[RFC] meta-handheld

[RFC] meta-handheld

[Paul Eggleton]

Hi all,

I've discussed the possibility with a few people on IRC of creating a meta-
handheld layer for support of older handheld devices - this would include 
Zaurus, iPAQ, SimPad, etc; possibly even EZX phones if they aren't in a layer 
of their own. These machines share a fair amount of commonality and I don't 
think that they will all survive if each one has to be put into a BSP of its 
own. Response so far seems to be positive but I'd like to hear thoughts from 
others on this.

I have access to many of these devices, and this is something I am prepared to 
maintain on my own time, although I would appreciate help from others. I have 
the basis of a layer from what was left in oe-core and meta-extras but quite a 
lot of it would need to be brought over from OE.

If people agree that this is a good idea, the next question would be where to 
host it - is this something that could be within the meta-oe repository?

Cheers,
Paul

-- 

Paul Eggleton
Intel Open Source Technology Centre (UK)

Re: [RFC] meta-handheld

[Graeme Gregory]

On Thu, Apr 21, 2011 at 05:04:37PM +0100, Paul Eggleton wrote:
> Hi all,
> 
> I've discussed the possibility with a few people on IRC of creating a meta-
> handheld layer for support of older handheld devices - this would include 
> Zaurus, iPAQ, SimPad, etc; possibly even EZX phones if they aren't in a layer 
> of their own. These machines share a fair amount of commonality and I don't 
> think that they will all survive if each one has to be put into a BSP of its 
> own. Response so far seems to be positive but I'd like to hear thoughts from 
> others on this.
> 
> I have access to many of these devices, and this is something I am prepared to 
> maintain on my own time, although I would appreciate help from others. I have 
> the basis of a layer from what was left in oe-core and meta-extras but quite a 
> lot of it would need to be brought over from OE.
> 
> If people agree that this is a good idea, the next question would be where to 
> host it - is this something that could be within the meta-oe repository?
> 
Grabbing all of these barely supported machines into their own layer sounds
like a good plan to me.

Id suggest just hosting it on gitorious, if its inside OE then there is the
magic assumption from the public that they should work (which is what happens
now). Then people get really dissapointed when I tell them getting it to work
is likely to be a hard slog.

Graeme

Re: [RFC] meta-handheld

[Koen Kooi]

Op 21 apr 2011, om 18:57 heeft Graeme Gregory het volgende geschreven:

> On Thu, Apr 21, 2011 at 05:04:37PM +0100, Paul Eggleton wrote:
>> Hi all,
>> 
>> I've discussed the possibility with a few people on IRC of creating a meta-
>> handheld layer for support of older handheld devices - this would include 
>> Zaurus, iPAQ, SimPad, etc; possibly even EZX phones if they aren't in a layer 
>> of their own. These machines share a fair amount of commonality and I don't 
>> think that they will all survive if each one has to be put into a BSP of its 
>> own. Response so far seems to be positive but I'd like to hear thoughts from 
>> others on this.
>> 
>> I have access to many of these devices, and this is something I am prepared to 
>> maintain on my own time, although I would appreciate help from others. I have 
>> the basis of a layer from what was left in oe-core and meta-extras but quite a 
>> lot of it would need to be brought over from OE.
>> 
>> If people agree that this is a good idea, the next question would be where to 
>> host it - is this something that could be within the meta-oe repository?
>> 
> Grabbing all of these barely supported machines into their own layer sounds
> like a good plan to me.
> 
> Id suggest just hosting it on gitorious, if its inside OE then there is the
> magic assumption from the public that they should work (which is what happens
> now). Then people get really dissapointed when I tell them getting it to work
> is likely to be a hard slog.

I agree with Graeme on the hosting bit. I also have all the devices you mentioned near my desk, so testing should be easy :)

regards,

Koen

Re: [RFC] meta-handheld

[Richard Purdie]

On Thu, 2011-04-21 at 19:27 +0200, Koen Kooi wrote:
> Op 21 apr 2011, om 18:57 heeft Graeme Gregory het volgende geschreven:
> 
> > On Thu, Apr 21, 2011 at 05:04:37PM +0100, Paul Eggleton wrote:
> >> Hi all,
> >> 
> >> I've discussed the possibility with a few people on IRC of creating a meta-
> >> handheld layer for support of older handheld devices - this would include 
> >> Zaurus, iPAQ, SimPad, etc; possibly even EZX phones if they aren't in a layer 
> >> of their own. These machines share a fair amount of commonality and I don't 
> >> think that they will all survive if each one has to be put into a BSP of its 
> >> own. Response so far seems to be positive but I'd like to hear thoughts from 
> >> others on this.
> >> 
> >> I have access to many of these devices, and this is something I am prepared to 
> >> maintain on my own time, although I would appreciate help from others. I have 
> >> the basis of a layer from what was left in oe-core and meta-extras but quite a 
> >> lot of it would need to be brought over from OE.
> >> 
> >> If people agree that this is a good idea, the next question would be where to 
> >> host it - is this something that could be within the meta-oe repository?
> >> 
> > Grabbing all of these barely supported machines into their own layer sounds
> > like a good plan to me.
> > 
> > Id suggest just hosting it on gitorious, if its inside OE then there is the
> > magic assumption from the public that they should work (which is what happens
> > now). Then people get really dissapointed when I tell them getting it to work
> > is likely to be a hard slog.
> 
> I agree with Graeme on the hosting bit. I also have all the devices you mentioned near my desk, so testing should be easy :)

I'm going to disagree, this is exactly the kind of layer fragmentation
I'd not like to see layers cause. Which website do I go to where I can
see a list of layers that exist? Who is going to maintain that list? If
its off on gitorious nobody will be able to easily find it.

The easiest solution is if git.openembedded.org or git.yoctoproject.org
can host the majority of the meta-* repositories.

Equally, I don't really see why this can't be a specific meta- group in
meta-oe. The README can clearly spell out what the expectations of the
layer are...

Cheers,

Richard

Re: [RFC] meta-handheld

[Andrea Adami]

snip

>> >> If people agree that this is a good idea, the next question would be where to
>> >> host it - is this something that could be within the meta-oe repository?
>> >>
>> > Grabbing all of these barely supported machines into their own layer sounds
>> > like a good plan to me.
>> >
>> > Id suggest just hosting it on gitorious, if its inside OE then there is the
>> > magic assumption from the public that they should work (which is what happens
>> > now). Then people get really dissapointed when I tell them getting it to work
>> > is likely to be a hard slog.
>>
>> I agree with Graeme on the hosting bit. I also have all the devices you mentioned near my desk, so testing should be easy :)
>
> I'm going to disagree, this is exactly the kind of layer fragmentation
> I'd not like to see layers cause. Which website do I go to where I can
> see a list of layers that exist? Who is going to maintain that list? If
> its off on gitorious nobody will be able to easily find it.
>
> The easiest solution is if git.openembedded.org or git.yoctoproject.org
> can host the majority of the meta-* repositories.
>
> Equally, I don't really see why this can't be a specific meta- group in
> meta-oe. The README can clearly spell out what the expectations of the
> layer are...
>
> Cheers,
>
> Richard

I agree with Richard, for keeping it as meta- group.
About expectations, at least for Zaurus I don't see problems: vanilla
2.6.38 kernel, actual udev, imminent move to xorg-xserver if kdrive is
finally doomed.

Ipaq's seem suffering more bitrot...

Andrea

Re: [RFC] meta-handheld

[Paul Eggleton]

On Saturday 23 April 2011 00:09:38 Andrea Adami wrote:
> I agree with Richard, for keeping it as meta- group.
> About expectations, at least for Zaurus I don't see problems: vanilla
> 2.6.38 kernel, actual udev, imminent move to xorg-xserver if kdrive is
> finally doomed.
>
> Ipaq's seem suffering more bitrot...

This is a problem, yes. Most of the iPAQs require 2.6.21-hh which is ancient 
and unmaintained and as we know presents difficulties with modern udev. Some of 
the iPAQs will work with mainline but many functions will be unavailable. This 
is something I hope can be sorted out - surely by now a lot of the "ip blocks" 
used within iPAQs and other PDAs are now in mainline already and all that's 
necessary is some plumbing...?

Anyway, I'm concerned that if these machines get pushed out to some random 
repo it's only going to make things worse. I think we can mitigate the 
concerns about people rushing in only to find that the level of support for 
their device is less than they hoped simply by being crystal clear about each 
device's status in documentation (e.g. a wiki page), something which we have 
not been good at recently. (Am I volunteering to maintain this information as 
well? Naturally :) )

Cheers,
Paul

-- 

Paul Eggleton
Intel Open Source Technology Centre (UK)

Re: [RFC] meta-handheld

[Khem Raj]

On Sun, Apr 24, 2011 at 2:25 AM, Paul Eggleton
<paul.eggleton@linux.intel.com> wrote:
> On Saturday 23 April 2011 00:09:38 Andrea Adami wrote:
>> I agree with Richard, for keeping it as meta- group.
>> About expectations, at least for Zaurus I don't see problems: vanilla
>> 2.6.38 kernel, actual udev, imminent move to xorg-xserver if kdrive is
>> finally doomed.
>>
>> Ipaq's seem suffering more bitrot...
>
> This is a problem, yes. Most of the iPAQs require 2.6.21-hh which is ancient
> and unmaintained and as we know presents difficulties with modern udev. Some of
> the iPAQs will work with mainline but many functions will be unavailable. This
> is something I hope can be sorted out - surely by now a lot of the "ip blocks"
> used within iPAQs and other PDAs are now in mainline already and all that's
> necessary is some plumbing...?
>
> Anyway, I'm concerned that if these machines get pushed out to some random
> repo it's only going to make things worse. I think we can mitigate the
> concerns about people rushing in only to find that the level of support for
> their device is less than they hoped simply by being crystal clear about each
> device's status in documentation (e.g. a wiki page), something which we have
> not been good at recently. (Am I volunteering to maintain this information as
> well? Naturally :) )
>

we can host it on its own repo and not under meta-openembedded
umbrella. As long as it is maintained its good


> Cheers,
> Paul
>
> --
>
> Paul Eggleton
> Intel Open Source Technology Centre (UK)
>
> _______________________________________________
> Openembedded-core mailing list
> Openembedded-core@lists.openembedded.org
> http://lists.linuxtogo.org/cgi-bin/mailman/listinfo/openembedded-core
>