+++ Dave Martin [2011-12-02 11:19 +0000]:
On Thu, Dec 1, 2011 at 7:14 PM, David Zinman david.zinman@linaro.org wrote:
A request has been received to discontinue Linaro's support for the Beagleboard and Beagleboard-xM hardware.
Although they are starting to be replaced, Beagle and Beagle xM are particularly popular boards in the community at large. What is the rationale behind discontinuing support for these boards at this particular point in time? Will we still continue to produce best-effort "community" builds (as for some other boards)?
EOL-ing popular boards could weaken our links to the community, so we need to think carefully about it.
Indeed. Part of the benefit of Free Software is that you don't get support for hardware dropped at the whim of manufacturers who only care about their latest, greatest products.
Now of course linaro's focus is on the newer stuff to a large degree, not least because that's our member's focus and they are ultimately paying the bills, but we are primarily here as an interface between members and their corporate 'only the _next_ product counts' view of the world and the community/users/engineers who care about what they have on their desks, and getting work done with it.
So ideally, at this point we would have successfully upstreamed everything of significance on beagleboards and could leave it to others to maintain suitable distro support. Is that in fact the case? (I admit I haven't been following the details at all (I have, and thus only care about, nslu2/panda/n900/freerunner :-) )
It seems that both Debian and Ubuntu beagleboard installs are available: http://elinux.org/BeagleBoardDebian http://elinux.org/BeagleBoardUbuntu but it's not clear (to me) to what degree this support is in-distro. Should we be helping complete that work so that there really is proper in-distro support. I'm not convinced that our work is really done until that's the case, but we could declare that we only upstream stuff. distro-support is distro's problem.
I'm not sure we've really thought that much about the criteria for declaring a board 'done' in the sense that Linaro has done its bit of knwoledge and code-transfer from the vendors to upstreams and distros. This issue is going to come up regularly so we probably ought to work out what the rules are.
I think it's quite important that we don't act just like pure vendors and drop stuff the moment there is a new thing to care about: that's not being a good community member. But it is our job to get shot of that maintenance as soon as it's been successfully transferred. There is clearly a continuum between very popular indeed hardware like beagleboard at one end and snowball v3 (which never got out into the wild) at the other, and it's reasonable not to treat them the same.
If we really did only send code from vendors to upstreams this would be simpler, but we've also put a lot of effort into making board images which is distro-like behaviour. Both aspects of our work need to be considered: perhaps they should even have different end-dates?
Are we happy that our job is done in this case and users are not being (unduly) left in the lurch?
Do we have our 'end-of-support' criteria written down somewhere? I think they should contain something more than 'member says so', which seems to be where we are currently at. But if in fact that really is how it's going to work then we should say so.
Wookey