[Toybox] [PATCH] Implement file(1).

[Rob Landley]

On Sat, Feb 13, 2016 at 3:57 PM, Isaac Dunham <ibid.ag at gmail.com> wrote:
> On Sat, Feb 13, 2016 at 09:56:29AM -0800, enh wrote:
>> Unlike the POSIX file(1), there's no magic file here, just hard-coded
>> common (non-obsolete) file formats. Personally most of my use of file(1)
>> is as a one-line readelf(1) summarizer, so although I assume a full POSIX
>> file(1) is out of scope (because just the database would likely be larger
>> than all the rest of toybox), a subset that only supports in-use file types
>> actually covers most of the use cases I encounter personally.
>
> FYI:
> binary cpio (almost never seen) starts with a short having the octal value
> 070707; there are big- and little-endian variants, so 0143561 is
> 'byte-swapped cpio archive'
> text-based cpio magic is a 6-byte string starting with "0707"
> "ASCII cpio archive"
> 070707 is 'odc' - "(pre-SVR4 orr odc)"
> 070701 is 'newc' - "(SVR4 with no CRC)",
> 070702 is 'newc' - "(SVR4 with CRC)"

I've been meaning to add another one, since there's a periodic thread
on linux-kernel about the fact cpio has no xattr support, and while
we're at it fixing 32 bit timestamps, lack of nanosecond timestamp
granularity, and 32 bit filesize limitations seemed like a good idea.

I think the first go-round was here:
http://lkml.iu.edu/hypermail/linux/kernel/1501.1/03905.html

That's come up a couple times since then but no actual motion on it.
(I should just sit down and do it. And send it to the posix list.)

Anyway, the option of an external format file for the "file" command
is not the _requirement_ for one. I wouldn't object to adding a check
to read /etc/file.dat or whatever it is, processing what it finds
there first (the file not existing being silently ignored), and
falling back to the built-in types if it's not there.

(The main reason it's still on the todo list is I could never quite
work out what format said file should have...)

Rob

 1455475227.0