Re: [PATCH 2/2] python3: add bluez5-native to DEPENDS conditionally

["Guðni Már Gilbert" <gudni.m.g@gmail.com>]

[-- Attachment #1: Type: text/plain, Size: 1727 bytes --]

Hi all,

I've thought some more about this issue. I think it is best to remove these two patches I submitted from master-next if you haven't done so already. Using bluez5-native is wrong, I was misunderstanding the 'native' part a little bit.

Moving forward I think there are really only two possible ways to resolve this (both seem quite cumbersome):

1. Build BlueZ, and it dependencies by using a 'target' Python 3. Then after BlueZ is done building,  Python should somehow be re-configured and re-built in order to compile with bluetooth support.
2. Create a new recipe called libbluetooth which modifies BlueZ such that only the lib folder is built. This recipe only packages the library. In this case, we need to create a custom Makefile and a target which only builds the bluetooth library. It should have minimal dependencies. The goal with this approach is to avoid depending on Python at all such that we can avoid building it until we have libbluetooth available for the target build.

In my opinion I prefer option 2. I am thinking if it is worth it to push a patch to upstream BlueZ to better support only building the library. I'll work on this weekend to build the library locally and see what dependencies it needs. If it doesn't depend on glib and dbus, then that's perfect.

For option 1, I am not sure how this can be done. I'm open to any ideas.

This could take me a few weeks to implement as I don't have much free time except on weekends. But I feel this is an important issue to resolve and the pybluez library is no longer being maintained. Having native bluetooth support in Python really helps when developing Bluetooth scripts for development purposes or even for released products.

[-- Attachment #2: Type: text/html, Size: 1930 bytes --]