[Date Prev][Date Next] [Thread Prev][Thread Next] [Date Index] [Thread Index]

Re: Bits (Nybbles?) from the Vancouver release team meeting

Hi Dann,

On Sun, Mar 13, 2005 at 11:51:30PM -0700, dann frazier wrote:

> > - the release architecture must be publicly available to buy new

> I'd like to see if we can clarify this text.  I can see some of the
> problems this requirement would solve, such as guaranteed availability
> of replacement parts for Debian machines, etc.  However, I wouldn't want
> to see us drop an arch with a large user community because a company
> decides to discontinue a product.

> For example, although HP announced they are going to discontinue alpha,
> there maybe a large number of alpha users/developers who want to see
> stable releases continue, even after debian could no longer purchase a
> new one[1].

> If its really some side-effect of the discontinuation that's an issue,
> let's list that side-effect instead; e.g., if the real issue is that we
> need to have a reliable source for spare parts or replacement machines,
> lets say that - there are other ways to assure part availability.  If
> this requirement is a guideline to help us get down to a target number
> of architectures, lets explicitly say that, versus listing it as a firm
> requirement.

Sure; I believe there are actually a couple of requirements that are
being conflated here, the first being the availability of replacement
hardware, and the second being concerns about having hardware that's
actually able to keep up with the build queues since the buildd load is
increasing pretty steadily over time.

But the second requirement has already been stated separately, as "no
more than 2 buildds required to keep up"; and if we require that to
remain true throughout the preparation of the stable release, it will
certainly be true in terms of keeping up with stable build
*post*-release, which are of much, much smaller volume; so it shouldn't
matter if we can't buy more new hardware, if we know we already have
enough of it to keep up with the present release.

That leaves this as a proxy for "must be able to replace machines that
fail".  I think I'd prefer to let the ftpmaster folks who came up with
this criterion speak to this one, as I have only an outsider's view of
the actual hardware failures that the project has been fighting with.

> [1] I have no idea how many alpha users we have, nor how many want to
> have a stable release - this is just an example.

FWIW, until I talked with people about it on IRC, I did rather expect
that alpha would fail this requirement by the time etch was released,
but it seems that's not the case after all.  Color me pleasantly
surprised as an alpha aficionado, but all the same I don't actually run
stable on my alpha right now, so it doesn't directly affect me either.

Steve Langasek
postmodern programmer

Attachment: signature.asc
Description: Digital signature

Reply to: