[m-rev.] for post-commit review: improve test coverage of base_string_to_{int, uint}/3

[Julien Fischer]

On Wed, 19 Aug 2026 at 13:09, Zoltan Somogyi <zoltan.somogyi at runbox.com> wrote:
>
>
>
> On Wed, 19 Aug 2026 12:54:15 +1000, Julien Fischer <jfischer at opturion.com> wrote:
>
> > Improve test coverage of base_string_to_{int,uint}/3.
> >
> > The existing test coverage of base_string_to_int/3 is pretty minimal, and that
> > of base_string_to_uint/3 is almost non-existent. Recent changes have
> > improved test coverage for the overflow-related aspects of these predicates.
> > This diff improves test coverage for the non-overflow-related aspects.
>
> This is a good idea. I have two suggestions.
>
> First, I can't use "diff" to compare the input data sets of the two test cases,
> but they seem to be at least similar. I would suggest you replace those two
> test cases with a single test case that tests both the signed and unsigned
> version of each operation on the same set of test data (the unions of the
> current signed and unsigned sets, if they are different), and report just
> one result if the signed and unsigned op yield equivalent results.
> This should eliminate the need for double maintenance in the future.
>
> Second, I would specify minimum widths for all the inputs in the calls
> to io.format. It is easier to check the results if results that should be
> identical all start in the same column.

I've committed the attached diff which addresses both the points you
raised above.

Julien.
-------------- next part --------------
A non-text attachment was scrubbed...
Name: DIFF.string_to_int
Type: application/octet-stream
Size: 20852 bytes
Desc: not available
URL: <http://lists.mercurylang.org/archives/reviews/attachments/20260819/35867e03/attachment-0001.obj>