`std::path::absolute` #91673

ChrisDenton · 2021-12-08T21:16:07Z

Implements #59117 by adding a std::path::absolute function that creates an absolute path without reading the filesystem. This is intended to be a drop-in replacement for std::fs::canonicalize in cases where it isn't necessary to resolve symlinks. It can be used on paths that don't exist or where resolving symlinks is unwanted. It can also be used to avoid circumstances where canonicalize might otherwise fail.

On Windows this is a wrapper around GetFullPathNameW. On Unix it partially implements the POSIX pathname resolution specification, stopping just short of actually resolving symlinks.

rust-highfive · 2021-12-08T21:16:11Z

r? @Mark-Simulacrum

(rust-highfive has picked a reviewer for you, use r? to override)

the8472 · 2021-12-08T21:30:51Z

absolute(a.join(b)) will give resolve()-like behavior which some other languages have, right? If so then the comment could mention that combination as a way to resolve relative paths to something other than the current working directory.

ChrisDenton · 2021-12-08T21:46:17Z

Good point! But I'm slightly wary of the Windows behaviour here. I'm thinking of legacy paths such as D:foo which will resolve to the D drive's current directory if the application is run from the command prompt (or the root of D otherwise).

E.g.

let path = Path::new(r"C:\path").join("D:foo");
// path is "D:foo"
let absolute = absolute(path);
// absolute is like "D:\something\foo"
// actual value depends on the `D` drive's current directory.

My point being that the given path is relative but the base directory is still ignored.

the8472 · 2021-12-08T21:55:53Z

Is that an issue with join or with absolute? Or with using them to achieve resolve?

ChrisDenton · 2021-12-08T22:12:48Z

Hm, probably with using them to implement resolve. The command prompt can have a different current directory for each letter from A to Z (though it may not have them all). So to capture this you'd need to pass in mappings of (optional) current directories for each drive.

join could work differently. E.g. it could treat D:foo as either .\D:foo or D:\foo. But it doesn't and if it did it may be unexpected. I guess absolute could error on such paths. However, I don't think it should do any thing different from the OS as the goal here is to be as close to canonicalize as possible.

To be clear, I'm uncertain how much of an issue this is but it does make me wary of suggesting it.

bjorn3 · 2021-12-09T13:44:19Z

library/std/src/sys/windows/path.rs

+        // SAFETY: `fill_utf16_buf` ensures the `buffer` and `size` are valid.
+        // `lpfilename` is a pointer to a null terminated string that is not
+        // invalidated until after `GetFullPathNameW` returns successfully.
+        |buffer, size| unsafe { c::GetFullPathNameW(lpfilename, size, buffer, ptr::null_mut()) },


Does this work for non \\?\ paths bigger than MAX_PATH? It would be nice to be able to do std::path::absolute() and then prefix \\?\ if necessary to handle long paths.

Indeed it does! The std already does exactly that. Unfortunately the current docs for GetFullPathNameW suggests otherwise but this was an oversight that's going to be corrected when the online docs are next rebuilt (see the sdk-api source).

Mark-Simulacrum · 2022-01-10T20:09:44Z

In general this seems "fine", though I don't know if it's quite the API we want. Adding it as unstable to start though seems quite reasonable; we can discuss prior to stabilization whether this hits the right point on the API spectrum. Some initial questions I might think about:

Should we prepend the current directory, or skip that step? The user can typically join with the current directory afterwards, right?
Is the avoidance of stripping .. actually the right call? It does seem consistent with the docs linked, but may not be the most intuitive behavior (particularly given that Windows doesn't do that).
Probably some other questions based on the thread here so far, you probably have a better sense than me.

Can you file a tracking issue, and then I can r+?

ChrisDenton · 2022-01-10T22:43:27Z

@Mark-Simulacrum I've filed a tracking issue. I've added your points to the unresolved questions and I'll try to give my thoughts on them soon.

Mark-Simulacrum · 2022-01-10T22:47:34Z

@bors r+ rollup

Thanks! Going to go ahead and approve this PR.

bors · 2022-01-10T22:47:35Z

📌 Commit 0660732 has been approved by Mark-Simulacrum

…imulacrum `std::path::absolute` Implements rust-lang#59117 by adding a `std::path::absolute` function that creates an absolute path without reading the filesystem. This is intended to be a drop-in replacement for [`std::fs::canonicalize`](https://doc.rust-lang.org/std/fs/fn.canonicalize.html) in cases where it isn't necessary to resolve symlinks. It can be used on paths that don't exist or where resolving symlinks is unwanted. It can also be used to avoid circumstances where `canonicalize` might otherwise fail. On Windows this is a wrapper around [`GetFullPathNameW`](https://docs.microsoft.com/en-us/windows/win32/api/fileapi/nf-fileapi-getfullpathnamew). On Unix it partially implements the POSIX [pathname resolution](https://pubs.opengroup.org/onlinepubs/9699919799/basedefs/V1_chap04.html#tag_04_13) specification, stopping just short of actually resolving symlinks.

ehuss · 2022-01-12T05:40:24Z

@bors r-

Failed in #92785 (comment)

The wasm32-unknown-unknown target reuses unix/path.rs but itself is not unix.

Let me know if you have any questions about that or how to test it.

ChrisDenton · 2022-01-12T14:40:09Z

Thanks! I got wasm32-unknown-unknown to build locally. The fix was fortunately simple. I also added the tracking issue number and fixed a typo while I was there.

bors · 2022-02-08T13:41:35Z

💔 Test failed - checks-actions

ChrisDenton · 2022-02-08T14:25:11Z

#87869 changed the way of creating a const io::Error. So I've rebased this PR and switch to using io::const_io_error! in library/std/src/path.rs. Nothing else has changed.

Mark-Simulacrum · 2022-02-08T20:44:45Z

@bors r+

bors · 2022-02-08T20:44:46Z

📌 Commit 81cc3af has been approved by Mark-Simulacrum

bors · 2022-02-13T12:03:55Z

⌛ Testing commit 81cc3af with merge 1f4681a...

bors · 2022-02-13T14:28:34Z

☀️ Test successful - checks-actions
Approved by: Mark-Simulacrum
Pushing 1f4681a to master...

rust-timer · 2022-02-13T16:00:52Z

Finished benchmarking commit (1f4681a): comparison url.

Summary: This benchmark run did not return any relevant results.

If you disagree with this performance assessment, please file an issue in rust-lang/rustc-perf.

@rustbot label: -perf-regression

axetroy · 2022-03-05T15:12:33Z

library/std/src/path/tests.rs

+    assert_eq!(absolute("//a/b/c").unwrap(), Path::new("//a/b/c"));
+    assert_eq!(absolute("///a/b/c").unwrap(), Path::new("/a/b/c"));
+    assert_eq!(absolute("/a/b/c/").unwrap(), Path::new("/a/b/c/"));
+    assert_eq!(absolute("/a/./b/../c/.././..").unwrap(), Path::new("/a/b/../c/../.."));


Unix test without dot prefix.

assert_eq!(absolute("./a").unwrap(), Path::new("/pwd/a")); // return /pwd/./a assert_eq!(absolute("../a").unwrap(), Path::new("/parent_pwd/a")); // return /pwd/../a

I have tried in darwin. it is wrong

Thank you for reporting this. I'll add more tests and fix the dot prefix case.

* https://docs.rs/normpath * rust-lang/rust#91673

rust-highfive assigned Mark-Simulacrum Dec 8, 2021

rust-highfive added the S-waiting-on-review Status: Awaiting review from the assignee but also interested parties. label Dec 8, 2021

apiraino added the T-libs Relevant to the library team, which will review and decide on the PR/issue. label Dec 9, 2021

bjorn3 reviewed Dec 9, 2021

View reviewed changes

dtolnay added T-libs-api Relevant to the library API team, which will review and decide on the PR/issue. and removed T-libs Relevant to the library team, which will review and decide on the PR/issue. labels Dec 10, 2021

ChrisDenton mentioned this pull request Dec 14, 2021

Windows device path and long-path meta issue. rust-lang/cargo#9770

Open

Mark-Simulacrum added S-waiting-on-author Status: This is awaiting some action (such as code changes or more information) from the author. and removed S-waiting-on-review Status: Awaiting review from the assignee but also interested parties. labels Jan 10, 2022

ChrisDenton mentioned this pull request Jan 10, 2022

Tracking Issue for std::path::absolute #92750

Closed

3 tasks

bors added S-waiting-on-bors Status: Waiting on bors to run and complete tests. Bors will change the label on completion. and removed S-waiting-on-author Status: This is awaiting some action (such as code changes or more information) from the author. labels Jan 10, 2022

matthiaskrgr mentioned this pull request Jan 11, 2022

Rollup of 16 pull requests #92784

Closed

matthiaskrgr mentioned this pull request Jan 11, 2022

Rollup of 15 pull requests #92785

Closed

bors added S-waiting-on-author Status: This is awaiting some action (such as code changes or more information) from the author. and removed S-waiting-on-bors Status: Waiting on bors to run and complete tests. Bors will change the label on completion. labels Jan 12, 2022

bors added S-waiting-on-review Status: Awaiting review from the assignee but also interested parties. and removed S-waiting-on-bors Status: Waiting on bors to run and complete tests. Bors will change the label on completion. labels Feb 8, 2022

This comment has been minimized.

Sign in to view

ChrisDenton force-pushed the path-absolute branch from fd5bb45 to c17568e Compare February 8, 2022 14:25

This comment has been minimized.

Sign in to view

ChrisDenton added 2 commits February 8, 2022 14:57

std::path::absolute

d59d32c

Fix absolute issues

81cc3af

ChrisDenton force-pushed the path-absolute branch from c17568e to 81cc3af Compare February 8, 2022 14:59

bors added S-waiting-on-bors Status: Waiting on bors to run and complete tests. Bors will change the label on completion. and removed S-waiting-on-review Status: Awaiting review from the assignee but also interested parties. labels Feb 8, 2022

bors added the merged-by-bors This PR was explicitly merged by bors. label Feb 13, 2022

bors merged commit 1f4681a into rust-lang:master Feb 13, 2022

rustbot added this to the 1.60.0 milestone Feb 13, 2022

bors mentioned this pull request Feb 13, 2022

const Path initialisation #92930

Closed

ChrisDenton deleted the path-absolute branch February 13, 2022 16:16

jyn514 mentioned this pull request Feb 13, 2022

Move some more bootstrap logic from python to rust #92260

Merged

andylizi mentioned this pull request Feb 17, 2022

Add function to make paths absolute, which is different from canonicalization #59117

Open

axetroy reviewed Mar 5, 2022

View reviewed changes

op3 mentioned this pull request Aug 23, 2022

Do not resolve symlinks and especially hard links and bind mounts for --cwd and cwd wez/wezterm#2449

Closed

messense added a commit to messense/maturin that referenced this pull request Oct 18, 2022

Use normpath crate instead of fs::canonicalize

006e5dd

* https://docs.rs/normpath * rust-lang/rust#91673

messense added a commit to messense/maturin that referenced this pull request Oct 18, 2022

Use normpath crate instead of fs::canonicalize

451a25f

* https://docs.rs/normpath * rust-lang/rust#91673

matthiaskrgr mentioned this pull request Nov 10, 2023

ice: calling const FnDef errored when it shouldn't #117795

Open

Provide feedback

Saved searches

Use saved searches to filter your results more quickly

`std::path::absolute` #91673

`std::path::absolute` #91673

ChrisDenton commented Dec 8, 2021 •

edited

Loading

rust-highfive commented Dec 8, 2021

the8472 commented Dec 8, 2021 •

edited

Loading

ChrisDenton commented Dec 8, 2021 •

edited

Loading

the8472 commented Dec 8, 2021 •

edited

Loading

ChrisDenton commented Dec 8, 2021

bjorn3 Dec 9, 2021

ChrisDenton Dec 9, 2021

Mark-Simulacrum commented Jan 10, 2022

ChrisDenton commented Jan 10, 2022

Mark-Simulacrum commented Jan 10, 2022

bors commented Jan 10, 2022

ehuss commented Jan 12, 2022

ChrisDenton commented Jan 12, 2022

bors commented Feb 8, 2022

This comment has been minimized.

ChrisDenton commented Feb 8, 2022

This comment has been minimized.

Mark-Simulacrum commented Feb 8, 2022

bors commented Feb 8, 2022

bors commented Feb 13, 2022

bors commented Feb 13, 2022

rust-timer commented Feb 13, 2022

axetroy Mar 5, 2022 •

edited

Loading

ChrisDenton Mar 5, 2022

std::path::absolute #91673

std::path::absolute #91673

Conversation

ChrisDenton commented Dec 8, 2021 • edited Loading

rust-highfive commented Dec 8, 2021

the8472 commented Dec 8, 2021 • edited Loading

ChrisDenton commented Dec 8, 2021 • edited Loading

the8472 commented Dec 8, 2021 • edited Loading

ChrisDenton commented Dec 8, 2021

bjorn3 Dec 9, 2021

Choose a reason for hiding this comment

ChrisDenton Dec 9, 2021

Choose a reason for hiding this comment

Mark-Simulacrum commented Jan 10, 2022

ChrisDenton commented Jan 10, 2022

Mark-Simulacrum commented Jan 10, 2022

bors commented Jan 10, 2022

ehuss commented Jan 12, 2022

ChrisDenton commented Jan 12, 2022

bors commented Feb 8, 2022

This comment has been minimized.

ChrisDenton commented Feb 8, 2022

This comment has been minimized.

Mark-Simulacrum commented Feb 8, 2022

bors commented Feb 8, 2022

bors commented Feb 13, 2022

bors commented Feb 13, 2022

rust-timer commented Feb 13, 2022

axetroy Mar 5, 2022 • edited Loading

Choose a reason for hiding this comment

ChrisDenton Mar 5, 2022

Choose a reason for hiding this comment

`std::path::absolute` #91673

`std::path::absolute` #91673

ChrisDenton commented Dec 8, 2021 •

edited

Loading

the8472 commented Dec 8, 2021 •

edited

Loading

ChrisDenton commented Dec 8, 2021 •

edited

Loading

the8472 commented Dec 8, 2021 •

edited

Loading

axetroy Mar 5, 2022 •

edited

Loading