf on 64 bit platforms by ignoring it :-)
>
> We could try treating floats in spf grades the same way we treat
> int32s, allowing them to be packed together with other sub-word-sized
> values, but that would be a separate change. I for one don't think
> that change is at all urgent.

No, it was just curiosity on my part.

Julien.


More information about the reviews mailing list