bors [Tue, 16 Mar 2021 20:55:20 +0000 (20:55 +0000)]
Auto merge of #9275 - ehuss:features-non-member, r=alexcrichton
Fix --feature pkg/feat for V1 resolver for non-member.
#8997 had an unintended regression where `-p foo --feature foo/feat` syntax where `foo` is an **optional non-member** fails with an error that `foo` did not match any packages. The issue is that the member/feature selection routine needed to slot this into the features for the package in the current working directory (it was incorrectly treating `foo` as a workspace member).
V2 outright does not allow specifying features for non-workspace members.
bors [Tue, 16 Mar 2021 18:23:15 +0000 (18:23 +0000)]
Auto merge of #9276 - ehuss:fix-collision-root-removal, r=alexcrichton
Fix doc duplicate removal of root units.
The doc collision removal code didn't consider the possibility that there was a collision with a root unit. This caused problems in conjunction with #9142 where cargo would panic because a root unit got removed from the graph because of a filename collision. The solution here is to avoid removing root units during the doc collision sweep.
This has a downside that this reintroduces a filename collision.
An alternate solution would be to make the set of root units mutable, and remove from that list, but I figured this is simpler and more conservative.
bors [Thu, 11 Mar 2021 19:57:49 +0000 (19:57 +0000)]
Auto merge of #9255 - ehuss:fix-deps-filtering, r=Eh2406
Fix issue with filtering exclusive target dependencies.
#8777 incorrectly changed the filtering logic for dependencies. Essentially it split `filter(any(A && B && C && D))` into two parts `filter(any(A && B)).filter(any(C && D))` which doesn't have the same meaning. The solution here is to pass a closure so that the conditions are joined again.
bors [Mon, 22 Feb 2021 22:05:24 +0000 (22:05 +0000)]
Auto merge of #9196 - ehuss:backports, r=alexcrichton
[beta] backports for 1.51
Beta backports for the following:
* Fix panic with doc collision orphan. (#9142)
* This is an important regression that is fairly easy to hit.
* Do not exit prematurely if anything failed installing. (#9185)
* This is not a regression, but I think an important fix.
* Add schema field to the index (#9161)
* This is only the first commit from the PR which checks for the `v` field in the index, and skips entries that are not understood. The reason to backport is to get this in as early as possible so that if we do decide to start using it in the future, it works as early as possible. This otherwise doesn't do anything, so I think it should be safe.
* Fix warnings of the new non_fmt_panic lint (#9148)
* Fixes CI for a new warning in nightly.
bors [Fri, 5 Feb 2021 18:11:29 +0000 (18:11 +0000)]
Auto merge of #9142 - ehuss:fix-doc-orphan, r=Eh2406
Fix panic with doc collision orphan.
As I feared, the collision removal added in #9077 caused problems due to orphans left in the unit graph. Ironically, it was the collision warning detection code which failed, as it iterates over all the keys of the graph.
The solution is to remove all orphans from the graph after collisions have been removed.
bors [Wed, 3 Feb 2021 22:50:15 +0000 (22:50 +0000)]
Auto merge of #9112 - alexcrichton:split-debuginfo, r=ehuss
Add split-debuginfo profile option
This commit adds a new `split-debuginfo` option to Cargo compilation
profiles which gets forwarded to the `-Csplit-debuginfo` codegen option
in rustc. This commit also sets the default, only on macOS, to be
`-Csplit-debuginfo=unpacked`. The purpose of this change is to leverage
rust-lang/rust#79570 to avoid running `dsymutil` on incremental builds
while also preserving a pleasant debugging experience by default. This
should lead to much faster incremental build times on macOS since
`dsymutil` isn't exactly the speediest tool in the world.
This is technically a breaking change in Cargo because we're no longer
by-default producing the `*.dSYM` folders on macOS. If those are still
desired, however, authors can always run `dsymutil` themselves or
otherwise configure `split-debuginfo = 'packed'` in their
manifest/profile configuration.
bors [Wed, 3 Feb 2021 15:56:05 +0000 (15:56 +0000)]
Auto merge of #9126 - ehuss:registry-builder, r=alexcrichton
Add RegistryBuilder for tests, and update crates-io error handling.
This adds `RegistryBuilder` to the test suite to make it more flexible to create different registry setups, and to reuse code a little more easily.
This also makes a small adjustment to the registry API to add a `ResponseError` type to make it easier to work with API errors. As part of this, some tests were added to validate the API behavior for response errors. There are only a few very small changes here:
* Extra newlines are removed from the headers printed in the error message.
* The UTF-8 error now also includes the text "invalid response from server".
* The "file too large" crates.io publish error now displays the tarball size. (There is no test for this because it is only issued for talking to `crates.io`.)
bors [Wed, 3 Feb 2021 02:09:41 +0000 (02:09 +0000)]
Auto merge of #9122 - ehuss:fix-multiple-run-custom-build, r=alexcrichton
Fix env/cfg set for `cargo test` and `cargo run`.
There are some situations where a build script may need to run multiple times for the same package during the same `cargo` session. There was a bug in that some of the values in the `Compilation` struct didn't handle this case. The solution here is to be more careful about how this extra data is associated with `Unit`s, instead of assuming a package's build script only runs once.
The things that were not being handled properly:
* `Compilation::extra_env`, which is the output of `cargo:rustc-env` in build scripts. The solution here is to use the `Metadata` hash which is used elsewhere for distinguishing build script outputs.
* `Compilation::cfgs`, which is the output of `cargo:rustc-cfg` in build scripts and the features to be set, and this is only used for doctests. The solution here is to just add those `--cfg` flags directly in the `Doctest` struct.
The situations that cause a build script to be run multiple times:
* A package needed for both the host and target. In the test here, this was accomplished with a proc-macro (which has to be `host`) and a cyclical dev dependency, but there are many other ways to trigger this.
* Something built with different features (with the new feature resolver), usually due to host/target differences.
* Something built with different profile settings, usually due to host/target differences.
Alex Crichton [Mon, 1 Feb 2021 18:39:36 +0000 (10:39 -0800)]
Cache failures in the rustc info cache
This commit updates the rustc info cache to cache failures to execute
rustc as well as successes. This fixes a weird issue where if you're
probing for flags the `rustc_info_cache` test fails on channels which
don't have the flag since previously a failure to execute rustc resulted
in never caching the result.
Alex Crichton [Fri, 29 Jan 2021 21:24:56 +0000 (13:24 -0800)]
Add split-debuginfo profile option
This commit adds a new `split-debuginfo` option to Cargo compilation
profiles which gets forwarded to the `-Csplit-debuginfo` codegen option
in rustc. This commit also sets the default, only on macOS, to be
`-Csplit-debuginfo=unpacked`. The purpose of this change is to leverage
rust-lang/rust#79570 to avoid running `dsymutil` on incremental builds
while also preserving a pleasant debugging experience by default. This
should lead to much faster incremental build times on macOS since
`dsymutil` isn't exactly the speediest tool in the world.
This is technically a breaking change in Cargo because we're no longer
by-default producing the `*.dSYM` folders on macOS. If those are still
desired, however, authors can always run `dsymutil` themselves or
otherwise configure `split-debuginfo = 'packed'` in their
manifest/profile configuration.
bors [Mon, 1 Feb 2021 16:24:34 +0000 (16:24 +0000)]
Auto merge of #9108 - CPerezz:locked_warn, r=alexcrichton
Impl warn for locked install without Cargo.lock
If we're installing in --locked mode and there's no `Cargo.lock` published
ie. the bin was published before https://github.com/rust-lang/cargo/pull/7026
the cargo install errors were not stating that it was due to the lack of
the `Cargo.lock` file. Instead, the error seemed completely unrelated.
Therefore, this tries to address this by adding a warn in the stderr
output.
Closes #9106
I will need some help on the testing side (assuming the code I added for the warning is correct).
It looks to me that the publish function implemented for testing purposes does not publish `Cargo.lock` which is the actual convention. Should this be updated too? See #7026
bors [Mon, 1 Feb 2021 15:06:07 +0000 (15:06 +0000)]
Auto merge of #9120 - dimo414:patch-1, r=alexcrichton
Flip 'foo' and 'bar' to be consistent
The "Renaming dependencies" section initially uses 'foo' as the crate name and 'bar' as a rename, but then swaps them and uses 'bar' as the example crate name in the context of optional dependencies. Now both examples in this section treat 'foo' as the original crate name.
Michael Diamond [Mon, 1 Feb 2021 07:27:56 +0000 (23:27 -0800)]
Flip 'foo' and 'bar' to be consistent
The "Renaming dependencies" section initially uses 'foo' as the crate name and 'bar' as a rename, but then swaps them and uses 'bar' as the example crate name in the context of optional dependencies. Now both examples in this section treat 'foo' as the original crate name.
CPerezz [Thu, 28 Jan 2021 01:44:12 +0000 (02:44 +0100)]
Impl warn for locked install withoud Cargo.lock
If we're installing in --locked mode and there's no `Cargo.lock` published
ie. the bin was published before https://github.com/rust-lang/cargo/pull/7026
the cargo install errors were not stating that it was due to the lack of
the `Cargo.lock` file. Instead, the error seemed completely unrelated.
Therefore, this tries to address this by adding a warn in the stderr
output.
bors [Mon, 25 Jan 2021 16:16:43 +0000 (16:16 +0000)]
Auto merge of #9097 - ehuss:tracking-issue-template-update, r=alexcrichton
Minor update to tracking issue template.
Just some minor tweaks, move the "about" to the bottom since it isn't that important (I think the summary should be first). Also add a link to an RFC if it is an RFC.
bors [Thu, 21 Jan 2021 21:04:12 +0000 (21:04 +0000)]
Auto merge of #9092 - ehuss:unstable-updates, r=alexcrichton
Unstable updates
This is a collection of updates for unstable/nightly feature support, intended to provide better messages for users and better internal and external documentation. Separated by commit, in summary:
* Added comments and new docstrings for improved internal documentation.
* Added new documentation to the reference guide on how unstable things work.
* Also added redirects for stabilized features so any external links won't be broken.
* Add a targeted error message if you put `cargo-features` in the wrong place in `Cargo.toml`.
* Remove `publish-lockfile`. The feature was stabilized without the key in #7026 about 1.5 years ago. Also added "removed" support for features, which prints out a more helpful error message.
* Add help messages about stabilized `-Z` flags (instead of spitting out an unhelpful error message).
* Add help messages about stabilized `cargo-features` features.
* Add more context to the error when using `cargo-features` on stable.
* Unhide nightly CLI flags. I changed my mind on how these should work. I think it is useful to "advertise" the existence of these options on stable. The error message if you try to use it should help guide on what to do.
Eric Huss [Thu, 21 Jan 2021 03:46:50 +0000 (19:46 -0800)]
Add more helpful message with stabilized -Z flags.
Previously, when something was stabilized, Cargo would spit out a very
unhelpful error message about an unknown -Z flag. This changes it so
that it displays a helpful warning (or error).
Eric Huss [Thu, 21 Jan 2021 03:16:54 +0000 (19:16 -0800)]
Unhide nightly-only flags.
I changed my mind on how these should work. I think it is useful to
"advertise" the existence of these options on stable. The error message
if you try to use it should help guide on what to do.
bors [Wed, 20 Jan 2021 19:02:26 +0000 (19:02 +0000)]
Auto merge of #9077 - ehuss:fix-doc-resolver2-proc-macro, r=alexcrichton
Fix some issues with `cargo doc` and the new feature resolver.
This contains two related commits fixing some issues with `cargo doc` and the new feature resolver.
The first commit fixes a panic. The old code was using `UnitFor::new_normal` when computing doc units, but this was wrong. That essentially circumvents the new feature resolver, and breaks determining the correct features to use. I don't remember exactly what I was thinking when I wrote that, other than "rustdoc shouldn't care", but it does matter.
Unfortunately changing that makes collisions worse because it aggravates #6313, so the second commit adds some logic for removing some collisions automatically. Specifically:
- Prefers dependencies for the target over dependencies for the host (such as proc-macros).
- Prefers the "newest" version if it comes from the same source.
There are still plenty of situations where there can be collisions, but I'm uncertain on the best way to handle those.
bors [Wed, 20 Jan 2021 16:12:56 +0000 (16:12 +0000)]
Auto merge of #8037 - djc:rfc-2495, r=ehuss
Implement support for rust-version field in project metadata
Needs a bunch more work, but I'd like some early feedback! Remaining work:
- [x] Bikeshedding name (picked `rust-version` here over `msrv` or `min-rust-version`)
- [x] Failing test for local dependency with unsatisfied MSRV
- [x] Requirement cannot be smaller than 1.27
- [x] Requirement cannot be smaller than initial release of edition used
- [x] Check for `run`, `verify` and `publish` subcommands
- [x] More tests (would love feedback on this)
- [x] Handle pre-release identifiers
- [x] Disallow semver range operators
- [x] Feature gate it
- [x] Add documentation for unstable feature
Minimal version of `@moxian's` earlier take in #7801.
bors [Mon, 18 Jan 2021 19:51:07 +0000 (19:51 +0000)]
Auto merge of #9075 - alexcrichton:fix-cycles, r=ehuss
Fix a bug in Cargo's cyclic dep graph detection
Cargo's cyclic dependency graph detection turns out to have had a bug
for quite a long time as surfaced by #9073. The bug in Cargo has to do
with how dev-dependencies are handled. Cycles are "allowed" through
dev-dependencies because the dev-dependency can depend on the original
crate. Our cyclic graph detection, however, was too eagerly flagging a
package as known to not have a cycle before we had processed everything
about it.
The fix here was basically to just simplify the graph traversal. Instead
of traversing the raw `Resolve` this instead creates an alternate
in-memory graph which has the actual edges we care about for cycle
detection (e.g. every edge that wasn't induced via a dev-dependency).
With this simplified graph we then apply the exact same algorithm, but
this time it should be less buggy because we're not trying to do funky
things about switching sets about what's visited halfway through a
traversal.
Eric Huss [Thu, 14 Jan 2021 21:12:24 +0000 (13:12 -0800)]
Remove some doc collisions.
There are some cases where `cargo doc` will try to document two things
with the same crate_name. This attempts to automatically remove some of
those duplicates based on some rules:
- Prefers dependencies for the target over dependencies for the host
(such as proc-macros).
- Prefers the "newest" version if it comes from the same source.
There are still plenty of situations where there can be collisions, but
I'm uncertain on the best way to handle those.
Alex Crichton [Thu, 14 Jan 2021 16:40:12 +0000 (08:40 -0800)]
Fix a bug in Cargo's cyclic dep graph detection
Cargo's cyclic dependency graph detection turns out to have had a bug
for quite a long time as surfaced by #9073. The bug in Cargo has to do
with how dev-dependencies are handled. Cycles are "allowed" through
dev-dependencies because the dev-dependency can depend on the original
crate. Our cyclic graph detection, however, was too eagerly flagging a
package as known to not have a cycle before we had processed everything
about it.
The fix here was basically to just simplify the graph traversal. Instead
of traversing the raw `Resolve` this instead creates an alternate
in-memory graph which has the actual edges we care about for cycle
detection (e.g. every edge that wasn't induced via a dev-dependency).
With this simplified graph we then apply the exact same algorithm, but
this time it should be less buggy because we're not trying to do funky
things about switching sets about what's visited halfway through a
traversal.
bors [Tue, 12 Jan 2021 23:45:39 +0000 (23:45 +0000)]
Auto merge of #9066 - rubenrua:hotfix_sort_bins, r=ehuss
Sort available binaries when multiple
From:
```
error: `cargo run` could not determine which binary to run. Use the `--bin` option to specify a binary, or the `default-run` manifest key.
available binaries: basic-tutorial-13, basic-tutorial-6, basic-tutorial-1, basic-tutorial-4, basic-tutorial-9, basic-tutorial-2, basic-tutorial-3, basic-tutorial-5, basic-tutorial-12, playback-tutorial-4, basic-tutorial-8, basic-tutorial-7
```
To:
```
error: `cargo run` could not determine which binary to run. Use the `--bin` option to specify a binary, or the `default-run` manifest key.
available binaries: basic-tutorial-1, basic-tutorial-12, basic-tutorial-13, basic-tutorial-2, basic-tutorial-3, basic-tutorial-4, basic-tutorial-5, basic-tutorial-6, basic-tutorial-7, basic-tutorial-8, basic-tutorial-9, playback-tutorial-4
```
bors [Tue, 12 Jan 2021 17:14:29 +0000 (17:14 +0000)]
Auto merge of #9065 - alexcrichton:fix-build-bug-again, r=ehuss
Fix `links` vars showing up for testing packages
If a package is tested and the library for the package wasn't built
(e.g. only tested or it wasn't present) then the `links` env vars from
dependencies weren't showing up to the build script by accident. This
was an accidental regression from #8969.
The intention of #8969 was to exclude connections to build scripts
connected via dev-dependencies, but it only applied a heuristic because
the unit graph doesn't retain information about dev-dependencies. The
fix here is to instead actually retain information about
dev-dependencies which is only used for constructing the unit graph and
connecting build script executions to one another.
Alex Crichton [Mon, 11 Jan 2021 18:42:38 +0000 (10:42 -0800)]
Fix `links` vars showing up for testing packages
If a package is tested and the library for the package wasn't built
(e.g. only tested or it wasn't present) then the `links` env vars from
dependencies weren't showing up to the build script by accident. This
was an accidental regression from #8969.
The intention of #8969 was to exclude connections to build scripts
connected via dev-dependencies, but it only applied a heuristic because
the unit graph doesn't retain information about dev-dependencies. The
fix here is to instead actually retain information about
dev-dependencies which is only used for constructing the unit graph and
connecting build script executions to one another.
rubenrua [Tue, 12 Jan 2021 07:46:25 +0000 (08:46 +0100)]
Sort available binaries when multiple
From:
```
error: `cargo run` could not determine which binary to run. Use the `--bin` option to specify a binary, or the `default-run` manifest key.
available binaries: basic-tutorial-13, basic-tutorial-6, basic-tutorial-1, basic-tutorial-4, basic-tutorial-9, basic-tutorial-2, basic-tutorial-3, basic-tutorial-5, basic-tutorial-12, playback-tutorial-4, basic-tutorial-8, basic-tutorial-7
```
To:
```
error: `cargo run` could not determine which binary to run. Use the `--bin` option to specify a binary, or the `default-run` manifest key.
available binaries: basic-tutorial-1, basic-tutorial-12, basic-tutorial-13, basic-tutorial-2, basic-tutorial-3, basic-tutorial-4, basic-tutorial-5, basic-tutorial-6, basic-tutorial-7, basic-tutorial-8, basic-tutorial-9, playback-tutorial-4
```
bors [Mon, 11 Jan 2021 15:12:09 +0000 (15:12 +0000)]
Auto merge of #9059 - ehuss:fix-unit-for-host, r=alexcrichton
Fix unit_for computation on proc-macros in shared workspace.
There was a bug where the UnitFor was not being computed properly for proc-macros in a workspace with shared dependencies, integration tests, and a mixture of various flags (`--workspace --all-targets --all-features`). The issue is that [this line](https://github.com/rust-lang/cargo/blob/faf05ac316cd71100a691799cd8da02fce6dd85d/src/cargo/core/compiler/unit_dependencies.rs#L474), which is used when attaching the implicit dependency from an integration test to its library, was using the wrong unit_for value (it was not checking if the implicit lib is a proc-macro). The consequence is that the graph could be built inconsistently, causing features to be randomly selected incorrectly if the integration test happened to be the first unit processed.
The solution here is to use a common function for transitioning the unit_for value. The with_for_host/with_host_features split was mostly a consequence of how things evolved over time, and keeping them separate wasn't really necessary.
bors [Sun, 10 Jan 2021 20:25:54 +0000 (20:25 +0000)]
Auto merge of #9000 - JohnTitor:owner-add-docs, r=ehuss
Document `could not find the github team` error on `cargo owner --add`
When running `cargo owner --add`, the `could not find the github team` error often occurs due to lack of the permissions.
Example cases: https://github.com/rust-lang/cargo/issues/5297#issuecomment-653709183, https://github.com/rust-lang/crates.io/issues/2928
I think it's worth documenting that error and the possible solution.
bors [Fri, 8 Jan 2021 16:44:12 +0000 (16:44 +0000)]
Auto merge of #9053 - sharnoff:master, r=alexcrichton
[doc] add note about empty environment variables for missing manifest keys
This is a small addition to the cargo book.
In [the section](https://doc.rust-lang.org/cargo/reference/environment-variables.html#environment-variables-cargo-sets-for-crates) about the environment variables cargo sets to provide information about a crate, there's no mention about the value of those environment variables is if the key isn't present in the manifest.
As can be seen [here](https://github.com/rust-lang/cargo/blob/94e21ada550f91f5d32f2f51635792a79aeb8d63/src/cargo/core/compiler/compilation.rs#L287)¹, the values default to the empty string (instead of, say, not being present), so this change just adds a sentence in the book to clarify that.
If the terminology could be improved or there's a better place for this to be included, I'm happy to change it :)
---
¹ To be clear, I did also play around with this and found that the environment variables were empty. The code reference was more to double-check that the behavior I saw was a consistent pattern.
bors [Tue, 5 Jan 2021 21:39:20 +0000 (21:39 +0000)]
Auto merge of #8997 - ehuss:stabilize-features, r=alexcrichton
Stabilize -Zfeatures and -Zpackage-features.
This follows through with [RFC 2957](https://github.com/rust-lang/rfcs/pull/2957) to stabilize the new feature resolver, and `-Zpackage-features` command-line changes.
This also rewrites the "Features" chapter to try to expand it a little.
I decided to leave the `-Zfeatures` flag in for now for testing, but it can be removed at a later date.
There is a code change related to the `package-name/feature-name` syntax for the `--features` flag. I wanted to stabilize that for `resolver = "1"`, but I previously neglected to separate that behavior out, so it required change to `Workspace::members_with_features` to make that work (see the `resolver1_member_features` test).