[Buildroot] [PATCH 1/1] package/panel-mipi-dbi-firmware: new package

[Fiona Klute]

Am 25.01.25 um 12:49 schrieb Julien Olivain:
> Hi Fiona,
>
> On 24/01/2025 22:03, Fiona Klute wrote:
>> Hi Julien!
>>
>> Am 24.01.25 um 21:05 schrieb Julien Olivain:
>>> Hi Fiona,
>>>
>>> Thanks for the patch!
>>>
>>> On 23/01/2025 15:40, Fiona Klute via buildroot wrote:
>>> [...]
>>>> diff --git a/package/panel-mipi-dbi-firmware/panel-mipi-dbi-
>>>> firmware.mk b/package/panel-mipi-dbi-firmware/panel-mipi-dbi-
>>>> firmware.mk
>>>> new file mode 100644
>>>> index 0000000000..c3a9fa7afc
>>>> --- /dev/null
>>>> +++ b/package/panel-mipi-dbi-firmware/panel-mipi-dbi-firmware.mk
>>>> @@ -0,0 +1,35 @@
>>>> +################################################################################
>>>> +#
>>>> +# panel-mipi-dbi-firmware
>>>> +#
>>>> +################################################################################
>>>> +
>>>> +PANEL_MIPI_DBI_FIRMWARE_VERSION =
>>>> 1cbd40135a8c7f25d7b444a7fac77fd3c3ad471e
>>>> +PANEL_MIPI_DBI_FIRMWARE_SITE = https://github.com/notro/panel-mipi-
>>>> dbi.git
>>>> +PANEL_MIPI_DBI_FIRMWARE_SITE_METHOD = git
>>>> +PANEL_MIPI_DBI_FIRMWARE_LICENSE = CC0-1.0
>>>> +# license info is directly in the only source file
>>>> +PANEL_MIPI_DBI_FIRMWARE_LICENSE_FILES = mipi-dbi-cmd
>>>> +
>>>> +PANEL_MIPI_DBI_FIRMWARE_DEPENDENCIES = host-python3
>>>
>>> The script mipi-dbi-cmd does not seem to depend on extra host python
>>> modules. See:
>>> https://github.com/notro/panel-mipi-dbi/blob/main/mipi-dbi-cmd#L15
>>>
>>> Do you think we could use BR2_PYTHON3_HOST_DEPENDENCY instead? See:
>>> https://gitlab.com/buildroot.org/buildroot/-/commit/
>>> b98062f7301bc72d2a59cca733f36525405d3bb9
>>>
>>> This could save the time of the host-python compilation, in some
>>> situation.
>>>
>>> What do you think?
>> In principle that should work, but as far as I can see I'd have to add
>> some logic to search the Python interpreter in that case:
>> $(HOST_DIR)/bin/python, python3, or python?
>>
>> I suppose the elegant solution would be to store the detected
>> interpreter in a make variable from
>> support/dependencies/check-host-python3.mk, and the host-python3 one
>> otherwise. I can add a patch to do that, but the naming scheme varies
>> between the other check-host-* files. Would PYTHON3 be acceptable, or
>> should I use BR2_PYTHON3? Something else?
>>
>> Or would it be acceptable to just use "python3" and break in the (likely
>> rare) case someone only has "python"? I don't really like that option.
>
> My understanding of PEP 394:
> https://peps.python.org/pep-0394/
> is that a system with the "python" binary alone could only be a old
> Python 2 interpreter (which would not be sufficient and would build
> a host-python3).
Thanks, I wasn't aware that's an actual standard, not just common
practice. In that case it's indeed simple, I've just sent the patch. :-)

Best regards,
Fiona