Re: ZFS in Buster

[Ian Jackson <ijackson@chiark.greenend.org.uk>]

Mo Zhou writes ("Re: ZFS in Buster"):
> I made a mistake at this point. There is no SIMD bug in zfs
> 0.7.12-2. The true bug lies in the stable kernel update that
> breaks stuff. We debian ZoL maintainers decided to do nothing
> before the Buster release, and file an RC bug against the
> kernel when 10.1 comes out.

I am hoping I have misunderstood.

Here is what I read from your message:

 * Prior to Daniel (ganchya@gmail.com)'s message of last Tuesday, the
   Debian ZoL maintainers were aware of this precise difficulty
   surrounding the upstream __*fpu* symbols and ZoL.

 * The Debian ZoL maintainers collectively knew that this problem
   would not affect buster immediately, but would very likely affect a
   Linux kernel version which the kernel team would want to ship in a
   buster point releease.

 * The intention seems perhaps to have been to allow this latent
   problem to release with buster; and, then, to use the "released"
   status of ZoL in buster contrib, as a lever to try to have the
   kernel symbol change reverted in Debian's version of a forthcoming
   buster Linux kernel update, in that expected buster point
   release. [1]

 * The Debian ZoL maintainers did not seek a conversation with Debian
   kernel maintainers or the wider Debian project about this issue.

 * This lack of communication was deliberate.

I have a lot of respect for the energy you personally have brought to
your Debian work and the contributions you have made.  But I would
find behaviour such as I have described, if that is what occurred,
totally unacceptable.

If any of us in Debian become aware that any kind of problem or
adverse interaction will arise between our work, and that of other
contributors, it is our duty to promptly and candidly inform those
other contributors.

That is true even if we think those other contributors may disagree
with us, and if their likely response will be difficult for us.
Indeed, if we expect that their likely response will be adverse,
hiding the information is *more* rather than less serious.

It is very much Not OK to try to create Facts On The Ground while our
co-contributors remain in ignorance of our intent.

Please tell me I have misunderstood your message.

Regretfully,
Ian.

[1] As a matter of practical politics, I think any such strategy would
be clearly doomed, and not just because of the evidence of bad faith,
but also because Debian's overall attitude to GPL compliance issues
like this one is, as others have noted, very firm, and because of the
secondary status of contrib.

-- 
Ian Jackson <ijackson@chiark.greenend.org.uk>   These opinions are my own.

If I emailed you from an address @fyvzl.net or @evade.org.uk, that is
a private address which bypasses my fierce spamfilter.