> My concern would be that without it being an official target from
the
> start, we run the risk of brokenness being found late in the game.
> Does it seem possible to add an "official" designation to ARM (and
> i686 for that matter) if things prove to be working by whatever cutoff
> date we have? That would seem to put more impetus on the people doing
> the things you suggest without us declaring either of those
> architectures by default.
Yeah, I think that it would be an error to commit to it until there's
been a demonstration that this is achievable - and the onus should be
on the ARM people to demonstrate that. I don't want to end up with one
of our first deliverables being a sub-par experience. If it can't run
Shell reliably and with adequate performance then it buys us nothing to
ship it.
I agree with this sentiment, but like Dennis I wouldn't want to see it
excluded entirely because it wasn't discussed at the beginning.
> I'm not as optimistic as some when it comes to viable
accelerated
> graphics hardware on ARM in the F21 timeframe. If testing of
> Workstation ARM can't even begin until things are merged, and that
> happens at the tail end of the development window, I don't really want
> us to be stuck in the blocker/demotion game if it doesn't happen.
> Opportunistic "promotion" seems a decent compromise.
Not going to disagree.
There are two possible candidates of HW class that would (could) make
Workstation a nice option on ARM in the timeframe of F-21 but as both
are out of my control I'm not going to get excited until I have them
working in front of me let alone be naive to throw the hat in the
ring. We've got around 6 months until the release of F-21 if it's
something that evolves into that in a reasonable time in the interim
and the moving planets align at the right time with enough time I
believe it's worth reassessment then.
Peter