Skip to content

Commit

Permalink
perf: Quick prefix validation check (#2400)
Browse files Browse the repository at this point in the history
This pr will add a hash of the lock file to the `EnvironmentFile`
(`.pixi/envs/default/conda-meta/pixi`)

It will look like this:
```json
{
  "manifest_path": "/home/rarts/dev/pixi/pixi.toml",
  "environment_name": "default",
  "pixi_version": "0.34.0",
  "environment_lock_file_hash": "4f36ee620f10329d"
}
```
And that hash will be compared with the current lockfile on `pixi run`
`pixi shell` `pixi shell-hook`.

### Profile result

```shell
❯ hyperfine "pixi run echo" "old-pixi run echo"
Benchmark 1: pixi run echo
  Time (mean ± σ):     381.6 ms ±  22.1 ms    [User: 193.8 ms, System: 246.3 ms]
  Range (min … max):   344.5 ms … 414.3 ms    10 runs
 
Benchmark 2: old-pixi run echo
  Time (mean ± σ):     868.2 ms ±  58.2 ms    [User: 480.0 ms, System: 557.0 ms]
  Range (min … max):   791.1 ms … 950.8 ms    10 runs
 
Summary
  pixi run echo ran
    2.28 ± 0.20 times faster than old-pixi run echo
```
> [!NOTE]  
> The remaining `381ms` is the activation which is fixed by
#2367

### UX
- It's turned on by default
- You can request a re-validate on `pixi run/shell/shell-hook` with
`--revalidate`
- All commands designed to update the lock file or `pixi install` will
always re-validate.

### TODO:

- [x] : Add tests: chosen python integration tests as I was to fed up
with the extreme amount of time spent on writing tests in Rust
- [x] : use Enum instead of booleans. Using `UpdateMode::QuickValidate`
and `UpdateMode::Revalidate`
- [x] : Document behavior
- [x] : Extend logic to cli. `--revalidate`

---------

Co-authored-by: Hofer-Julian <30049909+Hofer-Julian@users.noreply.github.com>
  • Loading branch information
ruben-arts and Hofer-Julian authored Nov 6, 2024
1 parent c2ea77b commit d7230cf
Show file tree
Hide file tree
Showing 24 changed files with 436 additions and 93 deletions.
2 changes: 1 addition & 1 deletion docs/advanced/explain_info_command.md
Original file line number Diff line number Diff line change
Expand Up @@ -57,7 +57,7 @@ In that case, if pixi cannot find the `__cuda` virtual package on your machine t
### Cache dir

The directory where pixi stores its cache.
Checkout the [cache documentation](../features/environment.md#caching) for more information.
Checkout the [cache documentation](../features/environment.md#caching-packages) for more information.

### Auth storage

Expand Down
27 changes: 26 additions & 1 deletion docs/features/environment.md
Original file line number Diff line number Diff line change
Expand Up @@ -40,6 +40,31 @@ These directories are conda environments, and you can use them as such, but you
Pixi will always make sure the environment is in sync with the `pixi.lock` file.
If this is not the case then all the commands that use the environment will automatically update the environment, e.g. `pixi run`, `pixi shell`.

### Environment Installation Metadata
On environment installation, pixi will write a small file to the environment that contains some metadata about installation.
This file is called `pixi` and is located in the `conda-meta` folder of the environment.
This file contains the following information:
- `manifest_path`: The path to the manifest file that describes the project used to create this environment
- `environment_name`: The name of the environment
- `pixi_version`: The version of pixi that was used to create this environment
- `environment_lock_file_hash`: The hash of the `pixi.lock` file that was used to create this environment

```json
{
"manifest_path": "/home/user/dev/pixi/pixi.toml",
"environment_name": "default",
"pixi_version": "0.34.0",
"environment_lock_file_hash": "4f36ee620f10329d"
}
```

The `environment_lock_file_hash` is used to check if the environment is in sync with the `pixi.lock` file.
If the hash of the `pixi.lock` file is different from the hash in the `pixi` file, pixi will update the environment.

This is used to speedup activation, in order to trigger a full revalidation pass `--revalidate` to the `pixi run` or `pixi shell` command.
A broken environment would typically not be found with a hash comparison, but a revalidation would reinstall the environment.
By default, all lock file modifying commands will always use the revalidation and on `pixi install` it always revalidates.

### Cleaning up

If you want to clean up the environments, you can simply delete the `.pixi/envs` directory, and pixi will recreate the environments when needed.
Expand Down Expand Up @@ -186,7 +211,7 @@ For the `[pypi-dependencies]`, `uv` implements `sdist` building to retrieve the
For this building step, `pixi` requires to first install `python` in the (conda)`[dependencies]` section of the `pixi.toml` file.
This will always be slower than the pure conda solves. So for the best pixi experience you should stay within the `[dependencies]` section of the `pixi.toml` file.

## Caching
## Caching packages

Pixi caches all previously downloaded packages in a cache folder.
This cache folder is shared between all pixi projects and globally installed tools.
Expand Down
4 changes: 4 additions & 0 deletions docs/reference/cli.md
Original file line number Diff line number Diff line change
Expand Up @@ -208,6 +208,8 @@ You cannot run `pixi run source setup.bash` as `source` is not available in the
- `--locked`: only install if the `pixi.lock` is up-to-date with the [manifest file](project_configuration.md)[^1]. It can also be controlled by the `PIXI_LOCKED` environment variable (example: `PIXI_LOCKED=true`). Conflicts with `--frozen`.
- `--environment <ENVIRONMENT> (-e)`: The environment to run the task in, if none are provided the default environment will be used or a selector will be given to select the right environment.
- `--clean-env`: Run the task in a clean environment, this will remove all environment variables of the shell environment except for the ones pixi sets. THIS DOESN't WORK ON `Windows`.
- `--revalidate`: Revalidate the full environment, instead of checking the lock file hash. [more info](../features/environment.md#environment-installation-metadata)

```shell
pixi run python
pixi run cowpy "Hey pixi user"
Expand Down Expand Up @@ -636,6 +638,7 @@ To exit the pixi shell, simply run `exit`.
- `--frozen`: install the environment as defined in the lock file, doesn't update `pixi.lock` if it isn't up-to-date with [manifest file](project_configuration.md). It can also be controlled by the `PIXI_FROZEN` environment variable (example: `PIXI_FROZEN=true`).
- `--locked`: only install if the `pixi.lock` is up-to-date with the [manifest file](project_configuration.md)[^1]. It can also be controlled by the `PIXI_LOCKED` environment variable (example: `PIXI_LOCKED=true`). Conflicts with `--frozen`.
- `--environment <ENVIRONMENT> (-e)`: The environment to activate the shell in, if none are provided the default environment will be used or a selector will be given to select the right environment.
- `--revalidate`: Revalidate the full environment, instead of checking lock file hash. [more info](../features/environment.md#environment-installation-metadata)

```shell
pixi shell
Expand Down Expand Up @@ -664,6 +667,7 @@ This command prints the activation script of an environment.
- `--environment <ENVIRONMENT> (-e)`: The environment to activate, if none are provided the default environment will be used or a selector will be given to select the right environment.
- `--json`: Print all environment variables that are exported by running the activation script as JSON. When specifying
this option, `--shell` is ignored.
- `--revalidate`: Revalidate the full environment, instead of checking lock file hash. [more info](../features/environment.md#environment-installation-metadata)

```shell
pixi shell-hook
Expand Down
4 changes: 2 additions & 2 deletions src/cli/add.rs
Original file line number Diff line number Diff line change
Expand Up @@ -22,7 +22,7 @@ use crate::{
cli::cli_config::{DependencyConfig, PrefixUpdateConfig, ProjectConfig},
environment::verify_prefix_location_unchanged,
load_lock_file,
lock_file::{filter_lock_file, LockFileDerivedData, UpdateContext},
lock_file::{filter_lock_file, LockFileDerivedData, UpdateContext, UpdateMode},
project::{grouped_environment::GroupedEnvironment, DependencyType, Project},
};

Expand Down Expand Up @@ -294,7 +294,7 @@ pub async fn execute(args: Args) -> miette::Result<()> {
&& default_environment_is_affected
{
updated_lock_file
.prefix(&project.default_environment())
.prefix(&project.default_environment(), UpdateMode::Revalidate)
.await?;
}

Expand Down
14 changes: 14 additions & 0 deletions src/cli/cli_config.rs
Original file line number Diff line number Diff line change
@@ -1,5 +1,6 @@
use crate::cli::has_specs::HasSpecs;
use crate::environment::LockFileUsage;
use crate::lock_file::UpdateMode;
use crate::DependencyType;
use crate::Project;
use clap::Parser;
Expand Down Expand Up @@ -98,6 +99,10 @@ pub struct PrefixUpdateConfig {

#[clap(flatten)]
pub config: ConfigCli,

/// Run the complete environment validation. This will reinstall a broken environment.
#[arg(long)]
pub revalidate: bool,
}
impl PrefixUpdateConfig {
pub fn lock_file_usage(&self) -> LockFileUsage {
Expand All @@ -114,6 +119,15 @@ impl PrefixUpdateConfig {
pub(crate) fn no_install(&self) -> bool {
self.no_install || self.no_lockfile_update
}

/// Which `[UpdateMode]` to use
pub(crate) fn update_mode(&self) -> UpdateMode {
if self.revalidate {
UpdateMode::Revalidate
} else {
UpdateMode::QuickValidate
}
}
}
#[derive(Parser, Debug, Default)]
pub struct DependencyConfig {
Expand Down
9 changes: 8 additions & 1 deletion src/cli/install.rs
Original file line number Diff line number Diff line change
@@ -1,5 +1,6 @@
use crate::cli::cli_config::ProjectConfig;
use crate::environment::update_prefix;
use crate::lock_file::UpdateMode;
use crate::Project;
use clap::Parser;
use fancy_display::FancyDisplay;
Expand Down Expand Up @@ -52,7 +53,13 @@ pub async fn execute(args: Args) -> miette::Result<()> {
let environment = project.environment_from_name_or_env_var(Some(env))?;

// Update the prefix by installing all packages
update_prefix(&environment, args.lock_file_usage.into(), false).await?;
update_prefix(
&environment,
args.lock_file_usage.into(),
false,
UpdateMode::Revalidate,
)
.await?;

installed_envs.push(environment.name().clone());
}
Expand Down
2 changes: 2 additions & 0 deletions src/cli/project/channel/add.rs
Original file line number Diff line number Diff line change
@@ -1,5 +1,6 @@
use crate::{
environment::{update_prefix, LockFileUsage},
lock_file::UpdateMode,
Project,
};

Expand All @@ -16,6 +17,7 @@ pub async fn execute(mut project: Project, args: AddRemoveArgs) -> miette::Resul
&project.default_environment(),
LockFileUsage::Update,
args.no_install,
UpdateMode::Revalidate,
)
.await?;
project.save()?;
Expand Down
2 changes: 2 additions & 0 deletions src/cli/project/channel/remove.rs
Original file line number Diff line number Diff line change
@@ -1,3 +1,4 @@
use crate::lock_file::UpdateMode;
use crate::{
environment::{update_prefix, LockFileUsage},
Project,
Expand All @@ -16,6 +17,7 @@ pub async fn execute(mut project: Project, args: AddRemoveArgs) -> miette::Resul
&project.default_environment(),
LockFileUsage::Update,
args.no_install,
UpdateMode::Revalidate,
)
.await?;
project.save()?;
Expand Down
2 changes: 2 additions & 0 deletions src/cli/project/platform/add.rs
Original file line number Diff line number Diff line change
Expand Up @@ -2,6 +2,7 @@ use std::str::FromStr;

use crate::{
environment::{update_prefix, LockFileUsage},
lock_file::UpdateMode,
Project,
};
use clap::Parser;
Expand Down Expand Up @@ -48,6 +49,7 @@ pub async fn execute(mut project: Project, args: Args) -> miette::Result<()> {
&project.default_environment(),
LockFileUsage::Update,
args.no_install,
UpdateMode::Revalidate,
)
.await?;
project.save()?;
Expand Down
2 changes: 2 additions & 0 deletions src/cli/project/platform/remove.rs
Original file line number Diff line number Diff line change
@@ -1,5 +1,6 @@
use std::str::FromStr;

use crate::lock_file::UpdateMode;
use crate::{
environment::{update_prefix, LockFileUsage},
Project,
Expand Down Expand Up @@ -47,6 +48,7 @@ pub async fn execute(mut project: Project, args: Args) -> miette::Result<()> {
&project.default_environment(),
LockFileUsage::Update,
args.no_install,
UpdateMode::Revalidate,
)
.await?;
project.save()?;
Expand Down
2 changes: 2 additions & 0 deletions src/cli/remove.rs
Original file line number Diff line number Diff line change
Expand Up @@ -6,6 +6,7 @@ use crate::DependencyType;
use crate::Project;

use crate::cli::cli_config::{DependencyConfig, PrefixUpdateConfig, ProjectConfig};
use crate::lock_file::UpdateMode;

use super::has_specs::HasSpecs;

Expand Down Expand Up @@ -82,6 +83,7 @@ pub async fn execute(args: Args) -> miette::Result<()> {
&project.default_environment(),
prefix_update_config.lock_file_usage(),
prefix_update_config.no_install,
UpdateMode::Revalidate,
)
.await?;
}
Expand Down
16 changes: 10 additions & 6 deletions src/cli/run.rs
Original file line number Diff line number Diff line change
@@ -1,12 +1,11 @@
use std::collections::hash_map::Entry;
use std::collections::HashSet;
use std::convert::identity;
use std::{collections::HashMap, string::String};

use clap::Parser;
use dialoguer::theme::ColorfulTheme;
use itertools::Itertools;
use miette::{Diagnostic, IntoDiagnostic};
use std::collections::hash_map::Entry;
use std::collections::HashSet;
use std::convert::identity;
use std::{collections::HashMap, string::String};

use crate::cli::cli_config::{PrefixUpdateConfig, ProjectConfig};
use crate::environment::verify_prefix_location_unchanged;
Expand Down Expand Up @@ -169,7 +168,12 @@ pub async fn execute(args: Args) -> miette::Result<()> {
Entry::Occupied(env) => env.into_mut(),
Entry::Vacant(entry) => {
// Ensure there is a valid prefix
lock_file.prefix(&executable_task.run_environment).await?;
lock_file
.prefix(
&executable_task.run_environment,
args.prefix_update_config.update_mode(),
)
.await?;

let command_env = get_task_env(
&executable_task.run_environment,
Expand Down
2 changes: 2 additions & 0 deletions src/cli/shell.rs
Original file line number Diff line number Diff line change
Expand Up @@ -9,6 +9,7 @@ use rattler_shell::{
};

use crate::cli::cli_config::{PrefixUpdateConfig, ProjectConfig};
use crate::lock_file::UpdateMode;
use crate::{
activation::CurrentEnvVarBehavior, environment::update_prefix,
project::virtual_packages::verify_current_platform_has_required_virtual_packages, prompt,
Expand Down Expand Up @@ -241,6 +242,7 @@ pub async fn execute(args: Args) -> miette::Result<()> {
&environment,
args.prefix_update_config.lock_file_usage(),
false,
UpdateMode::QuickValidate,
)
.await?;

Expand Down
11 changes: 6 additions & 5 deletions src/cli/shell_hook.rs
Original file line number Diff line number Diff line change
Expand Up @@ -10,12 +10,12 @@ use rattler_shell::{
use serde::Serialize;
use serde_json;

use crate::cli::cli_config::{PrefixUpdateConfig, ProjectConfig};
use crate::project::HasProjectRef;
use crate::activation::CurrentEnvVarBehavior;
use crate::environment::update_prefix;
use crate::{
activation::{get_activator, CurrentEnvVarBehavior},
environment::update_prefix,
project::Environment,
activation::get_activator,
cli::cli_config::{PrefixUpdateConfig, ProjectConfig},
project::{Environment, HasProjectRef},
Project,
};

Expand Down Expand Up @@ -113,6 +113,7 @@ pub async fn execute(args: Args) -> miette::Result<()> {
&environment,
args.prefix_update_config.lock_file_usage(),
false,
args.prefix_update_config.update_mode(),
)
.await?;

Expand Down
Loading

0 comments on commit d7230cf

Please sign in to comment.