GNU bug report logs - #24400
cp: default to --reflink=auto?

Previous Next

To add a comment to this bug, you must first unarchive it, by sending
a message to control AT debbugs.gnu.org, with unarchive 24400 in the body.
You can then email your comments to 24400 AT debbugs.gnu.org in the normal way.

Toggle the display of automated, internal messages from the tracker.

Message #5 received at submit <at> debbugs.gnu.org (full text, mbox):

From: Paul Eggert <eggert <at> cs.ucla.edu>
To: bug-coreutils <at> gnu.org
Subject: cp: default to --reflink=auto?
Date: Fri, 9 Sep 2016 15:54:32 -0700

Currently 'cp' defaults to not using reflinks, which can be 
significantly slower on a system that supports file cloning. I expect 
most users would prefer 'cp' to default to --reflink=auto for spped. 
Although there are some special cases otherwise, on balance how about if 
we change cp to default to --reflink=auto, and add an option 
--reflink=no for users who prefer the current default?

Message #8 received at 24400 <at> debbugs.gnu.org (full text, mbox):

From: Pádraig Brady <P <at> draigBrady.com>
To: Paul Eggert <eggert <at> cs.ucla.edu>, 24400 <at> debbugs.gnu.org
Subject: Re: bug#24400: cp: default to --reflink=auto?
Date: Sat, 10 Sep 2016 10:55:01 +0100

On 09/09/16 23:54, Paul Eggert wrote:
> Currently 'cp' defaults to not using reflinks, which can be 
> significantly slower on a system that supports file cloning. I expect 
> most users would prefer 'cp' to default to --reflink=auto for spped. 
> Although there are some special cases otherwise, on balance how about if 
> we change cp to default to --reflink=auto, and add an option 
> --reflink=no for users who prefer the current default?

Up until now we didn't do that so that we actually get a data copy.
I.E. traditionally such copies were more resilient to partial hardware failures.
However depending on the underlying hardware (which may do deduplication)
we're not guaranteed that an actual copy is made.
In discussions with kernel devs about this (in relation to efficient defaults
for copy_file_range), their current thinking is that separate devices are
required to give such guarantees, so reflinking by default is a valid
thing to do.

It might not be an appropriate thing to do in a minor release.
We're due one of those quite soon now, so I'd suggest waiting
a short while before changing?

thanks,
Pádraig

Message #11 received at 24400 <at> debbugs.gnu.org (full text, mbox):

From: Paul Eggert <eggert <at> cs.ucla.edu>
To: Pádraig Brady <P <at> draigBrady.com>, 24400 <at> debbugs.gnu.org
Subject: Re: bug#24400: cp: default to --reflink=auto?
Date: Sat, 10 Sep 2016 19:18:22 -0700

Pádraig Brady wrote:
> It might not be an appropriate thing to do in a minor release.
> We're due one of those quite soon now, so I'd suggest waiting
> a short while before changing?

Sure, there's no rush, we can leave this bug open.

I just now installed a similar change into Emacs for its copy-file function, 
into the master branch that is (i.e., the release after next).

Previous Next

GNU bug tracking system
Copyright (C) 1999 Darren O. Benham, 1997,2003 nCipher Corporation Ltd, 1994-97 Ian Jackson.