Kevin Turcios
ca198ce5ab
fix: update test expectations for multi-round benchmark plugin
...
- test_trace_multithreaded_benchmark: SELECT DISTINCT collapses all 10
threaded sorter calls to 1 row (identical metadata), change 10 → 1
- test_trace_benchmark_decorator: accept zero timing when func_time >
total_time triggers the overflow guard in validate_and_format
2026-04-02 11:18:33 -05:00
Kevin Turcios
74c29b20b1
fix: update tests for multi-round benchmark plugin
...
The benchmark plugin now runs multiple rounds with calibrated
iterations. Tests need SELECT DISTINCT for row counts and must
extract median_ns from BenchmarkStats before validation.
2026-04-02 07:24:55 -05:00
claude[bot]
887e6384b6
fix: apply consistent conn safety pattern in trace benchmark tests
...
Initialize conn to None before try blocks and guard finally with
if conn is not None to prevent NameError if sqlite3.connect() raises.
Co-authored-by: Kevin Turcios <KRRT7@users.noreply.github.com>
2026-03-04 07:37:03 +00:00
claude[bot]
59ec38eec4
fix: resolve mypy type error for conn variable in test_trace_benchmarks
...
Initialize conn as Optional before try block to allow None assignment.
Co-authored-by: Kevin Turcios <undefined@users.noreply.github.com>
2026-03-04 07:15:22 +00:00
Kevin Turcios
17730663ec
fix: close SQLite connections in finally blocks for Windows compatibility
...
Ensures SQLite connections are always closed before file cleanup to
prevent PermissionError on Windows where open handles block file deletion.
2026-03-04 02:12:18 -05:00
misrasaurabh1
198487bf81
format and lint all
2026-01-29 01:39:48 -08:00
Kevin Turcios
328ccb2079
windows shenanigans
2025-09-28 05:49:36 +00:00
Kevin Turcios
be1ae9332b
normalize paths and tmp_dir here too
2025-09-28 05:40:20 +00:00
Kevin Turcios
16d21e7464
reinsert
2025-07-29 23:49:22 -07:00
Kevin Turcios
41d314ddbf
cleanup benchmark
2025-07-28 16:46:05 -07:00
Kevin Turcios
c6776dc06b
cleanup plugin
2025-07-23 17:45:00 -07:00
aseembits93
4e8f2e7f55
cleaning up
2025-05-30 15:06:41 -07:00
aseembits93
c566436aac
modify existing test
2025-05-30 15:03:50 -07:00
Alvin Ryanputra
e5ca10fbb6
marked multithreaded trace benchmarks test to be skipped during CI as its flaky with github action machines
2025-04-17 11:11:12 -04:00
Alvin Ryanputra
f14cf010ee
skip multithreaded benchmark test if machine is single threaded (fixes flaky github actions test, as sometimes the machines allocated are different)
2025-04-17 10:56:20 -04:00
Alvin Ryanputra
0146d828bc
line number test is off by 1 for python versions 39 and 310, removed the check
2025-04-17 10:16:56 -04:00
Alvin Ryanputra
3158f9cc1c
end to end test that proves picklepatcher works. example shown is a socket (which is unpickleable) that's used or not used
2025-04-10 21:43:56 -04:00
Alvin Ryanputra
d6ed1c33c4
Support recursive functions, and @benchmark / @pytest.mark.benchmark ways of using benchmark. created tests for all of them
2025-04-09 17:19:20 -07:00
Alvin Ryanputra
972ef4648e
benchmark flow is working. changed paths to use module_path instead of file_path for Benchmarkkey
2025-04-09 17:19:18 -07:00
Alvin Ryanputra
b3c83204f8
Use module path instead of file path for benchmarks, improved display to console.
2025-04-09 17:19:13 -07:00
Alvin Ryanputra
75c1be7b11
replay tests are now grouped by benchmark file. each benchmark test file will create one replay test file.
2025-04-09 17:19:12 -07:00
Alvin Ryanputra
e5a8260ac9
reworked matching benchmark key to test results.
2025-04-09 17:19:07 -07:00
Alvin Ryanputra
a82e9f0d49
modified the benchmarking approach. codeflash_trace and codeflash_benchmark_plugins are singleton instances that will both handle writing to disk. enables flushing to disk once a limit is reached. also added various details to the tracer
2025-04-09 17:19:02 -07:00
Alvin Ryanputra
357f5863b4
refactored get_function_benchmark_timings and get_benchmark_timings into BenchmarkDatabaseUtils class
2025-04-09 17:18:56 -07:00
Alvin Ryanputra
baac96451e
works with multithreading, added test
2025-04-09 17:18:56 -07:00
Alvin Ryanputra
7601895651
modified paths in test
2025-04-09 17:18:56 -07:00
Alvin Ryanputra
82cb7a9e0e
cleanup tests better
2025-04-09 17:18:56 -07:00
Alvin Ryanputra
adffb9d501
replay test functionality working for functions, methods, static methods, class methods, init. basic instrumentation logic for codeflash_trace done.
2025-04-09 17:18:53 -07:00
Alvin Ryanputra
26b2c4fa61
initial replay test version working
2025-04-09 17:18:51 -07:00
Alvin Ryanputra
d6217e8f8e
work on new replay_test logic
2025-04-09 17:18:22 -07:00
Alvin Ryanputra
1f3fd4d2db
initial attempt for codeflash_trace_decorator
2025-04-09 17:18:22 -07:00
Alvin Ryanputra
965e2c818c
initial implementation for tracing benchmarks using a plugin, and projecting speedup
2025-04-09 17:18:08 -07:00
