samgiles/rs-mktemp

mktemp for Rust

mktemp

This module provides a simple way of creating temporary files and directories where their lifetime is defined by the scope they exist in.

Once the variable goes out of scope, the underlying file system resource is removed.

See documentation for full API, and other use cases.

Example

use mktemp::Temp;

{
  let temp_file = Temp::new_file().unwrap();
  let file = fs::File::open(temp_file).unwrap();
} // temp_file is cleaned from the fs here

Contributors

Special thanks to our contributors! Contributors

License

MPL v2

rs-mktemp's Issues

mktemp templates & security

Hiya, this isn't an 'issue' but I wanted to ask about the design / what use cases you aim to cover.

First, some "mktemp" functions allow a template filename to be passed, e.g. C's mkstemp and the shell command. Allowing this would be nice.

Second, there are several security issues to consider.

It may turn out better just to wrap C's mkstemp or mkstemps?

Release flag and method are inefficient; operate by value instead

It’d be more efficient to have release() consume self and return the PathBuf:

pub fn release(self) -> PathBuf {
    let path = mem::replace(&mut self.path, PathBuf::new());
    mem::forget(self);
    path
}

That way you don’t need to allocate twice just to get the path out as a PathBuf.

This also allows you to scrap the flag _release altogether.

Provide a mechanism for retaining file

The default behaviour is nice for writing tests since everything is cleaned up afterwards. But painful if the tests fail -- it becomes hard to debug because it's not possible to see what is in the file that you've just deleted.

It would be nice to have a mechanism to stop file deletions after, say, a panic.

crates.io links are not updated

Fix is already commited to master, but crates.io update is not done. v5.1 would fix that.

Upgrade uuid

uuid 0.8 is deprecated since april 2022. Can it be upgraded?

Don't panic on remove if the file does not exist

The current implementation panic in the Drop impl if removing the file failed. It panics even if the remove failed due to the file not existing. The purpose of the temp file is to make the file disappear, so why should that panic?

Support creating temporary file with particular extension

I have a use case where I'm generating a temporary file and then opening it with a text editor. The editor uses the file extension to decide which syntax highlighter to use so it would be useful to be able to specify the extension by hand.

[Suggestion] A way to provide a containing directory other than by changing the environment variable

Maybe something like

new_dir_in(&Path)

?

Or maybe a structure like

TempBuilder::new()
  .in(&Path) // optional
  .file() // or .dir

Split Temp into TempFile and TempDir

The _type flag is wasteful and makes for trivially more expensive construction and destruction. It’d be more efficient—and I would argue more ergonomic—to split the Temp type into two types, TempFile and TempDir. (I’m not going to suggest phantom types, because they’re more complex for no gain that is evident to me.)