NLUUG/ELCE: Embedded Linux and the community [LWN.net]

NLUUG/ELCE: Embedded Linux and the community

Posted Nov 13, 2008 14:32 UTC (Thu) by dwmw2 (subscriber, #2063)
In reply to: NLUUG/ELCE: Embedded Linux and the community by Jaffa
Parent article: NLUUG/ELCE: Embedded Linux and the community

Thanks for that. You're right that I didn't spend a huge amount of time on it, but neither should I have to. This is what I actually did:

On asking Google for "maemo kernel source" I was taken to a wiki page which said "THIS ARTICLE IS ONLY PARTLY UPDATED - DON'T EXPECT IT TO WORK", and had a link to another page which seemed to be telling me that I would have to install and use both ScratchBox and apt-get — which put it firmly beyond my attention span.

It would be great if the kernel was available in some form other than a distribution-specific package, and a little easier to find.

(Log in to post comments)

NLUUG/ELCE: Embedded Linux and the community

Posted Nov 13, 2008 14:44 UTC (Thu) by Jaffa (guest, #4327) [Link]

Thanks for the response. Though, I will take exception to:

You're right that I didn't spend a huge amount of time on it, but neither should I have to.

(emphasis added)

At the moment, none of the use cases for the existing maemo.org - or the version to replace it - have "kernel developer who wants to do some investigation work on how close to upstream the Maemo kernel is". Personally (and no offence to you, or the very interesting article that this is), I don't think that's a use case we would want to support particularly.

The important thing (IMHO) is that someone wanting to do kernel hacking for their Maemo device can get, patch and rebuild the kernel. This is doable (evident from the number of community members who have done exactly that), although the procedure may be inadequate (anything involving Scratchbox is, unless you've already got it setup).

Maemo use cases

Posted Nov 13, 2008 15:47 UTC (Thu) by corbet (editor, #1) [Link]

At the moment, none of the use cases for the existing maemo.org - or the version to replace it - have "kernel developer who wants to do some investigation work on how close to upstream the Maemo kernel is".

But...I think part of what David is trying to say is that "make it easier to get your useful changes back into the mainline" is a use case that projects like Maemo should be supporting. Making your kernel sources hard to find does not help in that regard.

Maemo use cases

Posted Nov 13, 2008 15:54 UTC (Thu) by Jaffa (guest, #4327) [Link]

"make it easier to get your useful changes back into the mainline" is a use case that projects like Maemo should be supporting. Making your kernel sources hard to find does not help in that regard.

Agreed with the first part. The second part is interesting - AIUI (although I'm not a kernel hacker on the Maemo kernel) the Maemo kernel is upstream with limited or no patches.

Nokia primarily do all the kernel work themselves, upstream in the omap tree. Making the kernel sources easier to find (although, come on, from maemo.org it wasn't that hard - the Googability of it notwithstanding) doesn't help get stuff upstream: upstream kernel hackers rarely go scouting around for patches. Instead, they get pushed stuff from people working in those trees day-in, day-out.

Maemo use cases

Posted Dec 3, 2008 5:56 UTC (Wed) by HalfMoon (guest, #3211) [Link]

upstream in the omap tree

That is, not upstream yet.

Maemo use cases

Posted Nov 14, 2008 11:09 UTC (Fri) by dneary (subscriber, #55185) [Link]

In that case, to be fair to Maemo, it's worth noting that Nokia ranked in the top 5 contributors to 2.6.27, ahead of Red Hat and Novell. Nokia's kernel guys are actually working quite well with upstream right now.

The kernel is one of the components of the Maemo platform which is developed essentially by Nokia engineers (working upstream mostly - see the omap list for a good example) rather than a community developed component.

Maemo use cases

Posted Nov 14, 2008 11:21 UTC (Fri) by dwmw2 (subscriber, #2063) [Link]

Nokia's kernel guys are actually working quite well with upstream right now.

This is a very good point — and although Jake doesn't mention it, I was careful to make precisely the same point on Friday during my presentation.

Nokia do very good work on the kernel, contributing heavily in areas like Bluetooth and flash storage (to mention just the parts I've noticed directly).

NLUUG/ELCE: Embedded Linux and the community

Posted Nov 14, 2008 9:24 UTC (Fri) by nhippi (subscriber, #34640) [Link]

To be fair, googlin "fedora kernel" takes me to unifficial howto page that tells me I would have to use yum and rpm and doesn't explain what to do if I happen to use something else that fedora. The next usefull looking search result on fedora wiki is pretty much identical.

Fedoras upstream track record is one of the best in communities, so I don't think googleability of sources is a valid indicator at all. Considering community is a social thing, perhaps asking people would better way ;) Then someone could have told you that it's not "maemo kernel" you want, it's the Linux-omap GIT tree and the omap vger mailing list.