Re: [meta-ti][master][PATCH] weston: Update to 15.0.0

On 4/22/2026 4:30 AM, Ernest Van Hoecke wrote:
> On Tue, Mar 24, 2026 at 02:43:27PM -0400, Denys Dmytriyenko wrote:
>> On Tue, Mar 24, 2026 at 01:18:32PM -0500, Ryan Eatmon via lists.yoctoproject.org wrote:
>>>
>>>
>>> On 3/24/2026 11:44 AM, Denys Dmytriyenko wrote:
>>>> On Tue, Mar 24, 2026 at 11:20:04AM -0500, Ryan Eatmon via lists.yoctoproject.org wrote:
>>>>> oe-core has bumped the version to 15.0.0 [1].
>>>>
>>>>
>>>>> We need to adjust the bbappend version to % to better match future version
>>>>> changes.
>>>>
>>>> You shouldn't be doing this rename.
>>>>
>>>>
>>>>> Additionally, the patch needs to be updated to match the weston code
>>>>> changes.
>>>>
>>>> The bbappend is version-specific on purpose exactly due to the fact that the
>>>> patch won't apply against different versions - it has to be massaged every
>>>> time there's a major version bump, we've done it many times before.
>>>
>>> But... if the patch won't apply then the build will fail.  But it
>>> fails in a manner that is more obvious then just an upfront error
>>> about the bbappend not applying to a recipe.
>>>
>>> Either way, with the new bbappend logic they put in, the build will
>>> not go through.  So what does it really matter if the bbappend if
>>> version specific or not?
>>
>> Yes, the build will fail in both cases now.
>>
>> I guess from my perspective, getting a patch failing to apply all of a sudden
>> is less clear indication that a major version update just happened upstream.
>> Versus an error message saying no recipe is available for this versioned
>> bbappend. Moreover, do_patch() error happens much further down the build
>> process, vs. a dangling bbappend error comes very early during parsing...
>>
>> On the other hand, with your way, if the patch happens to apply cleanly to the
>> next major version, no intervention will be required.
>>
>> So, pros and cons for both ways - I'll be fine with either...
>>
> 
> Hi Ryan,
> Hi Denys,
> 
> We have our own weston_9.0.0.bb recipe in a layer that is used alongside
> meta-ti. Because of this rename, the bbappend from meta-ti started to
> apply onto this recipe as well, and of course the patch is not
> compatible with that version of weston.
> 
> Probably we should not be keeping such an old version around, but while
> I initially also thought that it is just a choice of which error, this
> situation shows that there is a real reason to keep the bbappend version
> specific. It's not just about getting an error during parsing vs during
> patching (which personally I do prefer) but also about not applying to
> the wrong version where the patch might be irrelevant. Please let me
> know if I misunderstood the situation here.

You are spot on.  And this is a good reason to keep the bbappend version 
specific.  I'll send a patch to fix this.


> Thanks and kind regards,
> Ernest
>   

-- 
Ryan Eatmon                [email protected]
-----------------------------------------
Texas Instruments, Inc.  -  LCPD  -  MGTS