This is another experiment for #9930.
Comparing preferring over exclusively using MSRV compatible:
Benefits
- Better error messages
- `--ignore-rust-version` is implicitly sticky
Downsides
- Can't backtrack for MSRV compatible version
- Still requires workspace-wide MSRV (compared to our desired end state of declaring MSRV as yet another dependency)
This builds on #12930
Before, when running `cargo update <member>`, we'd not reuse the
previous resolve result and when the resolver started walking into the
dependencies, it would do a git fetch.
Now, we won't even try to resolve the workspace members and so we won't
look at those dependencies and do git fetch.
This will make `cargo update <workspace-member>`
match `cargo update --workspace`.
I considered whether there were other ways of handling this but I
figured aiming for consistency in approaches was the best way.
We can investigate improving those approaches separately.
There are other discrepancies in the different code paths (handling of
patches, adding sources) but I'm deferring looking over those.
Between this and #12602, this should finnally resolve#12599.
Fixes#12599
Only filter out target if its in the package root
### What does this PR try to resolve?
Only filter out target if its in the package root. Fixed https://github.com/rust-lang/cargo/issues/12790
### How should we test and review this PR?
Add two testcase in tests/testsuite/package.rs for this PR.
- `include_files_called_target_project` testcase test the logic for none git repository. By the way, our PR was based on git repository, so this testcase was a little irrelevant.
- `include_files_called_target_git` testcase was made for git repository. There are two cases here, one is the target in the uncommitted state, at this time should not be included, and one is the target in the committed state, at this time should be included
### Additional information
This fixes an issue where `--quiet` doesn't work with commands that have
subcommands. This is because `config_configure` only looks at the global
and top-level subcommand, and not deeper subcommands. The issue was that
`--quiet` was not defined as a global flag. This was changed in
https://github.com/rust-lang/cargo/pull/6358 in order to give a better
help message for `cargo test --quiet`. I don't remember if clap just
didn't support overriding at the time, or if we just didn't know how it
worked. Anyways, it seems to work to override it now, so I think it
should be fine to mark it as global.
This should bring in `--quiet` more in-line with how `--verbose` works.
This means that `--quiet` is now accepted with `cargo report`,
`cargo help`, and `cargo config`.
This also fixes `--quiet` with `cargo clean gc`.
This should also help with supporting `--quiet` with the new `cargo
owner` subcommands being added in
https://github.com/rust-lang/cargo/pull/11879.
Fixes#12957
This was requested to separate the interaction of `cargo clean` and the
cleaning of global cache data, and to minimize the impact of this
initial implementation.
This adds a garbage collector which will remove old files from cargo's
global cache.
A general overview of the changes here:
- `cargo::core::global_cache_tracker` contains the `GlobalCacheTracker`
which handles the interface to a sqlite database which stores
timestamps of the last time a file was used.
- `DeferredGlobalLastUse` is a type that implements an optimization for
collecting last-use timestamps so that they can be flushed to disk all
at once.
- `cargo::core::gc` contains the `Gc` type which is the interface for
performing garbage collection. It coordinates with the
`GlobalCacheTracker` for determining what to delete.
- Garbage collection can either be automatic or manual. The automatic
garbage collection supports some config options for defining when
it runs and how much it deletes.
- Manual garbage collection can be performed via options to `cargo
clean`.
- `cargo clean` uses the new package cache locking system to coordinate
access to the package cache to prevent interference with other cargo
commands running concurrently.
For `cargo install` we'll now show a more specific parse error for
semver, much like other parts of cargo.
This came out of my work on #12801. I was looking at what might be
appropriate to put in a `cargo-util-semver` crate and realized we have
the `ToSemver` trait that exists but doesn't do much, so I dropped it.
refactor(toml): Pull out the schema
### What does this PR try to resolve?
On its own, this PR is a net negative for readability / complexity. It moves all of the serde types related to manifest from `toml/mod.rs` to `toml/schema.rs`, leaving a lot of the functions and some trait implementations back in `toml/mod.rs` (some basic functions that made sense for the type on their own were also moved).
So why do this? This is an ooch towards having the schema broken out into a separate package (#12801). To do this, we need to
- Identify what all types need to be put in the package. This refactor highlights the dependence on `RustVersion` and `PackageIdSpec`
- Highlights what functionality we need to find a new home for
Follow up PRs would
- Find better homes for the logic in `toml/mod.rs`, moving us away from having `impl schema::Type` blocks.
- Pull out a `src/cargo/util_semver` package to own `PartialVersion` (at least) as prep for making a `cargo-util-semver` package
- Move `RustVersion` to `manifest-toml/schema.rs`, deciding what functionality needs to move with the type
- Move or copy `PackageIdSpec` into `manfest-toml/schema.rs`, deciding what functionality remain in `core/` and what moves over
- Move `toml/schema.rs` to `src/cargo/util_schema`
- Actually make `cargo-util-semver` and `cargo-util-manifest-schema` packages
So why do this now? This is a big change! By being incremental
- Reduce churn for me and others
- Be easier to review
- Collect feedback as we go on the whole plan to avoid more painful changes later
We *can* back this out if needed but the further we go, the more painful it will be.
### How should we test and review this PR?
### Additional information
fix(cli): Clarify --test is for targets, not test functions
We already refer to test targets as "test targets" instead of "tests" in `--test` but not `--tests` or in the error output. This makes it uniformly refer to them as "test targets", making it clearer that these aren't test functions.
Fixes#7864
We already refer to test targets as "test targets" instead of "tests" in
`--test` but not `--tests` or in the error output. This makes it
uniformly refer to them as "test targets", making it clearer that these
aren't test functions.
Fixes#7864
Add new packages to [workspace.members] automatically
### What does this PR try to resolve?
If a new package is created in a workspace, this change adds the package's path to the workspace's members list automatically.
It doesn't add the package to the list if the path is in the workspace's exclude list, or if the members list doesn't exist already. I noticed that a `cargo_new` test broke if I create the members list when it doesn't exist. This is because the workspace's manifest can be used for package templating. I think it's better to not break that use case.
Fixes#6378
### How should we test and review this PR?
I've included tests in the `cargo_new` suite.
When a user runs `cargo new` or `cargo init` within a workspace, Cargo will automatically add the new package to the members list in the workspace if necessary. The heuristic to add the new package is as follows:
- If there is no `members` list in the workspace yet, a new `members` list is created.
- If there is an `exclude` statement, Cargo checks if the new package should be excluded. If it doesn't match the `exclude` list, the package is added to the `members` list.
- If there is a glob expression in the `members` list that matches the new package, the package is not added to the `members` list.
- If the existent `members` list is sorted, Cargo tries to preserve the ordering when it adds the new package.
This change doesn't try to format the resulting `members` list in any way, leaving the formatting decissions to the user.
Signed-off-by: David Calavera <david.calavera@gmail.com>
To have a separate manifest API (#12801), we can't rely on interning
because it might be used in longer-lifetime applications.
I consulted https://github.com/rosetta-rs/string-rosetta-rs when
deciding on what string type to use for performance.
Originally, I hoped to entirely replacing string interning. For that, I
was looking at `arcstr` as it had a fast equality operator. However,
that is only helpful so long as the two strings we are comparing came
from the original source. Unsure how likely that is to happen (and
daunted by converting all of the `Copy`s into `Clone`s), I decided to
scale back.
Concerned about all of the small allocations when parsing a manifest, I
assumed I'd need a string type with small-string optimizations, like
`hipstr`, `compact_str`, `flexstr`, and `ecow`.
The first three give us more wiggle room and `hipstr` was the fastest of
them, so I went with that.
I then double checked macro benchmarks, and realized `hipstr` made no
difference and switched to `String` to keep things simple / with lower
dependencies.
When doing this, I had created a type alias (`TomlStr`) for the string
type so I could more easily swap it out if needed
(and not have to always write out a lifetime).
With just using `String`, I went ahead and dropped that.
I had problems getting the cargo benchmarks running, so I did a quick
and dirty benchmark that is end-to-end, covering fresh builds, clean
builds, and resolution. I ran these against a fresh clone of cargo's
code base.
```console
$ ../dump/cargo-12801-bench.rs run
Finished dev [unoptimized + debuginfo] target(s) in 0.07s
Running `target/debug/cargo -Zscript -Zmsrv-policy ../dump/cargo-12801-bench.rs run`
warning: `package.edition` is unspecified, defaulting to `2021`
Finished dev [unoptimized + debuginfo] target(s) in 0.04s
Running `/home/epage/.cargo/target/0a/7f4c1ab500f045/debug/cargo-12801-bench run`
$ hyperfine "../cargo-old check" "../cargo-new check"
Benchmark 1: ../cargo-old check
Time (mean ± σ): 119.3 ms ± 3.2 ms [User: 98.6 ms, System: 20.3 ms]
Range (min … max): 115.6 ms … 124.3 ms 24 runs
Benchmark 2: ../cargo-new check
Time (mean ± σ): 119.4 ms ± 2.4 ms [User: 98.0 ms, System: 21.1 ms]
Range (min … max): 115.7 ms … 123.6 ms 24 runs
Summary
../cargo-old check ran
1.00 ± 0.03 times faster than ../cargo-new check
$ hyperfine --prepare "cargo clean" "../cargo-old check" "../cargo-new check"
Benchmark 1: ../cargo-old check
Time (mean ± σ): 20.249 s ± 0.392 s [User: 157.719 s, System: 22.771 s]
Range (min … max): 19.605 s … 21.123 s 10 runs
Benchmark 2: ../cargo-new check
Time (mean ± σ): 20.123 s ± 0.212 s [User: 156.156 s, System: 22.325 s]
Range (min … max): 19.764 s … 20.420 s 10 runs
Summary
../cargo-new check ran
1.01 ± 0.02 times faster than ../cargo-old check
$ hyperfine --prepare "cargo clean && rm -f Cargo.lock" "../cargo-old check" "../cargo-new check"
Benchmark 1: ../cargo-old check
Time (mean ± σ): 21.105 s ± 0.465 s [User: 156.482 s, System: 22.799 s]
Range (min … max): 20.156 s … 22.010 s 10 runs
Benchmark 2: ../cargo-new check
Time (mean ± σ): 21.358 s ± 0.538 s [User: 156.187 s, System: 22.979 s]
Range (min … max): 20.703 s … 22.462 s 10 runs
Summary
../cargo-old check ran
1.01 ± 0.03 times faster than ../cargo-new check
```
I've wanted something like this myself. I dislike using `--open`
because I tend to move up to re-run my `cargo doc` run but then have to
edit it to remove `--open`.
Also makes it annoying when opening docs when `cargo doc` is wrapped by
a tool like `make`.
This was previously attempted in #5592:
- Unlike the request in #5562, this aligns with #5592 in always printing
rather than using a flag as this seems generally useful
- Unlike #5592, this prints as an alternative to "Opening" to keep
things light
- Unlike #5592, this prints afterwards as the link is only valid then
Fixes#5562
feat(toml): Allow version-less manifests
### What does this PR try to resolve?
Expected behavior with this PR:
- `package.version` defaults to `0.0.0`
- `package.publish` is defaulted to `version.is_some()`
This also updates "cargo script" to rely on this new behavior.
My motivation is to find ways to close the gap between "cargo script" and `Cargo.toml`. With "cargo script", we want to allow people to only write however much of a manifest is directly needed for the work they are doing (which includes having no manifest). Each difference between "cargo script" and `Cargo.toml` is a cost we have to pay in our documentation and a hurdle in a users understanding of what is happening.
There has been other interest in this which I also find of interest (from #9829):
- Lower boilerplate, whether for [cargo xtasks](https://github.com/matklad/cargo-xtask), nested packages (rust-lang/rfcs#3452), etc
- Unmet expectations from users because this field is primarily targeted at registry operations when they want it for their marketing version (#6583).
- Make "unpublished" packages stand out
This then unblocks unifying `package.publish` by making the field's default based on the presence of a version as inspired by the proposal in #9829. Without this change, we were trading one form of boilerplate (`version = "0.0.0"`) for another (`publish = false`).
Fixes#9829Fixes#12690Fixes#6153
### How should we test and review this PR?
The initial commit has test cases I thought would be relevant for this change and you can see how each commit affects those or existing test cases. Would definitely be interested in hearing of other troubling cases to test
Implementation wise, I made `MaybeWorkspaceVersion` deserializer trim spaces so I could more easily handle the field being an `Option`. This is in its own commit.
### Additional information
Alternatives considered
- Making the default version "stand out more" with it being something like `0.0.0+HEAD`. The extra noise didn't seem worth it and people would contend over what the metadata field *should be*
- Make the default version the lowest version possible (`0.0.0-0`?). Unsure if this will ever really matter especially since you can't publish
- Defer defaulting `package.publish` and instead error
- Further unifying more fields made this too compelling for me :)
- Put this behind `-Zscript` and make it a part of rust-lang/rfcs#3502
- Having an affect outside of that RFC, I wanted to make sure this got the attention it deserved rather than getting lost in the noise of a large RFC.
- Don't just default the version but make packages versionless
- I extended the concept of versionless to `PackageId`'s internals and saw no observable difference
- I then started to examine the idea of version being optional everywhere (via `PackageId`s API) and ... things got messy especially when starting to look at the resolver. This would have also added a lot of error checks / asserts for "the version must be set here". Overall, the gains seemed questionable and the cost high, so I held off.
