Hey guys, I'd like to implement hostname. It that ok ?

thanks :)

I think this is an excellent project that many people can contribute to, since it can be developed in parallel by its nature.

To avoid the possibility of multiple people working on the same util, it'd be nice to have a way for people to "claim" an item.

Hey guys,

While I was trying to extract code from id.rs to use it in groups.rs I stumble upon a problem with the macros that we currently use. I've exposed this problem in the most recent meetup and also in the IRC, and the conclusion was basically the same.

The problem is the following.

groups.rs will include c_types.rs
c_types.rs will use crash! that is defined in util.rs which is being included by groups.rs
When I try to compile the program, I get this error:

sh -c 'rustc --opt-level=3 -o build/id id/id.rs'
common/util.rs:16:62: 16:66 error: unresolved name `NAME`. Did you mean `free`?
common/util.rs:16         safe_write!(&mut ::std::io::stderr(), "{}: error: ", NAME);
                                                                               ^~~~
note: in expansion of format_args!
<std macros>:4:9: 5:6 note: expansion site
<std macros>:1:1: 1:1 note: in expansion of write!
common/util.rs:32:15: 32:40 note: expansion site
common/util.rs:30:1: 37:2 note: in expansion of safe_write!
common/util.rs:16:9: 16:68 note: expansion site
common/util.rs:13:1: 19:2 note: in expansion of show_error!
common/util.rs:24:9: 24:44 note: expansion site
common/util.rs:22:1: 27:2 note: in expansion of crash!
common/c_types.rs:48:17: 48:59 note: expansion site
common/util.rs:16:62: 16:66 error: unresolved name `NAME`. Did you mean `free`?
common/util.rs:16         safe_write!(&mut ::std::io::stderr(), "{}: error: ", NAME);
                                                                               ^~~~

1. This macro shouldn't work at all because of macros hygiene. It works however...
2. Solution proposed is to pass the name of the program as an argument, not as a constant.

Suggestion: add answer to readme.

If target rust version is not current then create branch for current version.

I ask because for example there is no from_utf8_slice_opt anymore

I'm using rust -v 0.9-pre (aa4455e...)

When I do make on the main directory, cat fails with this.

sh -c 'rustc --opt-level=3 -o build/cat cat/cat.rs'
cat/cat.rs:185:32: 185:53 error: unresolved name `std::path::Path::init`.
cat/cat.rs:185     match result(|| File::open(&std::path::Path::init(path.as_slice()))) {
                                               ^~~~~~~~~~~~~~~~~~~~~
error: aborting due to previous error
task 'rustc' failed at 'explicit failure', /Users/alan/lang/rust/src/libsyntax/diagnostic.rs:102
task '<main>' failed at 'explicit failure', /Users/alan/lang/rust/src/librustc/lib.rs:390
make: *** [build/cat] Error 101

Is it me or the program ?

Hi! I will take uniq unless anyone objects.

I personally don't like it.

So in nutshell what about adding

notifications:
  email: false

to .travis

I'll do this.

I'd like to work on this. (I already started in fact).
I saw there was already #34 but no answer since 3 months so I'd like to take the liberty to do it if there is no problem with that.

What about an logo image?
What do you want to see as logo image?

What about the ring from Rust logo and some instruments in the middle?

Some of utils are currently linux only. whoami , hostname ... We should separate them from crossplatform list somehow.

And ideally make the make to automatically detect to use both lists or crossplatform only

Working on it...

I already implemented a version for linux only. First I use strftime to format time but removed it because of simplicity and less dependencies.
It uses /proc/uptime for uptime, code from users for the number of users and getloadavg() for load average. Both /proc/uptime and getloadavg are unix only (linux only?) so in the meantime I added uptime to UNIX_PROGS.
Based roughly on the code from C coreutils.

I'm working on base64

I'd like to do this.

What about setting it as linux-only?

I know its a rust issue, but looking at about 3MB for each binary, versus around 100KB for coreutils binaries on my system (ubuntu 14.04), I wonder if there is anything that can be done to cut it down.

I'll do this as well.

I'll work on this.

Taking on du!

I'd like to work on this.

logname is very similar to whoami. Do i need to use a common folder, like stated in #103?

I'll do this.

for example for crash! : Heather@f6c4602

Reviewing other pull requests to not implement the same is hard.
Faster reaction on them will help.

Hi,

I would like to get my hand dirty on ls stuff? but not sure if someone is working on it yet

After thinking about this, I'm pretty sure that rather than a dev branch, we should have individual branches for big projects that are not yet "complete" (e.g. cp). What does everyone else think?

I'm working on cp.

For example, some utilities display errors like utilname: error msg, while others do it like error: error msg, while yet more do error msg.

IMHO, they should all be utilname: error: error msg or something of the sort. That way, people will know which utility failed if multiple are executed in a shell script and for normal usage people will be able to tell that there was an error.

I'll do this too, seeing as it's so closely related to rm.

We should have the Makefile prevent certain programs from building on the various operating systems (most specifically Windows).

I'm working on tee.

May I take head ?

thanks.

Dev branch could be used as some centralized place for testing / reviewing pull requests.

For example my and this #77 pull request can be there together so we will see how will they work together.

What do you think?

The id program has options that will do the same as groups program.

Also, the id program has a bunch of structs that could be reused not only by groups, but other programs.

Before proceeding with the groups program, I want to ask for some help to decide what to do.

1. We can make public structs and put them in their own file. This way whoami , id and potentially groups could those structs.
2. Make some of the id program functions public so that groups can reuse them with no effort.

Thanks.

After commit rust-lang/rust@e09a8e8, the link attribute is obsolete. It's an easy fix, but the author parameter is lost. I don't know if there's another attribute to use for that. In any case it doesn't seem significant.

WIP.

I'd like to implement basename.

What exactly are we aiming for in these implementations: are we aiming to replicate the quirks of the GNU implementations exactly? Or trying to be POSIXly correct? Or just doing our best for now and worrying about the details later?

My 2p: the GNU quirks aren't all good, e.g. GNU uniq -i thinks 'Γ' and 'γ' are distinct.

E.g. I think cat -n is particularly slow because it is doing a syscall to read 2 bytes at a time, wrapping the file reader in a std::rt::io::buffered::BufferedReader would likely fix this. (NB this is probably unnecessary overhead for cat without an arguments.)

I will be working on echo. Just wanted to make sure no one else is working on it :)