Re: [PATCH 1/2] meta-isar: Separate images per MACHINE

[Claudius Heine <claudius.heine.ext@siemens.com>]

Hi Jan,

On 14/03/2019 08.28, Jan Kiszka wrote:
> On 14.03.19 08:21, [ext] Claudius Heine wrote:
>> Hi Maxim,
>>
>> On 13/03/2019 17.42, Maxim Yu. Osipov wrote:
>>> Hi everybody,
>>>
>>> Any feedback on this patch?
>>> This is a "fast track" patch as it fixes problem which delayed the 
>>> release.
>>>
>>> Thanks,
>>> Maxim.
>>>
>>> On 3/12/19 9:27 PM, Maxim Yu. Osipov wrote:
>>>> Image directory gets overwritten when running for two targets
>>>> with the same pair DISTRO and DISTRO_ARCH, resulting
>>>> start_vm script failure.
>>>>
>>>> Note:
>>>> This patch affects the bitbake multiconfig target calling syntax:
>>>> PREVIOUS: "multiconfig:$MACHINE-$DISTRO:isar-image-base"
>>>> NEW: "multiconfig:$MACHINE-$DISTRO:isar-image-base-$MACHINE"
>>>>
>>>> Suggested-by: Jan Kiszka <jan.kiszka@siemens.com>
>>>> Signed-off-by: Maxim Yu. Osipov <mosipov@ilbers.de>
>>>> ---
>>>>   meta-isar/conf/conf-notes.txt                    |  6 ++--
>>>>   meta-isar/recipes-core/images/isar-image-base.bb |  2 ++
>>>>   meta-isar/recipes-core/images/isar-image-ubi.bb  |  2 ++
>>>>   scripts/ci_build.sh                              | 40 
>>>> ++++++++++++------------
>>>>   scripts/start_vm                                 | 16 +++++-----
>>>>   5 files changed, 35 insertions(+), 31 deletions(-)
>>>>
>>>> diff --git a/meta-isar/conf/conf-notes.txt 
>>>> b/meta-isar/conf/conf-notes.txt
>>>> index 87bd2dc..84049e1 100644
>>>> --- a/meta-isar/conf/conf-notes.txt
>>>> +++ b/meta-isar/conf/conf-notes.txt
>>>> @@ -1,4 +1,4 @@
>>>>   Common targets are:
>>>> -    multiconfig:qemuarm-stretch:isar-image-base
>>>> -    multiconfig:qemuamd64-stretch:isar-image-base
>>>> -    multiconfig:rpi-jessie:isar-image-base
>>>> +    multiconfig:qemuarm-stretch:isar-image-base-qemuarm
>>>> +    multiconfig:qemuamd64-stretch:isar-image-base-qemuamd64
>>>> +    multiconfig:rpi-jessie:isar-image-base-rpi
>>
>> I am not a big fan of having to specify the machine when building a 
>> image.
>>
>> Would it be possible to have virtual recipes?
>>
>> IMO all isar-image-base-* would provide a isar-image-base.
> 
> That would be useful if it has no unexpected side effects. Can we 
> exclude that we end up with multiple providers in some scenario? It 
> seems so, but I'm not 100% sure ATM.

IIUC want is needed here is that the workdir of the images contain the 
machine. We can do that indirectly via modifying the PN or directly via 
modifying the WORKDIR.

Maxim choose the first option in this patch and that changes how the 
target name of that recipe as well. I don't know why he choose to change 
the PN, but if that was what is required for other reasons, then ok. But 
I am not in favor of changing the target name since that breaks the UX. 
My suggestion is just the easiest way to fix the UX.

We will end up with multiple providers if someone writes recipes that 
provide the same target name, we cannot really do anything about this.

IMO this patch should not be merged from one rc to the next, because it 
breaks the user interface. Renaming the core targets is a pretty big change.

Cheers,
Claudius

-- 
DENX Software Engineering GmbH,      Managing Director: Wolfgang Denk
HRB 165235 Munich, Office: Kirchenstr.5, D-82194 Groebenzell, Germany
Phone: (+49)-8142-66989-54 Fax: (+49)-8142-66989-80 Email: ch@denx.de