xfs
[Top] [All Lists]

Re: [PATCH v2] xfsdump: use the full 32-bit generation number

To: Bill Kendall <wkendall@xxxxxxx>
Subject: Re: [PATCH v2] xfsdump: use the full 32-bit generation number
From: Christoph Hellwig <hch@xxxxxxxxxxxxx>
Date: Mon, 13 Feb 2012 11:45:20 -0500
Cc: Christoph Hellwig <hch@xxxxxxxxxxxxx>, xfs@xxxxxxxxxxx
In-reply-to: <4F393C08.6000801@xxxxxxx>
References: <1328813180-29095-1-git-send-email-wkendall@xxxxxxx> <20120212234720.GA3848@xxxxxxxxxxxxx> <4F393C08.6000801@xxxxxxx>
User-agent: Mutt/1.5.21 (2010-09-15)
On Mon, Feb 13, 2012 at 10:36:24AM -0600, Bill Kendall wrote:
> On 02/12/2012 05:47 PM, Christoph Hellwig wrote:
> >Looks generally good to me, but is there any good reason to not make
> >the -K argument extensible by requiring a version argument, even if
> >that one currently only supports 2 and 3 as valid values?
> 
> I thought about doing that, but given the low frequency of format
> changes decided it could wait until there's more than one old format
> to support. -K without an argument could be treated as "generate
> the previous media format", or with an argument it could generate
> a specific format. I'm okay with requiring a version argument though,
> being explicit is better.
> 
> FWIW, the way things have always worked is that xfsrestore provides
> backwards compatibility for all dump formats. xfsdump has, until
> now, always generated the current format. It's nice to provide
> the ability to generate old formats during a transition period,
> but long-term I don't see xfsdump retaining support for generating
> all old formats.

Ok, sounds fine.  I'll put it into the repository and after that we
should aim for an xfsdump 3.1.0 release ASAP.

<Prev in Thread] Current Thread [Next in Thread>