Re: [arch-dev-public] [signoff] coreutils 8.4-3

On 15/04/10 20:04, Allan McRae wrote:

On 15/04/10 17:10, Dieter Plaetinck wrote:

...

On Thu, 15 Apr 2010 09:28:04 +1000 Allan McRae wrote:

...

On 15/04/10 03:40, Thomas Bächler wrote:

...

Am 14.04.2010 18:42, schrieb Dieter Plaetinck:

...

...

On Tuesday 13 April 2010 14:04:44 Allan McRae wrote: > This changes /usr/bin/[ from a symlink to /usr/bin/test to the > actual binary provided upstream. The use of the symlink has > been in Arch for ages (probably for ever), but I can not see > what this change will break. Having the symlink does break > some stuff (FS#19063). Note that Fedora moved from the symlink > to the binary for "[" in 2004.

so what exactly does the binary do? is it an exact copy of /usr/bin/test, is it a small program that does exec(), or .. ? I downloaded the 8.4 tarball but couldn't quickly find the answer.

I don't think it matters. Neither [ nor test from coreutils are actually ever used - the bash (or whatever shell you use) builtins [ and test are used instead.

In general, this is the only real difference: NOTE: [ honors the --help and --version options, but test does not. test treats each of those as it treats any other nonempty STRING.

The rest I think is better compatibility with test synatx when your shell does not provide "[".

Anyway, we should be supplying what upstream installs.

I asked it just out of curiosity. I agree we should stick to upstream.

Anyway.... can I now have an i686 signoff? :P

This has taken so long the package is now out of date... I am moving it. Allan