Autobuilder bugs versus stable branch being updated

Autobuilder bugs versus stable branch being updated

[Philip Balister]

[-- Attachment #1: Type: text/plain, Size: 343 bytes --]

I think it is a but of stretch to close old autobuilder bugs with the 
comment the stable branch will not be updated.

These two issues are unrelated.

Autobuilder failures should be logged in tinderbox since they are not 
getting attention in bugzilla.

The stable branch updates are handled by users of the stable branch.

Philip

[-- Attachment #2: S/MIME Cryptographic Signature --]
[-- Type: application/x-pkcs7-signature, Size: 3303 bytes --]

Re: Autobuilder bugs versus stable branch being updated

[Rolf Leggewie]

Philip,

thank you for your comment.

Philip Balister wrote:
> I think it is a but of stretch to close old autobuilder bugs with the 
> comment the stable branch will not be updated.

That should not be the case (unless of course they were both of type 
.stable and -autobuild).

> These two issues are unrelated.

Agreed.

All closed bugs should have received a comment with an URL for further 
information as to why they were closed.

If you think there is a problem, please let me know a sample bug number 
and I'll see if there was anything that went wrong (which can of course 
happen when mass-handling bugs).

If there was an error, I'll gladly fix that.  If you think the closure 
message was too short, I can add a more verbose comment to all or a 
collection of the closed bugs.

Regards

Rolf

Re: Autobuilder bugs versus stable branch being updated

[Bernhard Guillon]

Rolf Leggewie wrote:
> Philip,
>
> thank you for your comment.
>
> Philip Balister wrote:
>> I think it is a but of stretch to close old autobuilder bugs with the 
>> comment the stable branch will not be updated.
>
> That should not be the case (unless of course they were both of type 
> .stable and -autobuild).
>
We should create a new stable branch. The old one drifted too much away 
from .dev which makes it hard to backport things from .dev to stable.

Best regards
Bernhard Guillon

Re: Autobuilder bugs versus stable branch being updated

[Koen Kooi]

On 17-02-09 20:54, Bernhard Guillon wrote:
> Rolf Leggewie wrote:
>> Philip,
>>
>> thank you for your comment.
>>
>> Philip Balister wrote:
>>> I think it is a but of stretch to close old autobuilder bugs with the
>>> comment the stable branch will not be updated.
>>
>> That should not be the case (unless of course they were both of type
>> .stable and -autobuild).
>>
> We should create a new stable branch. The old one drifted too much away
> from .dev which makes it hard to backport things from .dev to stable.

There's already work being done to do that, hold your horses please.

Re: Autobuilder bugs versus stable branch being updated

[Tom Rini]

On Tue, Feb 17, 2009 at 08:54:46PM +0100, Bernhard Guillon wrote:
> Rolf Leggewie wrote:
>> Philip,
>>
>> thank you for your comment.
>>
>> Philip Balister wrote:
>>> I think it is a but of stretch to close old autobuilder bugs with the 
>>> comment the stable branch will not be updated.
>>
>> That should not be the case (unless of course they were both of type  
>> .stable and -autobuild).
>>
> We should create a new stable branch. The old one drifted too much away  
> from .dev which makes it hard to backport things from .dev to stable.

The problem really is a policy one.  The old stable was "angstrom
snapped for 2007 here".  That's one way to do a stable branch, but not
what everyone thinks when they read "stable".  How much would it hurt to
have OE actually do releases, rather than to be, in kernel terms, mostly
like Andrew's tree (once stuff seems to work for some people, pull it
in, see what breaks)?

And of course, who would volunteer to actually try and maintain such
things and get everyone using the various branches that would have to
entail, correctly?

-- 
Tom Rini

Re: Autobuilder bugs versus stable branch being updated

[Bernhard Guillon]

Koen Kooi wrote:
>> We should create a new stable branch. The old one drifted too much away
>> from .dev which makes it hard to backport things from .dev to stable.
>
> There's already work being done to do that, hold your horses please.
>
>
Great :)
Is there some kind of todo list or masterplan.
How can I help?

Best regards
Bernhard Guillon

Re: Autobuilder bugs versus stable branch being updated

[Junqian Gordon Xu]

On 02/17/2009 03:55 PM, Tom Rini wrote:
> On Tue, Feb 17, 2009 at 08:54:46PM +0100, Bernhard Guillon wrote:
>> Rolf Leggewie wrote:
>>> Philip,
>>>
>>> thank you for your comment.
>>>
>>> Philip Balister wrote:
>>>> I think it is a but of stretch to close old autobuilder bugs with the 
>>>> comment the stable branch will not be updated.
>>> That should not be the case (unless of course they were both of type  
>>> .stable and -autobuild).
>>>
>> We should create a new stable branch. The old one drifted too much away  
>> from .dev which makes it hard to backport things from .dev to stable.
> 
> The problem really is a policy one.  The old stable was "angstrom
> snapped for 2007 here".  That's one way to do a stable branch, but not
> what everyone thinks when they read "stable".  How much would it hurt to
> have OE actually do releases, rather than to be, in kernel terms, mostly
> like Andrew's tree (once stuff seems to work for some people, pull it
> in, see what breaks)?

+1 for releases instead of snap short, which was proven not working for 
the users.

> 
> And of course, who would volunteer to actually try and maintain such
> things and get everyone using the various branches that would have to
> entail, correctly?
> 

I can maintain the user space stuffs and communicate/bug core members 
for stuff that I do not understand. In terms of machines, I will keep an 
eye on tosa and spitz. I've been away for a while since the git 
conversion (now half-way through the git manual), but I always try to 
keep up with the mailing lists.

Regards
Gordon

Re: Autobuilder bugs versus stable branch being updated

[Sledz, Steffen]

> > The problem really is a policy one.  The old stable was "angstrom
> > snapped for 2007 here".  That's one way to do a stable 
> > branch, but not what everyone thinks when they read "stable".
> > How much would it hurt to have OE actually do releases, rather
> > than to be, in kernel terms, mostly like Andrew's tree
> > (once stuff seems to work for some people, pull it
> > in, see what breaks)?
> 
> +1 for releases instead of snap short, which was proven not 
> working for the users.

+1 or +42 if this would be possible ;-)