perf(error): check SourceFragmentsHidden before formatting sources

[samber]

Summary

• ToMap, LogValue and formatVerbose evaluated if sources := o.Sources(); sources != "" && !SourceFragmentsHidden left-to-right: Sources() formatted the full source-fragment output - reading source files from disk and building per-line strings - and the result was then discarded, because SourceFragmentsHidden is true by default. Checking the flag first skips all of that work in the default configuration; behavior with SourceFragmentsHidden=false is unchanged.
• Also replaces fmt.Sprintf with direct concatenation in the per-line / per-block builders (getSourceFromFrame, framesToStacktraceBlocks, framesToSourceBlocks), benefiting the non-hidden path too.

Benchmarks

benchstat vs main (darwin/arm64, Apple M3, -count=10 -cpu=1):

         │ before3.txt │              after3.txt              │
         │   sec/op    │    sec/op     vs base                │
ToMap      3.494µ ± 5%   1.631µ ± 15%  -53.31% (p=0.002 n=10)
LogValue   3.499µ ± 4%   1.478µ ±  8%  -57.77% (p=0.000 n=10)
geomean    3.496µ        1.553µ        -55.60%

         │ before3.txt  │              after3.txt              │
         │     B/op     │     B/op      vs base                │
ToMap      5.214Ki ± 0%   2.789Ki ± 0%  -46.51% (p=0.000 n=10)
LogValue   5.357Ki ± 0%   2.745Ki ± 0%  -48.76% (p=0.000 n=10)
geomean    5.285Ki        2.767Ki       -47.65%

         │ before3.txt │             after3.txt             │
         │  allocs/op  │ allocs/op   vs base                │
ToMap       63.00 ± 0%   28.00 ± 0%  -55.56% (p=0.000 n=10)
LogValue    61.00 ± 0%   24.00 ± 0%  -60.66% (p=0.000 n=10)
geomean     61.99        25.92       -58.18%

BenchmarkNewStacktrace was also checked as a control (this path never calls Sources()): allocations and memory are byte-for-byte identical before/after (0% variance), confirming the change is isolated to the Sources()-gated paths.

Test coverage

The SourceFragmentsHidden=false path (and, by extension, oopsStacktrace.Source() and framesToSourceBlocks) had no coverage at all before this PR — every existing test ran with the package default (true), so the new gated branches and the two previously-0%-covered functions were untested. Added:

• TestOopsSourceFragmentsVisible / TestOopsSourceFragmentsHiddenByDefault: assert ToMap, LogValue and formatVerbose include/omit "sources" depending on the flag. Both run sequentially (no t.Parallel(), //nolint:paralleltest) since the former temporarily flips the package-level flag and the latter asserts on its default.
• TestOopsStacktraceSource: direct coverage of oopsStacktrace.Source() (empty-frames and with-frames cases).
• TestFramesToSourceBlocks: direct coverage of framesToSourceBlocks, including the no-body-skips-block and multi-block-reversal cases.

Package coverage: 90.8% -> 94.0%. Source and framesToSourceBlocks are now at 100%.

Test plan

Verified with go build ./..., go vet ./..., go test ./..., golangci-lint run ./... (0 issues), and go test -bench=BenchmarkToMap -bench=BenchmarkLogValue -benchmem -count=10 -cpu=1 ./benchmarks run before/after the change and compared with benchstat.

[codecov]

Codecov Report

❌ Patch coverage is 90.90909% with 1 line in your changes missing coverage. Please review.
✅ Project coverage is 91.20%. Comparing base (4579c2c) to head (d2050d3).
⚠️ Report is 1 commits behind head on main.

Files with missing lines	Patch %	Lines
error.go	87.50%	1 Missing ⚠️

Additional details and impacted files

@@            Coverage Diff             @@
##             main     #139      +/-   ##
==========================================
+ Coverage   90.50%   91.20%   +0.70%     
==========================================
  Files          15       15              
  Lines        1243     1251       +8     
==========================================
+ Hits         1125     1141      +16     
+ Misses         93       88       -5     
+ Partials       25       22       -3     

Flag	Coverage Δ
unittests	91.20% <90.90%> (+0.70%)	⬆️

Flags with carried forward coverage won't be shown. Click here to find out more.

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

🚀 New features to boost your workflow:

• ❄️ Test Analytics: Detect flaky tests, report on failures, and find test suite problems.

[Copilot]

Pull request overview

This PR optimizes OopsError formatting paths by avoiding expensive source-fragment generation when SourceFragmentsHidden is enabled (the default), while preserving behavior when source fragments are explicitly enabled.

Changes:

• Reorders the SourceFragmentsHidden check ahead of Sources() in LogValue, ToMap, and formatVerbose to skip unnecessary disk reads/string building by default.
• Replaces some fmt.Sprintf usages with direct string concatenation / strconv.Itoa in per-line and per-block formatting helpers.
• Removes now-unneeded fmt imports in files where formatting no longer uses it.

Reviewed changes

Copilot reviewed 3 out of 3 changed files in this pull request and generated no comments.

File	Description
error.go	Checks SourceFragmentsHidden before calling Sources() in LogValue, ToMap, and formatVerbose to avoid wasted work in the default configuration.
sources.go	Replaces fmt.Sprintf with strconv.Itoa + concatenation in per-line source formatting; removes fmt import.
stacktrace.go	Simplifies stacktrace/source block string assembly via concatenation; removes fmt import.

---

💡 Add Copilot custom instructions for smarter, more guided reviews. Learn how to get started.