lp:astral-uv

Author: Zanie
Author Date: 2024-02-13 04:15:06 UTC

Empty commit to test cache

Author: Andrew Gallant
Author Date: 2023-11-07 18:52:47 UTC

temporary patch for benchmarking

Author: Zanie
Author Date: 2024-02-05 14:36:38 UTC

Bump to merged PubGrub commit

Author: Charlie Marsh
Author Date: 2024-02-04 03:04:53 UTC

Pre-fetch

Author: Andrew Gallant
Author Date: 2024-01-27 01:35:45 UTC

hack: make rkyv work by ignoring DataWithCachePolicy

The DataWithCachePolicy poses some challenges for use with rkyv.
In the interest of getting some kind of measurement with rkyv,
this commit hacks around DataWithCachePolicy by stubbing it out.

This commit isn't meant to be merged and will likely be
completely thrown out.

Author: Zanie
Author Date: 2024-01-26 22:24:17 UTC

Hack additional version tracking into `solve`

Author: Zanie
Author Date: 2024-01-26 20:20:34 UTC

Drop retries for flaking tests

Author: Zanie
Author Date: 2024-01-26 20:46:58 UTC

Baseline

Author: Zanie
Author Date: 2024-01-26 20:07:07 UTC

Baseline

Author: Zanie
Author Date: 2024-01-26 19:30:12 UTC

Empty commit to test cache

Author: Zanie
Author Date: 2024-01-26 00:49:56 UTC

Bump packse

Author: Charlie Marsh
Author Date: 2024-01-26 14:22:39 UTC

Add dedicated bench

Author: Zanie
Author Date: 2024-01-25 16:16:10 UTC

Use 3.8.18 instead of 3.12.1

Not available in GitHUb

Author: Zanie
Author Date: 2024-01-25 19:26:43 UTC

Bootstrap in bash

Author: Charlie Marsh
Author Date: 2024-01-24 17:29:27 UTC

Restructure interpreter queries

Author: Zanie
Author Date: 2024-01-23 20:31:02 UTC

Add script to download Python versions

Author: Zanie
Author Date: 2024-01-24 00:33:57 UTC

Add support for downloading conda binaries

Author: Zanie
Author Date: 2024-01-22 23:00:39 UTC

Revert "Run tests on macOS too"

This reverts commit 8fc904459e3fc54a5b088d70a5d63482b1914723.

Author: Zanie
Author Date: 2024-01-22 01:44:17 UTC

DEMO: Include the installed version in the message

Author: Charlie Marsh
Author Date: 2024-01-20 00:32:29 UTC

Re-enable zlib-ng

Author: Zanie
Author Date: 2024-01-19 22:27:47 UTC

DEMO: Changing Python constraints changes deriviation tree ordering

Author: Zanie
Author Date: 2024-01-19 03:06:03 UTC

Commit with an edit in `puffin-traits`

Author: Charlie Marsh
Author Date: 2024-01-19 02:04:17 UTC

Ignore flaky test in pip_install_scenarios.rs

Author: Zanie
Author Date: 2024-01-18 18:29:37 UTC

Collapse no versions

Author: Zanie
Author Date: 2024-01-17 20:06:31 UTC

Link hookd.py to create

Author: Charlie Marsh
Author Date: 2024-01-17 15:14:11 UTC

Remove a stopship

Author: Charlie Marsh
Author Date: 2024-01-12 21:05:33 UTC

Split version map out from resolver index

Author: Bojan Serafimov
Author Date: 2024-01-11 21:14:50 UTC

no unwrap

Author: Bojan Serafimov
Author Date: 2024-01-11 19:19:15 UTC

WIP

Author: Zanie
Author Date: 2024-01-11 18:48:27 UTC

Empty commit

Author: Zanie
Author Date: 2024-01-10 21:07:00 UTC

Display "and" instead of "," in bound intersections

Author: Zanie
Author Date: 2024-01-10 17:09:08 UTC

Restore space in version range segments with intersections

Author: Zanie
Author Date: 2024-01-10 16:45:00 UTC

Fix installation of packages in Poetry which requires "==*" to not auto-pin

Author: Zanie
Author Date: 2024-01-09 17:13:47 UTC

Remove spaces from display of ranges

With the goal of disambiguating ranges from other clauses in the sentence. There remain problems with understanding the meaning of the `,` and `|` operators as well as precedence and general readability

Author: konstin
Author Date: 2023-12-21 10:47:37 UTC

Add pubgrub testing script

I'm using this script to provide feedback to pubgrub changes, and it's better to have it in tree than only on my machine.

Author: konstin
Author Date: 2023-12-27 20:11:00 UTC

Dir-file conflict

```
virtualenv --clear -p 3.10 .venv -q
cargo run --bin puffin -q -- pip-sync scripts/dir_file_conflict/requirements.txt
```

Author: Zanie
Author Date: 2023-12-20 18:22:32 UTC

Exclude the number of cleared cache entries from the snapshot

Author: Charlie Marsh
Author Date: 2023-12-17 20:11:36 UTC

Build editables first

Author: konstin
Author Date: 2023-12-15 09:32:35 UTC

Use dummy reporter instead of option reporter

I'm experimenting with getting with of the `if let Some(reporter) = self.reporter {...}` and `with_reporter()` methods. Added bonus is that we see where we're currently loosing information, as we currently do in `BuildDispatch::install`.

Author: konstin
Author Date: 2023-12-14 18:27:31 UTC

Allow injecting certificate for tests (#615)

Built on #609

When activating the `puffin-test-custom-ca-cert` feature, you can inject
a custom ssl certificate by setting `PUFFIN_TEST_CA_CERT_PEM` to a pem
file, e.g.

```bash
PUFFIN_TEST_CA_CERT_PEM=$(pwd)/mitmproxy-ca-cert.pem ./offlinepi record cargo test --features pypi --features puffin-test-custom-ca-cert -- --test-threads=1
```

This feature is off by default, so this is not possible in release
builds.

Author: Zanie
Author Date: 2023-12-12 23:16:21 UTC

Add handling for formatting `UnusableVersions` incompatibility

Author: Charlie Marsh
Author Date: 2023-11-10 19:33:53 UTC

Add an option to use Resolvo in lieu of PubGrub

Author: Andrew Gallant
Author Date: 2023-11-10 19:17:42 UTC

pep440-rs: switch Version::release to smallvec

This commit attempts an optimization that switches a version's `release`
field over to a `smallvec` optimization. The idea is that most versions
are very small and can be stored inline.

Interestingly, I was unable to observe any obvious benefit:

    $ hyperfine \
        "./target/profiling/puffin-dev-u32 resolve-many --cache-dir cache-docker-no-build --no-build pypi_top_8k_flat.txt --limit 1000 2> /dev/null" \
        "./target/profiling/puffin-dev-smallvec-release resolve-many --cache-dir cache-docker-no-build --no-build pypi_top_8k_flat.txt --limit 1000 2> /dev/null"
    Benchmark 1: ./target/profiling/puffin-dev-u32 resolve-many --cache-dir cache-docker-no-build --no-build pypi_top_8k_flat.txt --limit 1000 2> /dev/null
      Time (mean ± σ): 872.2 ms ± 26.5 ms [User: 14646.0 ms, System: 2516.0 ms]
      Range (min … max): 833.0 ms … 912.0 ms 10 runs

    Benchmark 2: ./target/profiling/puffin-dev-smallvec-release resolve-many --cache-dir cache-docker-no-build --no-build pypi_top_8k_flat.txt --limit 1000 2> /dev/null
      Time (mean ± σ): 882.3 ms ± 17.4 ms [User: 14764.4 ms, System: 2520.9 ms]
      Range (min … max): 859.7 ms … 912.7 ms 10 runs

    Summary
      './target/profiling/puffin-dev-u32 resolve-many --cache-dir cache-docker-no-build --no-build pypi_top_8k_flat.txt --limit 1000 2> /dev/null' ran
        1.01 ± 0.04 times faster than './target/profiling/puffin-dev-smallvec-release resolve-many --cache-dir cache-docker-no-build --no-build pypi_top_8k_flat.txt --limit 1000 2> /dev/null'

My hypothesis is that because of an earlier commit that switched the
global allocator to jemalloc, the cost of allocation had precipitously
decreased. To the point that the reduction in allocs from the smallvec
becomes a wash. To test my hypothesis, I dropped the jemalloc commit and
measured the perf of the smallvec optimization against main:

    $ hyperfine \
        "./target/profiling/puffin-dev-main resolve-many --cache-dir cache-docker-no-build --no-build pypi_top_8k_flat.txt --limit 1000 2> /dev/null" \
        "./target/profiling/puffin-dev-smallvec-release-no-jemalloc resolve-many --cache-dir cache-docker-no-build --no-build pypi_top_8k_flat.txt --limit 1000 2> /dev/null"
    Benchmark 1: ./target/profiling/puffin-dev-main resolve-many --cache-dir cache-docker-no-build --no-build pypi_top_8k_flat.txt --limit 1000 2> /dev/null
      Time (mean ± σ): 968.0 ms ± 20.0 ms [User: 17637.4 ms, System: 2151.9 ms]
      Range (min … max): 940.2 ms … 1005.3 ms 10 runs

    Benchmark 2: ./target/profiling/puffin-dev-smallvec-release-no-jemalloc resolve-many --cache-dir cache-docker-no-build --no-build pypi_top_8k_flat.txt --limit 1000 2> /dev/null
      Time (mean ± σ): 958.4 ms ± 15.7 ms [User: 17119.7 ms, System: 2246.1 ms]
      Range (min … max): 944.7 ms … 993.3 ms 10 runs

    Summary
      './target/profiling/puffin-dev-smallvec-release-no-jemalloc resolve-many --cache-dir cache-docker-no-build --no-build pypi_top_8k_flat.txt --limit 1000 2> /dev/null' ran
        1.01 ± 0.03 times faster than './target/profiling/puffin-dev-main resolve-many --cache-dir cache-docker-no-build --no-build pypi_top_8k_flat.txt --limit 1000 2> /dev/null'

Fiddlesticks. Even when allocation is (presumably) more expensive, the
smallvec optimization didn't help. This suggests something is off about
my mental model of the code. So there are more avenues to explore here!

Author: Zanie
Author Date: 2023-11-09 21:25:35 UTC

Update snapshots

Author: Zanie
Author Date: 2023-11-09 16:57:43 UTC

Happy clippy

Author: Zanie
Author Date: 2023-11-08 23:09:29 UTC

Switch to nightly rust for https://github.com/rust-lang/rust/issues/91611

Author: Zanie
Author Date: 2023-11-06 20:12:36 UTC

Improve error messages

Author: Zanie
Author Date: 2023-11-01 16:38:22 UTC

Add `PackageName::validate`