Refactor formatting

[pitdicker]

trafficstars

We currently have a 234-line format_inner function. In https://github.com/chronotope/chrono/pull/1163 I found a way to make it a bit nicer and faster. This is only the initial refactoring, that splits it up into three methods on DelayedFormat.

Every commit is as close to a copy-paste as compiles and passes rustfmt.

[codecov[bot]]

Codecov Report

Attention: Patch coverage is 94.08602% with 11 lines in your changes are missing coverage. Please review.

Project coverage is 91.78%. Comparing base (ffe2745) to head (9fb956e).

Files	Patch %	Lines
src/format/formatting.rs	94.08%	11 Missing :warning:

Additional details and impacted files

@@            Coverage Diff             @@
##             main    #1335      +/-   ##
==========================================
- Coverage   91.79%   91.78%   -0.01%     
==========================================
  Files          37       37              
  Lines       18175    18167       -8     
==========================================
- Hits        16683    16674       -9     
- Misses       1492     1493       +1     

:umbrella: View full report in Codecov by Sentry.
:loudspeaker: Have feedback on the report? Share it here.

[pitdicker]

Thank you for reviewing!

There were a few memory allocation changes from pass-by-reference to pass-by-copy (and back). I didn't track those through-full, but in the compiler I trust.

Yes, this is just a copy-paste and minimal changes from by-reference to by-value as needed to keep the code compiling.

v is short for value. It is used within format_numeric for whatever value needs to be formatted as an integer.

For context:

• #1196 was a large rewrite of our formatting code. Parts of it are performance improvements and making the code no_std compatible. The goal is to make it easier to re-use parsed format strings, and to avoid panics while formatting.
• #1163 was split out as a first step, including only the performance improvements and no_std compatibility.
• This PR is split out from #1163 as just the most basic refactor.
• I have another branch to build on this one and port over the performance improvements.

I hope this PR can land before it rots and needs to be redone from 0 to keep it a copy-paste. Hmmm, already has a conflict, I'll have to see about that.

[pitdicker]

Looks sane. Reviewing this is quite something, mostly because the huge indentation and lack of encapsulation of the old code. So this patch is surely a welcome improvement.

Rustfmt kept changing the formatting of this huge method. So part of the reason to split it over multiple methods is to avoid that.

[pitdicker]

I hope this PR can land before it rots and needs to be redone from 0 to keep it a copy-paste. Hmmm, already has a conflict, I'll have to see about that.

My own fault with #1473 :smile:. That's doable.

[pitdicker]

@djc Do you want to give this a second look?