Split DateTimeFormat keys

[zbraniecki]

Fixes #257

[zbraniecki]

@sffc does it look generally like the right direction to you? How should I handle StructProvider in such a composed bound scenario?

[sffc]

@sffc does it look generally like the right direction to you?

Yes, this looks like what I had in mind!

How should I handle StructProvider in such a composed bound scenario?

I would think that in order to make a StructProvider return multiple types of structs, one would have two StructProvider instances and then tie them together with another class that delegates to one or the other. Perhaps we could make a helper macro for that.

[jira-pull-request-webhook]

Notice: the branch changed across the force-push!

• components/datetime/tests/datetime.rs is different

View Diff Across Force-Push

~ Your Friendly Jira-GitHub PR Checker Bot

[codecov-commenter]

Codecov Report

Merging #774 (d1d8410) into main (ffd520f) will decrease coverage by 0.63%.
The diff coverage is 70.88%.

@@            Coverage Diff             @@
##             main     #774      +/-   ##
==========================================
- Coverage   76.16%   75.52%   -0.64%     
==========================================
  Files         189      192       +3     
  Lines       12153    12296     +143     
==========================================
+ Hits         9256     9287      +31     
- Misses       2897     3009     +112     

Impacted Files	Coverage Δ
components/datetime/src/provider/gregory.rs	74.64% <ø> (+2.64%)	⬆️
components/datetime/src/provider/helpers.rs	79.24% <ø> (ø)
provider/cldr/src/transform/mod.rs	10.98% <12.50%> (-0.27%)	⬇️
provider/cldr/src/transform/dates/mod.rs	41.66% <41.66%> (ø)
provider/cldr/src/transform/dates/patterns.rs	79.48% <79.48%> (ø)
provider/cldr/src/transform/dates/symbols.rs	80.88% <80.88%> (ø)
components/datetime/src/format/datetime.rs	50.00% <100.00%> (ø)
components/datetime/src/skeleton.rs	81.56% <100.00%> (ø)
provider/core/src/resource.rs	81.77% <100.00%> (+0.08%)	⬆️
...s/locale_canonicalizer/src/locale_canonicalizer.rs	76.29% <0.00%> (-19.08%)	⬇️
... and 10 more

---

Continue to review full report at Codecov.

Legend - Click here to learn more
Δ = absolute <relative> (impact), ø = not affected, ? = missing data
Powered by Codecov. Last update ffd520f...d1d8410. Read the comment docs.

[zbraniecki]

This is now ready for review.

It seems to give a 5-8% win in cargo criterion bench for me. I'm not sure why tbh because we still load both for every DateTimeFormat instance, but I assume it opens up ability to later implement logic to only load symbols if needed, right?

[coveralls]

Pull Request Test Coverage Report for Build 39aed657b0376692d800deb921232988975fa000-PR-774

• 241 of 340 (70.88%) changed or added relevant lines in 7 files are covered.
• 2 unchanged lines in 1 file lost coverage.
• Overall coverage decreased (-0.05%) to 75.598%

---

Changes Missing Coverage	Covered Lines	Changed/Added Lines	%
provider/cldr/src/transform/mod.rs	2	16	12.5%
provider/cldr/src/transform/dates/patterns.rs	93	117	79.49%
provider/cldr/src/transform/dates/symbols.rs	110	136	80.88%
provider/cldr/src/transform/dates/mod.rs	25	60	41.67%

Files with Coverage Reduction	New Missed Lines	%
provider/cldr/src/transform/mod.rs	2	10.99%

Totals
Change from base Build acf388649a30f8a2c3d39fa764a36c10996c7482:	-0.05%
Covered Lines:	9415
Relevant Lines:	12454

---

💛 - Coveralls

[sffc]

LGTM!

[jira-pull-request-webhook]

Notice: the branch changed across the force-push!

• components/datetime/src/datetime.rs is different
• components/datetime/tests/datetime.rs is different

View Diff Across Force-Push

~ Your Friendly Jira-GitHub PR Checker Bot

[gregtatum]

LGTM!