Formal request for approval of the multi-desktop DVD (Re: Board meeting recap, 2010 10-18)
Matt Domsch
matt at domsch.com
Thu Oct 21 14:51:08 UTC 2010
A number of concerns have been raised, I'd like to see these closed
out.
* It's coming in way late. This I agree with, and it may be enough to
be a showstopper for this release. But it doesn't have to be, if
rel-eng isn't expected to produce it, and if QA isn't expected to
test it. Aside from being possible now, here is nothing put forward
that would prevent this from being held until the F15 release, like
we do with other "features" - and I do think changing the
delivery model can be considered a feature change.
* Forget about calling this a remix. If we want our ambassadors to
hand this out at events, EMEA or elsewhere, it needs to be produced
by and for the Project as a whole. The EMEA ambassadors may be the
actors getting the ISOs pressed, and they may be the ones doing to
ISO creation and testing, but the end result must be a
Project-produced item, not treated like a 3rd-party produced item.
* GNOME vs Desktop naming, seems resolved or easily resolvable. Mo is
right though that we would want to see a good user experience, and
that being late lends itself to a more hasty, and possibly
sub-optimal experience. Given that the design team has had a hand
in producing the other sleeves and labels, it's fair to include them
in the production of such for this too (I could digress into
unfunded mandates here, when it's late in the cycle and everyone is
stressed...)
* Source distribution. This is an aggregation of the existing spin
images, so there are no _new_ obligations. Source is in the
Everything/ tree, and in git. We don't currently publish an ISO of
SRPMS corresponding to the contents of the other spins. I do still
encourage people to add their srpm lists to the correspondingsource
git tree on fedorahosted though.
* I'd argue that having QA test it is important. I agree the test plan
need not be huge and complex, and need not duplicate all the other
testing of the individual spins, but some formal test plan, covering
the difference between the existing spins and this, is critical.
And someone needs to both write the test plan, and find the manpower
to carry out those tests. (I could digress into unfunded mandates
here, when it's late in the cycle and everyone is stressed...)
* I don't see a need for Infrastructure to host the ISO for
downloadability, but I think we could. It's 5.2GB, which could pose
a problem in our serving environment (files > 4GB tend to cause
problems for apache), but alt.fp.o is currently running x86_64
apache, which should not have a problem with large files.
* How will liveusb-creator handle being given this image? Will it
still work? It doesn't have to, but it'd be nice to know if it'll
break...
So, my gut feeling is, this is possible, but way late, and is
introducing stress into the system exactly when it shouldn't be.
Could we pull it off? Likely yes. Will it be bumpy? Likely yes (as
this conversation has already shown).
Christoph, how can you and others help alleviate these stresses, so to
produce this now, rather than waiting for F15?
Thanks,
Matt
More information about the advisory-board
mailing list