Fix: Large numbers lose precision in localized messages

[naoNao89]

Big numbers like i64::MAX were showing up wrong in error messages because fluent-rs converts everything to f64 internally. This breaks numbers beyond ±2^53.

Quick fix: check the range first. If it's too big for f64 to handle safely, we pass it as a string instead. Now 9223372036854775807 shows correctly instead of 9223372036854776000.

This is a temporary workaround until fluent-rs fixes their precision issue upstream.

Closes #9294

Related: projectfluent/fluent-rs#337

[mfontanaar]

I think the approach is still bugged, because you are leaving out the case of having a u64 greater than i64::MAX.

I had created a (rough) PR which I think functionally is more correct, but whose code is very ugly (yours is much more polished). You can see the relevant commit here: temp-mfontanaar@4bb0e01

That being said, I'm puzzled about how does your approach works. In my tests, it seemed like FluentArgs was always converting to a f64 , regardless of whether the input was str or i64. But your passing tests are proving me wrong.

[naoNao89]

hmm

[sylvestre]

not super happy that it will be added this in every call of translate!()

[naoNao89]

Extracted logic to safe_set_fluent_arg() helper

[codspeed-hq]

CodSpeed Performance Report

Merging #9300 will improve performances by 8.22%

_{Comparing naoNao89:fix-9294-big-number-precision (52f4c2a) with main (29ec8f3)}

Summary

⚡ 2 improvements
✅ 121 untouched
⏩ 6 skipped¹

Benchmarks breakdown

	Benchmark	BASE	HEAD	Change
⚡	fold_custom_width[50000]	32.1 ms	30 ms	+6.96%
⚡	fold_many_lines[100000]	77.8 ms	71.9 ms	+8.22%

Footnotes

1. 6 benchmarks were skipped, so the baseline results were used instead. If they were deleted from the codebase, click here and archive them to remove them from the performance reports. ↩

[github-actions]

GNU testsuite comparison:

Skipping an intermittent issue tests/misc/tee (passes in this run but fails in the 'main' branch)

[github-actions]

GNU testsuite comparison:

Skip an intermittent issue tests/tail/overlay-headers (fails in this run but passes in the 'main' branch)
Skipping an intermittent issue tests/misc/tee (passes in this run but fails in the 'main' branch)

[sylvestre]

i think it has a bit too much complexity for a corner case :/

[github-actions]

GNU testsuite comparison:

Skip an intermittent issue tests/tail/overlay-headers (fails in this run but passes in the 'main' branch)
Skipping an intermittent issue tests/misc/tee (passes in this run but fails in the 'main' branch)