On Wed, 2024-07-17 at 10:32 +0800, Song, Jiaying (CN) via 
lists.openembedded.org wrote:
> From: Jiaying Song <[email protected]>
> 
> The /usr/bin/python cannot find the correct Python interpreter, so a symbolic
> link is added to point it to the Python3 interpreter.
> 
> Signed-off-by: Jiaying Song <[email protected]>
> ---
>  meta/recipes-devtools/python/python3/python3-manifest.json | 1 +
>  meta/recipes-devtools/python/python3_3.12.4.bb             | 1 +
>  2 files changed, 2 insertions(+)

There is a bit of a debate around this change. 

We're at a point where we patch anything using "python" to use python3.
This puts us in a fairly good deterministic position where everything
is explicit.

If we add this patch, on the plus side, we can drop some of the patches
we're carrying but we also lose the markup of the python versions. If
there ever was a python4, we'd be back in a mess again.

Adding it, starts discussions about whether it should just be present
unconditionally or whether it should be a configurable link with
update-alternatives. Should it be a separate recipe? Should any recipe
using "python" without a version have a specific dependency on this new
recipe/package?

Having listened to the discussions and given this is pretty much a
solved problem for the majority of our current metadata, I'm in favour
of maintaining the status quo and not taking this change, even if we
have to carry a few patches. It does at least make the situation quite
clear and explicit.

Cheers,

Richard
-=-=-=-=-=-=-=-=-=-=-=-
Links: You receive all messages sent to this group.
View/Reply Online (#202458): 
https://lists.openembedded.org/g/openembedded-core/message/202458
Mute This Topic: https://lists.openembedded.org/mt/107264938/21656
Group Owner: [email protected]
Unsubscribe: https://lists.openembedded.org/g/openembedded-core/unsub 
[[email protected]]
-=-=-=-=-=-=-=-=-=-=-=-

Reply via email to