xfs-masters
[Top] [All Lists]

Re: [PATCH 6/6] fsx: Update copyright statement to be the APSL 2.0

To: xfs-masters@xxxxxxxxxxx
Subject: Re: [PATCH 6/6] fsx: Update copyright statement to be the APSL 2.0
From: Theodore Tso <tytso@xxxxxxx>
Date: Tue, 12 May 2009 14:01:53 -0400
Cc: akpm@xxxxxxxxxxxxxxxxxxxx, Dave Jones <davej@xxxxxxxxxx>
In-reply-to: <1242144865-6967-6-git-send-email-tytso@xxxxxxx>
References: <1242144865-6967-1-git-send-email-tytso@xxxxxxx> <1242144865-6967-2-git-send-email-tytso@xxxxxxx> <1242144865-6967-3-git-send-email-tytso@xxxxxxx> <1242144865-6967-4-git-send-email-tytso@xxxxxxx> <1242144865-6967-5-git-send-email-tytso@xxxxxxx> <1242144865-6967-6-git-send-email-tytso@xxxxxxx>
User-agent: Mutt/1.5.18 (2008-05-17)
Since this commit is as much about the commit description as the
patch, let me fix some typo's that snuck in...

On Tue, May 12, 2009 at 12:14:25PM -0400, Theodore Ts'o wrote:
> The fsx test program was originally written by Avadis Tevanian, Jr.,
> at NeXT Computer, and carried a Copyright 1991, "All Rights Reserved"
> statement.  It was later rewritten and released by Conrad Minshall at
> Apple in 1998.  Those were simpler times, when people didn't really
> worried about the lack of a copyright permission statement.

s/worried/worry/

> NeXT Computer was acquired by Apple Computer in 1997, and so Jordan
> Hubbard, who was an employee of Apple Computer, was (presumably)
> empowered to officially release the fsx source under the Apple Public
> Source License (APSL) 2.0.

Let's cut and paste this paragraph [1] to point "A" below


> The lineage of this particular version of fsx.c comes from port of the
> original Conrad Minshall code, which was fixed up by David Jones, made
> its way into Andrew Morton's ext3tools, and eventually was sucked into
> xfsprogs.  A more detailed version of the history and lineage of fsx
> can be found here:
> 
>       http://www.codemonkey.org.uk/projects/fsx/
> 
> The "original" bits of fsx are substantially unchanged from the
> FreeBSD variant of fsx which is released under the APSL 2.0.  The APSL
> is an open source license, and while it is not clear was the intent of
> those people who published various changes and enhancements to fsx,
> presumably they didn't mean "All Rights Reserved Copyright NeXT
> Computer".  In any case, splicing the the APSL permission statement
            ^^^  
             |
Point "A"   Insert the above paragraph [1] here

> seems reasonable for the original bits of fsx, and seems perfectly
> justifiable.
> 
> There is also a version of fsx that is officially released by Apple
> at:
> 
>       http://fstools.macosforge.org/trac/wiki/FsxTool
> 
> this version of fsx is released has copyright label "Copyright
> 1998-2009 Apple, Inc.", and is released under the APSL 2.0.  It has
> the Apple-authored changes, and is missing some of the enhancement
> that came from the FreeBSD fork of fsx.c.  This is probably the
> cloesest thing to "upstream" for fsx.c, but the macosforge mailing
> list is totally silent.  It might be worthwhile to see if they will be
> interested in taking contributions from the non-Apple forks of fsx.
> In any case, this additional evidence that the "core" original bits of
                  ^^^
                   |
                  Insert the verb "is" here

> fsx are available under an OSS license, specifically the APSL 2.0.
> 
> As far as I can determine, the bits of functionality which were added
> to the xfsprogs fork of fsx-11, that are not in the FreeBSD fork, and

s/xfsprogs/xfstests/
s/fsx-11/fsx/

> which we don't have clear copyright ownership information and
> licensing intent, are:
> 
> *) The flush option (-f)
> *) The prealloc option (-x)
> *) The do_fsync option (-y)
> *) The AIO option (-A)
> *) The O_DIRECT option (-Z)
> 
> According to the SCM logs, the AIO and O_DIRECT changes came from
> "outside", and were merged into xfsprogs in March 18, 2004.  These

s/xfsprogs/xfstests/

> appear to have come from Andrew Morton's ext3tools fork of fsx.  The
> do_fsync option also appears to have come from Andrew Morton's
> ext3tools version of fsx.
> 
> It appears that the flush option was merged in by Allan Ranall from
> SGI from the "relevant CXFSQA tests", and the prealloc option was
> added by the "fsgqa" pseudo-user (presumably from SGI) on August 1,
> 2003.
> 
> So assuming we can track down and get permission from SGI for the
> flush and prealloc options, and Andrew Morton is willing to sign off
> on releasing his additions to the ext3tools' version of fsx under an
> APSL-compatible license, we may have finally dealt with all of the
> copyright problems.
> 
> n.b.: The above is not Legal Advice; I am not a lawyer, and I do not
> play one TV; this is just my best guess/analysis of the situation.
> People who are uncomfortable with the licensing situation should have
> a Real Attorney check my results; I have supplied enough pointers that
> it should be relatively easy to pull the sources from the various
> forks of fsx.
> 
> Signed-off-by: "Theodore Ts'o" <tytso@xxxxxxx>
> Cc: akpm@xxxxxxxxxxxxxxxxxxxx
> Cc: Dave Jones <davej@xxxxxxxxxx>

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