RE: [urgent] Parted 1.4.4

[Matt_Domsch]

> > The issue is that there's a CRC function which Intel uses.
> 
> This kind of stuff is probably insignificant for copyright purposes.
> (Obviously, we should talk to the FSF's lawyer, if we intend to
> use Intel code)
> 
> > Intel's EFI
> > Sample Implementation has a non-GPL-compatable license 
> (which Matt W.
> > caught).
> 
> Can we get them to change this?  I don't know anything about the
> politics here.

Yeah, right...  Already asked, already denied.

> > 
> > So no, it isn't public domain, it's copyrighted, with a 
> combination of "use
> > as you will" from Brown, and "All rights reserved" from ylo.
> 
> "All rights reserved" looks scary.

Yep.  What's there worth "reserving" on a single useful line of code which
is a function prototype?  Don't know.

> > 
> > I (and Dell) won't care about the 2 lines of code I 
> modified, nor I doubt
> > Brown or ylo will either.  But, IANAL, and am very confused here...
> 
> I think you're correct.  Eben Moglen (the FSF lawyer) said that code
> less than 30 lines is unlikely to be contentious.  (Obviously, it
> depends on the circumstances, like how much you fit on a line, etc.)
> 
> Aren't there any other (free) implementations of crc32?

Problem is the functional algorithm has to match what Intel's using.  It's
not just any CRC32, it's got to be the right CRC32 algorithm.

> 
> I did a search for "gnu crc32", and found a match in zlib, if that
> helps...
> 
>       http://www.info-zip.org/pub/infozip/zlib/
> 
> Is the crc32 thing here the same thing?

zlib-1.1.3 has a very similar-looking function, it'd take just a bit to
prove they're the same.  Of course, this is copyrighted too...

Interesting also to note that in the Red Hat kernel, the cipe patch
(linux/drivers/net/cipe) includes what appears to be an unmodified copy of
Brown's implementation.  That way we could avoid any conflict with ylo.  I'd
still need to make a 2-line modification, but that's allowed by Brown's
copyright.  And, I already know that this algorithm is functionally
identical to Intel's (the goal).

-Matt