<!--
Thank you for contributing to uv! To help us out with reviewing, please
consider the following:
- Does this pull request include a summary of the change? (See below.)
- Does this pull request include a descriptive title?
- Does this pull request include references to any relevant issues?
-->
## Summary
Separate exceptions for different timeouts to make it easier to debug
issues like #6105.
<!-- What's the purpose of the change? What does it do, and why? -->
## Test Plan
<!-- How was it tested? -->
Not tested at all.
## Summary
Avoid using work-stealing Tokio workers for bytecode compilation,
favoring instead dedicated threads. Tokio's work-stealing does not
really benefit us because we're spawning Python workers and scheduling
tasks ourselves — we don't want Tokio to re-balance our workers. Because
we're doing scheduling ourselves and compilation is a primarily
compute-bound task, we can also create dedicated runtimes for each
worker and avoid some synchronization overhead.
This is part of a general desire to avoid relying on Tokio's
work-stealing scheduler and be smarter about our workload. In this case
we already had the custom scheduler in place, Tokio was just getting in
the way (though the overhead is very minor).
## Test Plan
This improves performance by ~5% on my machine.
```
$ hyperfine --warmup 1 --prepare "target/profiling/uv-dev clear-compile .venv" "target/profiling/uv-dev compile .venv" "target/profiling/uv-dev-dedicated compile .venv"
Benchmark 1: target/profiling/uv-dev compile .venv
Time (mean ± σ): 1.279 s ± 0.011 s [User: 13.803 s, System: 2.998 s]
Range (min … max): 1.261 s … 1.296 s 10 runs
Benchmark 2: target/profiling/uv-dev-dedicated compile .venv
Time (mean ± σ): 1.220 s ± 0.021 s [User: 13.997 s, System: 3.330 s]
Range (min … max): 1.198 s … 1.272 s 10 runs
Summary
target/profiling/uv-dev-dedicated compile .venv ran
1.05 ± 0.02 times faster than target/profiling/uv-dev compile .venv
$ hyperfine --warmup 1 --prepare "target/profiling/uv-dev clear-compile .venv" "target/profiling/uv-dev compile .venv" "target/profiling/uv-dev-dedicated compile .venv"
Benchmark 1: target/profiling/uv-dev compile .venv
Time (mean ± σ): 3.631 s ± 0.078 s [User: 47.205 s, System: 4.996 s]
Range (min … max): 3.564 s … 3.832 s 10 runs
Benchmark 2: target/profiling/uv-dev-dedicated compile .venv
Time (mean ± σ): 3.521 s ± 0.024 s [User: 48.201 s, System: 5.392 s]
Range (min … max): 3.484 s … 3.566 s 10 runs
Summary
target/profiling/uv-dev-dedicated compile .venv ran
1.03 ± 0.02 times faster than target/profiling/uv-dev compile .venv
```
## Summary
This PR changes our user-facing representation for paths to use relative
paths, when the path is within the current working directory. This
mirrors what we do in Ruff. (If the path is _outside_ the current
working directory, we print an absolute path.)
Before:
```shell
❯ uv venv .venv2
Using Python 3.12.2 interpreter at: /Users/crmarsh/workspace/uv/.venv/bin/python3
Creating virtualenv at: .venv2
Activate with: source .venv2/bin/activate
```
After:
```shell
❯ cargo run venv .venv2
Finished dev [unoptimized + debuginfo] target(s) in 0.15s
Running `target/debug/uv venv .venv2`
Using Python 3.12.2 interpreter at: .venv/bin/python3
Creating virtualenv at: .venv2
Activate with: source .venv2/bin/activate
```
Note that we still want to use the existing `.simplified_display()`
anywhere that the path is being simplified, but _still_ intended for
machine consumption (e.g., when passing to `.current_dir()`).
Since Python 3.7, deterministic pycs are possible (see [PEP
552](https://peps.python.org/pep-0552/))
To select the bytecode invalidation mode explicitly by env var:
PYC_INVALIDATION_MODE=UNCHECKED_HASH uv pip install --compile ...
Valid values are TIMESTAMP (default), CHECKED_HASH, and UNCHECKED_HASH.
The latter options are useful for reproducible builds.
---------
Co-authored-by: konstin <konstin@mailbox.org>
Sometimes, the first time we read from the stdout of the bytecode
compiler python subprocess, we get an empty string back (no newline). If
we try to write to stdin, it will often be a broken pipe (#2245). After
we got an empty string the first time, we will get the same empty string
if we read a line again.
The details of this behavior are mysterious to me, but it seems that it
can be identified by the first empty string. We check by inserting
starting with a `Ready` message on the Python side. When we encounter
the broken state, we discard the interpreter and try again.
We have to introduce a third timeout check for the interpreter launch
itself.
Minimized test script:
```bash
#!/usr/bin/env bash
set -euo pipefail
while true; do
date --iso-8601=seconds # Progress indicator
rm -rf testenv
target/profiling/uv venv testenv -q --python 3.12
VIRTUAL_ENV=$PWD/testenv target/profiling/uv pip install -q --compile wheel==0.42.0
done
```
Run as
```
cargo build --profile profiling && bash compile_bug.sh
```
Fixes#2245
Follow-up to #2086: Don't use timeouts for the entire workers, but only
for the section that's about communicating with the (potentially broken)
`python` subprocess. I've also raised the timeout to 60s.
Add a `--compile` option to `pip install` and `pip sync`.
I chose to implement this as a separate pass over the entire venv. If we
wanted to compile during installation, we'd have to make sure that
writing is exclusive, to avoid concurrent processes writing broken
`.pyc` files. Additionally, this ensures that the entire site-packages
are bytecode compiled, even if there are packages that aren't from this
`uv` invocation. The disadvantage is that we do not update RECORD and
rely on this comment from [PEP 491](https://peps.python.org/pep-0491/):
> Uninstallers should be smart enough to remove .pyc even if it is not
mentioned in RECORD.
If this is a problem we can change it to run during installation and
write RECORD entries.
Internally, this is implemented as an async work-stealing subprocess
worker pool. The producer is a directory traversal over site-packages,
sending each `.py` file to a bounded async FIFO queue/channel. Each
worker has a long-running python process. It pops the queue to get a
single path (or exists if the channel is closed), then sends it to
stdin, waits until it's informed that the compilation is done through a
line on stdout, and repeat. This is fast, e.g. installing `jupyter
plotly` on Python 3.12 it processes 15876 files in 319ms with 32 threads
(vs. 3.8s with a single core). The python processes internally calls
`compileall.compile_file`, the same as pip.
Like pip, we ignore and silence all compilation errors
(https://github.com/astral-sh/uv/issues/1559). There is a 10s timeout to
handle the case when the workers got stuck. For the reviewers, please
check if i missed any spots where we could deadlock, this is the hardest
part of this PR.
I've added `uv-dev compile <dir>` and `uv-dev clear-compile <dir>`
commands, mainly for my own benchmarking. I don't want to expose them in
`uv`, they almost certainly not the correct workflow and we don't want
to support them.
Fixes#1788Closes#1559Closes#1928
Michal Čihař
Charlie Marsh
Ibraheem Ahmed
wim glenn
konsti