libstd/io/fs.rs: avoid reading directories as files on *BSD #19303
Conversation
| unreachable!() | ||
|
|
||
| let bytes = File::open(path).and_then(|f| | ||
| if cfg!(any(target_os = "freebsd", target_os = "dragonfly")) { |
There was a problem hiding this comment.
We typically try to not house logic such as this in the compiler itself but rather put it in the standard library. As std::io::File is a cross-platform abstraction and we generally try to enforce the same behavior across platforms. If read_to_end() is only available on freebsd/dragonfly, I'd personally be a little more in favor of putting the error all the way over into std::io.
Exposing the platform-specific behavior will be done through std::sys in time as we expose the raw underlying system primitives for this sort of operation.
nodakai
force-pushed
the
libsyntax-reject-dirs
branch
from
5fe5c79 to
5d1865d
Compare
|
@alexcrichton I restricted the change to |
| }, | ||
| Err(e) => Err(e) | ||
| } | ||
| ) |
There was a problem hiding this comment.
Sorry to be somewhat nitpickity, but could you move this to std::io instead? We're generally trying to keep the std::sys bindings a low-level as possible (no worries about cross-platform errors like this) while pushing the compatibility layer to std::io instead.
nodakai
force-pushed
the
libsyntax-reject-dirs
branch
from
5d1865d to
6080b7d
Compare
nodakai
changed the title
|
@alexcrichton The check logic was moved to |
nodakai
force-pushed
the
libsyntax-reject-dirs
branch
2 times, most recently
from
8c1b2fb to
e7e2d7a
Compare
|
I fixed |
The same goes for sys::fs::FileDesc::fstat() on Windows. Signed-off-by: NODA, Kai <nodakai@gmail.com>
On *BSD systems, we can open(2) a directory and directly read(2) from it due to an old tradition. We should avoid doing so by explicitly calling fstat(2) to check the type of the opened file. Opening a directory as a module file can't always be avoided. Even when there's no "path" attribute trick involved, there can always be a *directory* named "my_module.rs". Fix rust-lang#12460 Signed-off-by: NODA, Kai <nodakai@gmail.com>
nodakai
force-pushed
the
libsyntax-reject-dirs
branch
from
e7e2d7a to
3980cde
Compare
|
A unit test on OS X failed because it expected an error message modified by my patch. I updated the unit test as well. |
|
@cmr Please note that I updated my commit after @alexcrichton reviewed it. Update: oops, it was a classical case of race between two threads |
…hton On *BSD systems, we can `open(2)` a directory and directly `read(2)` from it due to an old tradition. We should avoid doing so by explicitly calling `fstat(2)` to check the type of the opened file. Opening a directory as a module file can't always be avoided. Even when there's no "path" attribute trick involved, there can always be a *directory* named `my_module.rs`. Incidentally, remove unnecessary mutability of `&self` from `io::fs::File::stat()`.
On *BSD systems, we can
open(2)a directory and directlyread(2)from it due to an old tradition. We should avoid doing so by explicitly callingfstat(2)to check the type of the opened file.Opening a directory as a module file can't always be avoided. Even when there's no "path" attribute trick involved, there can always be a directory named
my_module.rs.Incidentally, remove unnecessary mutability of
&selffromio::fs::File::stat().