refactor: update math/base/special/log10 to follow FreeBSD version 12.2.0

[gunjjoshi]

Description

What is the purpose of this pull request?

This pull request:

• updates existing JavaScript implementation to follow FreeBSD version 12.2.0.
• adds a corresponding test.

Related Issues

None.

Questions

Any questions for reviewers of this pull request?

No.

Other

Any other information relevant to this pull request? This may include screenshots, references, and/or implementation notes.

• Code coverage report:

• FreeBSD log10 12.2.0: https://svnweb.freebsd.org/base/release/12.2.0/lib/msun/src/e_log10.c?revision=367086&view=markup

• This might be useful to review this PR: refactor: update math/base/special/log2 to follow FreeBSD version 12.2.0 #2172

Checklist

Please ensure the following tasks are completed before submitting this pull request.

• Read, understood, and followed the contributing guidelines.

---

@stdlib-js/reviewers

[Planeshifter]

Left two small comments with regards to the code comments, otherwise looks good to me!

[gunjjoshi]

Left two small comments with regards to the code comments, otherwise looks good to me!

Thanks, I've made the changes in both main.c and main.js.

[kgryte]

@gunjjoshi @Planeshifter Correct me if I'm wrong, but hx is uint32_t here. Bitwise ops are defined for signed integers. Meaning, we should still be using ihx and ilx here and below.

[kgryte]

I see that we do this in log2 (https://github.com/stdlib-js/stdlib/blob/develop/lib/node_modules/%40stdlib/math/base/special/log2/src/main.c#L75), but I am not sure that is correct or works as intended.

[kgryte]

In

stdlib/lib/node_modules/@stdlib/math/base/special/acosf/src/main.c

Line 104 in d3215eb

hx = (int32_t)uhx;

, we do the explicit cast to int32_t. IIRC, the failure to explicitly cast words to signed integers resulted in test failures in other packages.

Update: after digging around a bit, looks like all bitwise operators are defined for unsigned integers. The issue arises when mixing signed and unsigned integer values. In that instance, unsigned integers will cast to signed integers which can cause unexpected behavior. My sense is that wherever FreeBSD uses signed integers, we should also use signed integers. In that case, we should use ihx (see https://svnweb.freebsd.org/base/release/12.2.0/lib/msun/src/e_log10.c?view=markup).

[gunjjoshi]

Yes @kgryte, the process of casting hx to a signed integer after these operations seems correct. I'll try to reiterate over this implementation once again, keeping this in mind.