Update PyO3 to 0.16 and to Rust stable channel

[rchildre3]

Addresses #17 and #11

[rchildre3]

Seeing a large regression on benchmarks on my box:

lsb_release -a
#No LSB modules are available.
#Distributor ID:	Ubuntu
#Description:	Ubuntu 21.10
#Release:	21.10
#Codename:	impish
uname -a
#Linux hostname 5.13.0-30-generic #33-Ubuntu SMP Fri Feb 4 17:03:31 UTC 2022 x86_64 x86_64 x86_64 GNU/Linux
cat /proc/cpuinfo
#model name      : Intel(R) Core(TM) i5-6200U CPU @ 2.30GHz
#siblings        : 4
#cpu cores       : 2
python3 -VV
#Python 3.9.7 (default, Sep 10 2021, 14:59:43) 
#[GCC 11.2.0]
python3 -m venv ./.venv
. ./.venv/bin/activate
pip install maturin tox
rm -r target
maturin build --no-sdist --release --strip -i python3
tox

-------------------------------------------------------------------------------- benchmark 'parse-bytes': 2 tests -------------------------------------------------------------------------------
Name (time in us)                Min                Max              Mean            StdDev            Median               IQR              Outliers  OPS (Kops/s)            Rounds  Iterations
-------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------
test_parse_bytes_uuid         1.2150 (1.0)      38.5570 (1.0)      1.4261 (1.0)      0.9790 (1.0)      1.2850 (1.0)      0.0490 (1.0)      4012;14871      701.2219 (1.0)      118977           1
test_parse_bytes_fastuuid     4.6490 (3.83)     45.9060 (1.19)     6.0721 (4.26)     2.8876 (2.95)     4.8880 (3.80)     0.7170 (14.63)     1744;3255      164.6872 (0.23)      15675           1
-------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------

-------------------------------------------------------------------------------- benchmark 'parse-bytes_le': 2 tests --------------------------------------------------------------------------------
Name (time in us)                   Min                 Max              Mean            StdDev            Median               IQR              Outliers  OPS (Kops/s)            Rounds  Iterations
-----------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------
test_parse_bytes_le_uuid         1.7220 (1.0)       61.4320 (1.0)      2.0211 (1.0)      1.2576 (1.0)      1.8170 (1.0)      0.0690 (1.0)      3898;15633      494.7810 (1.0)      109219           1
test_parse_bytes_le_fastuuid     5.7240 (3.32)     119.1090 (1.94)     6.6473 (3.29)     2.7640 (2.20)     5.9880 (3.30)     0.2280 (3.30)      1252;3231      150.4381 (0.30)      20146           1
-----------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------

------------------------------------------------------------------------------- benchmark 'parse-fields': 2 tests --------------------------------------------------------------------------------
Name (time in us)                 Min                Max              Mean            StdDev            Median               IQR              Outliers  OPS (Kops/s)            Rounds  Iterations
--------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------
test_parse_fields_uuid         1.7250 (1.0)      52.1830 (1.0)      2.0304 (1.0)      1.2770 (1.0)      1.8050 (1.0)      0.0650 (1.0)      5561;15965      492.5083 (1.0)      116782           1
test_parse_fields_fastuuid     6.5190 (3.78)     78.0640 (1.50)     7.6877 (3.79)     3.1784 (2.49)     6.8125 (3.77)     0.2970 (4.57)      1547;2988      130.0771 (0.26)      19526           1
--------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------

------------------------------------------------------------------------------- benchmark 'parse-hex': 2 tests ------------------------------------------------------------------------------
Name (time in us)              Min                Max              Mean            StdDev            Median               IQR            Outliers  OPS (Kops/s)            Rounds  Iterations
---------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------
test_parse_hex_uuid         1.5560 (1.0)      42.2360 (1.0)      1.7721 (1.0)      1.1177 (1.0)      1.6240 (1.0)      0.0460 (1.0)     1249;5101      564.3028 (1.0)       42842           1
test_parse_hex_fastuuid     7.5400 (4.85)     54.8400 (1.30)     8.8760 (5.01)     3.3834 (3.03)     7.8510 (4.83)     0.3770 (8.20)     816;1568      112.6627 (0.20)       9835           1
---------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------

[thedrow]

Is this because of the stable rust version or because of the PyO3 version upgrade?

[rchildre3]

Is this because of the stable rust version or because of the PyO3 version upgrade?

Actually seeing similar regressions back on master branch. Maybe the regression is just my machine, do you see the same results?

[thedrow]

Is this because of the stable rust version or because of the PyO3 version upgrade?

Actually seeing similar regressions back on master branch. Maybe the regression is just my machine, do you see the same results?

The previous benchmark runs were on i7. I'll give it a shot now.

[thedrow]

I ran the benchmarks and everything's fine.
When you compile a PyO3 extension in non-optimized mode, it tends to be very slow.
If you did this with tox, you'd get a debug build because with tox we just check that the benchmarks work right now.

[rchildre3]

When you compile a PyO3 extension in non-optimized mode, it tends to be very slow.

I was under the impression that the maturin build --release that I ran would make tox use the optimized version?

Potential other tasks to close this PR

• Regenerate benchmarks?
• Include script such that benchmarks can be easily reproduced?
• Version bump?

[thedrow]

The problem with reproducible benchmarks is that we only have VMs.
VMs are not reliable for measuring performance. See here and here

We can use https://github.com/marketplace/actions/continuous-benchmark to track the benchmarks results.

I think we can bump the version and tag.
Would you like an invite to this repository?

[rchildre3]

The idea of the a benchmarking github action is nice, and probably fitting, but should be its own, separate pull request

Will bump version in Cargo.toml but would prefer git taging stay with the main author.
Would 0.8.0 be appropriate?

[thedrow]

Yes. 0.8.0 would be appropriate.

[rchildre3]

added version bump with squash and rebase, manylinux2_24 job failing due to password, not sure for the fix?

[rchildre3]

resolved by backing out manylinux changes

[rchildre3]

would any other changes be required for this PR to be merged in?

[thedrow]

No. Thank you!