Re: Opie in Angstrom

View: New views
4 Messages — Rating Filter:   Alert me  

Parent Message unknown Re: Opie in Angstrom

by Bugzilla from bluelightning@bluelightning.org :: Rate this Message:

Reply to Author | View Threaded | Show Only this Message

Paul Sokolovsky wrote:
> 1. There's an unpleasant bug uncovered itself in 1.2.3 release - on
> some device models (e.g., h3900, h4000, some Palm models but not hx4700)
> after start the screen is rotated 90 degrees. This used to work in
> prereleases, so current suspect is h2200 rotation patch which was
> committed shortly before release. It could be that it regressed
> rotation handling for other models.

Argh :(

The patch I committed shortly before release actually rips out the change
I made that caused unexpected rotation of the screen (which should have
been h2200-specific anyway) and instead just changes the way that the
cursor keys are interpreted, again for h2200 only. For reference, here's
the diff:

http://handhelds.org/cgi-bin/cvsweb.cgi/opie/libopie2/opiecore/device/odevice_ipaq.cpp.diff?r1=1.31&r2=1.32&f=h

Now, it's possible that some of the other changes we made to odevice*
caused this to happen. In any case, something must be wrong here, and I'm
not suggesting that I'm definitely not to blame for this :)

I'm pretty sure you can add h2200 to the list of devices that are not
affected by this issue. Interesting that you mention h3900 - is that
device in a state where it can happily run Opie using the 2.6 kernel? I
have not tried that for some time, perhaps I should.

> This is top priority issue to be fixed, IMHO.

Agreed. Unfortunately this is the first time I've heard about this
problem. I don't know how Erik feels but I am happy to fast-track 1.2.4 a
little bit if we are able to solve this problem soon. Alternatively I
would be fine with patching 1.2.3 in OE and also solving it for 1.2.4 when
it comes out, too, whatever works best.

> Note that it's not in the tracker.

Since you mention you don't have permanent net access I have now filed one
on the Opie side for this:

http://opie-bugs.oszine.de/view.php?id=1894

>   Point 2 is crucial to get support in Angstrom - let me remind,  that
> per Angstrom policies, Angstrom maintainers don't do build engineering
> themselves - device mentors do, so many people doing regular builds
> and maintaining them what "supported" means.

Right. However, if there are genuine bugs preventing Opie from being
supported I would like to hear about them so I can help solve them. I
guess I also need to get up to speed on Angstrom policies so that I can
contribute more directly (I hope they are well documented somewhere).

Cheers,
Paul
_______________________________________________

http://opie.handhelds.org/cgi-bin/moin.cgi/DeveloperWikiIndex

Opie-devel mailing list
Opie-devel@...
https://handhelds.org/mailman/listinfo/opie-devel

Re: Re: Opie in Angstrom

by Erik Hovland :: Rate this Message:

Reply to Author | View Threaded | Show Only this Message

On Wed, Oct 24, 2007 at 09:40:09AM +1300, Paul Eggleton wrote:
> Agreed. Unfortunately this is the first time I've heard about this
> problem. I don't know how Erik feels but I am happy to fast-track 1.2.4 a
> little bit if we are able to solve this problem soon. Alternatively I
> would be fine with patching 1.2.3 in OE and also solving it for 1.2.4 when
> it comes out, too, whatever works best.

I am willing to fast track 1.2.4. But I would likely want most of the
bugs attached to the release shifted to 1.2.5, since I will likely not
get to many of them before putting out a fast-tracked 1.2.4.

E

--
Erik Hovland
mail: erik@...
web: http://hovland.org/
PGP/GPG public key available on request

_______________________________________________

http://opie.handhelds.org/cgi-bin/moin.cgi/DeveloperWikiIndex

Opie-devel mailing list
Opie-devel@...
https://handhelds.org/mailman/listinfo/opie-devel

Re: Opie in Angstrom

by Paul Sokolovsky :: Rate this Message:

Reply to Author | View Threaded | Show Only this Message

Hello Paul,

Tuesday, October 23, 2007, 11:40:09 PM, you wrote:

> Paul Sokolovsky wrote:
>> 1. There's an unpleasant bug uncovered itself in 1.2.3 release - on
>> some device models (e.g., h3900, h4000, some Palm models but not hx4700)
>> after start the screen is rotated 90 degrees. This used to work in
>> prereleases, so current suspect is h2200 rotation patch which was
>> committed shortly before release. It could be that it regressed
>> rotation handling for other models.

> Argh :(

> The patch I committed shortly before release actually rips out the change
> I made that caused unexpected rotation of the screen (which should have
> been h2200-specific anyway) and instead just changes the way that the
> cursor keys are interpreted, again for h2200 only. For reference, here's
> the diff:

> http://handhelds.org/cgi-bin/cvsweb.cgi/opie/libopie2/opiecore/device/odevice_ipaq.cpp.diff?r1=1.31&r2=1.32&f=h

> Now, it's possible that some of the other changes we made to odevice*
> caused this to happen. In any case, something must be wrong here, and I'm
> not suggesting that I'm definitely not to blame for this :)

  Well, I'm not talking about blaming, I'm talking about figuring this
issue out.

> I'm pretty sure you can add h2200 to the list of devices that are not
> affected by this issue.

> Interesting that you mention h3900 - is that
> device in a state where it can happily run Opie using the 2.6 kernel? I
> have not tried that for some time, perhaps I should.

  Happily - depends. Otherwise, there's no pocketpc device which is
supported by 2.4 and not supported by 2.6. The worst is h3800 - SD support
needs revamp, etc. Still, it should at least boot. As for h3900,
battery support is not there for example.

>> This is top priority issue to be fixed, IMHO.

> Agreed. Unfortunately this is the first time I've heard about this
> problem. I don't know how Erik feels but I am happy to fast-track 1.2.4 a
> little bit if we are able to solve this problem soon. Alternatively I
> would be fine with patching 1.2.3 in OE and also solving it for 1.2.4 when
> it comes out, too, whatever works best.

  I don't have strong opinion here, except that it's worth to invest
time into solving this issue, as it is clearly on critical path
towards making a PR campaign on prebuilt OPIE availability for
Angstrom.

>> Note that it's not in the tracker.

> Since you mention you don't have permanent net access I have now filed one
> on the Opie side for this:

> http://opie-bugs.oszine.de/view.php?id=1894

>>   Point 2 is crucial to get support in Angstrom - let me remind,  that
>> per Angstrom policies, Angstrom maintainers don't do build engineering
>> themselves - device mentors do, so many people doing regular builds
>> and maintaining them what "supported" means.

> Right. However, if there are genuine bugs preventing Opie from being
> supported I would like to hear about them so I can help solve them.

  No, apart from that unpleasant, but of course not really critical
bug, all requests from Angstrom maintainers were satisfied, so it is
only a matter of community interest and support.

> I
> guess I also need to get up to speed on Angstrom policies so that I can
> contribute more directly (I hope they are well documented somewhere).

  Not really, that's resume of IRC communications. But I think stuff
being done is primary, and bureaucracy afterwards, IMHO. Not talking
that it all just settles.

> Cheers,
> Paul


--
Best regards,
 Paul                            mailto:pmiscml@...

_______________________________________________

http://opie.handhelds.org/cgi-bin/moin.cgi/DeveloperWikiIndex

Opie-devel mailing list
Opie-devel@...
https://handhelds.org/mailman/listinfo/opie-devel

Re: Opie in Angstrom

by Bugzilla from bluelightning@bluelightning.org :: Rate this Message:

Reply to Author | View Threaded | Show Only this Message

On Sat, 27 Oct 2007, you wrote:
>   Well, I'm not talking about blaming, I'm talking about figuring this
> issue out.

Of course. However I am currently stuck without a working device that exhibits
this issue, so I would appreciate any help from those who do have such
devices. In the mean time I guess I can try to just look at odevice* to see
if there is anything obviously wrong there.

>   Happily - depends. Otherwise, there's no pocketpc device which is
> supported by 2.4 and not supported by 2.6. The worst is h3800 - SD support
> needs revamp, etc. Still, it should at least boot. As for h3900,
> battery support is not there for example.

BTW I did try building an Angstrom image for h3900, and it also does not seem
to be initialising the screen correctly on my h3970 - the backlight comes on
but otherwise it just stays black. I didn't connect to the serial console to
see how far it was actually getting. (Not that this is especially important
to our current discussion.)

>   No, apart from that unpleasant, but of course not really critical
> bug, all requests from Angstrom maintainers were satisfied, so it is
> only a matter of community interest and support.

OK. To open up an old can of worms, I did read an old comment just the other
day on the htc-pxa mailing list by Mickey saying that he did not want to go
back to the confusion of qt and non-qt packages for the same thing. I sort of
understand this, but where does it leave us with things like roadmap, SDL
games, etc.? I don't think we can completely eliminate these and still
provide the applications that users want.

Cheers,
Paul
_______________________________________________

http://opie.handhelds.org/cgi-bin/moin.cgi/DeveloperWikiIndex

Opie-devel mailing list
Opie-devel@...
https://handhelds.org/mailman/listinfo/opie-devel