GitLab

Mggmuggins/shadowfile

This removes fs2 and linux file locking and simply runs the tests
without concurrency rather than actually fixing file locking to make
them work like they should.The value of working file locking on Linux is
debatable anyway.

I'll open an issue to track file locking with this repo, and maybe go
and do some work to get it implemented for Redox.

I used `cfg(not(target_os = "redox"))` for all the non-redox locking
behavior (including fs2), even though we only use it for tests. I'm
hoping that is alright.

This is just changing a bunch of function headers to use T: AsRef<str>
instead of String, and making things more consistent in general.

This is at a point where the implementation should work. The only way to
know for sure that it doesn't is to write more tests :/

I basically went through all the documentation and checked it to make
sure it was up to date with the implementations, and I added some crate
level notes.

Wrote a few more tests, made sure that doing dumb things would fail,
etc.

This is still very much a WIP. There are a number of things that I know
shouldn't work, and that I haven't written tests for yet. That'll be
coming at a later date. However, the current test suite passes and it
does mostly work.

The internal API is kinda messy. I mean,
`Option<(String, Option<Encoded>)>??!! Anyway, there's some
stuff that is really not pretty, and I would be happy for advice on how
to make it look a little nicer and be more maintainable..

Travis; Rustfmt; Update README

Readme; Rustify

Added config files for those two services. I'm not going to require
formatting via travis, but I thought I'd add everything so it won't do
major things to code we'd already written if somebody runs it.

@jackpot51, how difficult would it be to get travis for this repo?

Tests! Fix #15

Fix Up read/write so that they will work on linux.

Wrote big tests!
These are designed to be run on linux, in a regular cargo environment.
Do not edit things in `tests/`, they will break!

Just added some cargo metadata

Also added a few lines for getting ready to test. Another commit will
come soon with a few basic tests to make sure that everything is working
alright.

Parsing via FromStr, impl IntoIterator for &T and &mut T, implemented Display, which gives us ToString.

Timout, Iters, Bugfix

Here I did two major things:

AllGroups and AllUsers now have `iter()` and `iter_mut()` to iterate
over them. Not really sure if this is the correct way to implement
iteration here, so advice would be really nice if it's not.

Implemented a timeout when passwords are incorrect. All we do is sleep
if the password is not verified.

Also fixed a little bug with Group::to_string()

Improve Write

This introduces even harder dependencies on syscall, but I think
it will be a much more robust solution to the problem of concurrent
access for `/etc/passwd` and `/etc/group`

Implement Unset

Implement unset passwords (see docs).

Also fixed a small bug where blank passwords would allow any password
rather than just a blank field.

Implement Remove

API Overhaul

Fixed a stupid mistake with User::shell_cmd

Implemented AllUsers::save and AllGroups::save for persisting changes to
the system. @jackpot51 @goyox86 I'd really like to know what you think
about this API. I don't nessesarily like it, but I'm not sure how to
implement something else that's more elegant. :/

Also did a bunch of stuff with docs

A full list of accomplishments for this commit can be viewed on my
message on the PR to redox_users. My main difficulty is figuring
out how to save state for AllUsers and AllGroups (Question of wether
to simply write on Drop or add a call for it). 5 other TODO's can be
found.

@goyox86 I'd like to know what you think about all this.

I rearranged most of the functions into AllGroups and AllUsers, and
adjusted a lot of the API's. I think this is a more sane way of doing
things, but comments are always appreciated.

I removed the Iterators and replaced them with the "get" methods.

Also lots of docs work. I think it should be fairly straightforward to
use this crate now.

@jackpot51 I also fixed the bug where empty passwords will accept
anything.

Improved error handling.

- Made a type alias Result<T>.
- Removed a few variant of the NotFound and AlreadyExists errors.
- Updated to failure v1.1.1.
- Removed some helper methods as building errors is simpler now.

TODO: Describe the details of the changes.

Added add_user, add_group, get_gid; fixed empty user list issue