Closed kevinzwang closed 1 day ago
Comparing kevin/pyobject-deser-seq
(31ed37c) with main
(7922d2d)
❌ 2
regressions
✅ 15
untouched benchmarks
:warning: Please fix the performance issues or acknowledge them on CodSpeed.
Benchmark | main |
kevin/pyobject-deser-seq |
Change | |
---|---|---|---|---|
❌ | test_iter_rows_first_row[100 Small Files] |
188.3 ms | 229.2 ms | -17.87% |
❌ | test_show[100 Small Files] |
14.8 ms | 31.4 ms | -52.79% |
Attention: Patch coverage is 0%
with 11 lines
in your changes missing coverage. Please review.
Project coverage is 77.17%. Comparing base (
a9bf7c0
) to head (31ed37c
). Report is 5 commits behind head on main.
Files with missing lines | Patch % | Lines |
---|---|---|
src/common/py-serde/src/python.rs | 0.00% | 11 Missing :warning: |
🚨 Try these New Features:
Seems fine, but should we really be using JSON as the serde mechanism instead of bytes?
I had this question too, but we can address that later. @samster25 wrote the IO Config pickling so maybe he should know the rationale.
I had this question too, but we can address that later. @samster25 wrote the IO Config pickling so maybe he should know the rationale.
I did serde_json
at the start since we all had were ints and strings and it was easy to read as a human and debug. However given that we now have bytes, pickles and whatnot inside of the payload, I think it makes sense to just encode it as bytes instead.
@kevinzwang maybe good as a quick follow-up to switch us to using bytes
@kevinzwang maybe good as a quick follow-up to switch us to using bytes
visit_seq
is used when using serde_json to serialize/deserialize Rust objects, since the byte buffer is just stored as a list of numbers in JSON.