Muir Manders 749a5cd4ad gopls/internal/regtest: fix completion benchmarks ... I fixed a check in sandbox.go to check for size of config.Files instead of nilness. The completion benchmarks run with an absolute path workdir and were failing this check due to a non-nil but empty config.Files. I tweaked the benchmark output so it is compatible with benchstat. In particular, the benchmark output now appears all on one line for an imaginary benchmark named BenchmarkStatistics. I also made a couple changes to the completion benchmarks: - Don't modify the buffer before every completion. Type checking completely dominates completion, so if it has to type check every time then you aren't benchmarking the completion code at all. - Don't try to exclude GC from the benchmark. I think amortized GC time should be included in the benchmark timing. Plus, I'm not sure that forcing a GC every 10 iterations was actually doing a good job excluding GC from the benchmark. Change-Id: I53718a5f6e25453146ccf5bb5fdfdfc65e244df3 Reviewed-on: https://go-review.googlesource.com/c/tools/+/323251 Trust: Rebecca Stambler <rstambler@golang.org> Trust: Cherry Mui <cherryyz@google.com> Run-TryBot: Rebecca Stambler <rstambler@golang.org> gopls-CI: kokoro <noreply+kokoro@google.com> TryBot-Result: Go Bot <gobot@golang.org> Reviewed-by: Robert Findley <rfindley@google.com>		2021-08-12 16:12:36 +00:00
..
bench_test.go	gopls/internal/regtest: fix completion benchmarks	2021-08-12 16:12:36 +00:00
completion_bench_test.go	gopls/internal/regtest: fix completion benchmarks	2021-08-12 16:12:36 +00:00
stress_test.go	internal/lsp/regtest: add a benchmark for didChange	2021-05-06 02:57:18 +00:00