[Openembedded-architecture] LSB is dead; when shall we drop it?

[Richard Purdie]

On Fri, 2017-04-07 at 13:58 +0300, Alexander Kanavin wrote:
> So I want to ask: how much effort should we spend on keeping it on
> life 
> support going forward? I'm all for dropping it entirely, but I 
> understand that it still may be an important checkbox to cross for 
> commercial vendors. Still, the components it requires (qt4 for
> instance) 
> are increasingly bitrotting, and at some point the question of
> dropping 
> them will become acute. What you think?

My proposal (for 2.4) is we put a file into OE-Core which states
exactly which pieces of lsb we continue to support. We'd drop qt4 as
its obsolete and mention openssl as one of the reasons. We could also
drop some things like mailx as it a dead upstream and openssl.

The key thing would be to document which pieces were no longer
supported and why. We'd then modify the code so for example the qt4
layer was no longer required.

I don't think the automated testing actually runs the lsb tests today,
that is handled by QA. QA would have a list of tests we know will never
pass and we'd exclude that test list from our results.

I think that should given everyone what they need, we'd just need to
discuss and agree the list of things we accept from LSB what we won't
do/meet.

Cheers,

Richard