On 27/06/2013 03:33, Frank Steinmetzger wrote:
> On Wed, Jun 26, 2013 at 06:40:02PM +0200, Florian Philipp wrote:
> 
>>>> Windows compatibility is not a must, but a nice-to-have. That would reduce 
>>>> my
>>>> remaining choices to ExFAT, I presume.
>>>
>>
>> BTW: What's the Linux status on that one?
> 
> Well, the German Wikipedia says that a stable 1.0 came out in January.
> It’s “only” a FUSE fs, but so is NTFS. I’ll do some testing with it.
> 
>>> That's how I see it too.
>>>
>>> I used to use ext4 for external media but quickly found that my notebook
>>> was the only box that could use them...
>>
>> Isn't group id 100 defined as the users group on most Linuxen nowadays?
>>
>> chgrp 100 $mount && chmod 2777 $mount
> 
> That still leaves UID.  I want it to “just work”[TM] and never encounter
> any problems when I can least use them, and never have to check any file
> attributes.
> 
>> should work reasonably well.
>>
>> Regards,
>> Florian Philipp
> 
> I’ll keep the uid and gid bit on the stack.  I would disable the x bit
> though.  Executables are lime in DIR_COLORS, overriding every other
> colouring (e.g. red archive, green text and purple media files). *g*
> 
> I’m more concerned about the behaviour of automounters. And I faintly
> remember some user-centric setting as to what the default chmod of new
> files is, so I would have to do some chmod -R from time to time.
> 

The Unix filesystem model simply does not allow you to do that easily -
it is designed to do something else entirely and do that thing well

You can't even override the uid/gid/perms at mount time. The central
premise is that the user must set those values on his own files whenever
he wants to and the rest of the universe must fall into line with those
wishes...

Look at what it takes to do something simple like set the default perms
on a new file in a shared directory to be 664 - you need to get dirty
with POSIX ACLs, and then a simple umask run in a shell session
overrules all of that.

exfat does what you want - it was designed to "just work" on the very
large removeable media we have nowadays (think 7G movie files) and
bypass all the nonsense like "does the user that created this file even
exist on the machine that is reading it?"

It also works pretty well


-- 
Alan McKinnon
[email protected]


Reply via email to