feature: expose SimpleSpanProcessor::new

[demurgos]

Fixes #2060

Changes

This commit update the visibility of the method SimpleSpanProcessor::new from pub(crate) to pub. This enables consumers to create SimpleSpanProcessor values for use with trace::provider::Builder::with_span_processor.

In particular, this fixes a consistency issue: the BatchSpanProcessor struct may already be built thanks to its pub builder. With this change, both processors in this repo may be built manually.

Merge requirement checklist

• CONTRIBUTING guidelines followed
• Unit tests added/updated (if applicable) (not applicable)
• Appropriate CHANGELOG.md files updated for non-trivial, user-facing changes
• Changes in public API reviewed (if applicable)

[codecov]

Codecov Report

All modified and coverable lines are covered by tests ✅

Project coverage is 78.3%. Comparing base (7ab5e0f) to head (6b40f20).
Report is 1 commits behind head on main.

Additional details and impacted files

@@          Coverage Diff          @@
##            main   #2119   +/-   ##
=====================================
  Coverage   78.3%   78.3%           
=====================================
  Files        121     121           
  Lines      20815   20815           
=====================================
+ Hits       16308   16309    +1     
+ Misses      4507    4506    -1     

☔ View full report in Codecov by Sentry.
📢 Have feedback on the report? Share it here.

[lalitb]

looks good, specifically that it allows us to remove with_simple_processor and with_batch_processor methods in future as reasoned here.

[cijothomas]

While I agree with this change (Thanks for fixing the inconsistency), I just want to call out that SimpleSpanProcessor is not going to scale well if you use it on any PROD scenarios. It makes blocking calls, so a Span.End might be blocked waiting for an http response, so Simpleprocessor is best left for debug/learning purposes. As long as you are leveraging it for test purposes no issues!

[demurgos]

@cijothomas Thanks, I confirm that I only use it in development. In production I use the bathed span processor.

The simple span processor has the benefit of being immediate, which helps when debugging locally with traces sent to stdout.