bug#14556: suggestion for 'ls' coreutil, comma'd filesizes in -l output

[Eric Blake]

[adding the list back in the loop; also adding bug-coreutils to get a
bug number assigned for 2 doc bugs, and slightly reformatting to avoid
excessive top-posting]

On 06/04/2013 01:44 PM, smu johnson wrote:
> I did reply on purpose to you, but only because I wasn't sure about some of
> the text you wrote in your original reply, and because I didn't realize it
> was a list until now.
> 
> I got the first email address from http://www.gnu.org/software/coreutils/,
> where it says, "Send general questions or suggestions to: address@hidden".
> It didn't warn me that sending an email is the equivalent to sending to a
> mailing list.  So I mistakenly thought you were in charge of answering
> emails.  My bad, sorry about that.

My take on this is that you have discovered a doc bug, in that our web
page is not making it clear that our bug reporting address is a publicly
archived mailing list (although this setup is typical for most open
source projects, it is not universal; and it takes an outsider's
perspective like yours to make us realize that we have unstated
assumptions that ended up surprising you).  We could probably add a
sentence or two to the web page that makes it clear that your post goes
to a list, that we are unable to scrub anything you post (so even though
we do our best to munge things to minimize the effects, we cannot
guarantee that spammers will not obtain your address from your post),
and that list policy is to reply-to-all to keep you in the loop without
forcing you to subscribe to the list.

>>>
>>> 2)  I tried the command you ran, but I seem to be getting different
>>> results.  I don't really know what I'm doing.
>>>
>>> address@hidden:~$ touch COPYING HACKING
>>> address@hidden:~$ stat -c "%'s" COPYING HACKING
>>> ?s
>>> ?s
>> 
>> What does 'stat --version' say?

> 
> To answer your other question, stat (GNU coreutils) 8.5.

You found another doc bug - I can confirm that the behavior has changed
in the meantime (I see a change in behavior between my testing on RHEL
6.4 with stat 8.4, vs. Fedora 19 with stat 8.17), but do not see the
change mentioned in NEWS.  I suspect the change happened with this
commit in 8.7:

commit ca9aa7596094c4af094edb7eb63648570d53c100
Author: Paul Eggert <address@hidden>
Date:   Sat Nov 6 13:57:08 2010 -0700

    stat: do not rely on undefined behavior in printf formats

As for myself, I'm headed on vacation for most of June, so one of the
other list readers will probably beat me to writing these two patches
(more details on the list, and supplying the missing NEWS entry).  I
also found this recently, and thought it was a-propos to this situation:
https://rwmj.wordpress.com/2010/11/08/want-help-dont-email-me-directly/

-- 
Eric Blake   eblake redhat com    +1-919-301-3266
Libvirt virtualization library http://libvirt.org

signature.asc
Description: OpenPGP digital signature