[oe] OEDEM: Summary of the 'Splitting the tree' session

[Michael 'Mickey' Lauer]

Hi guys,

here's the summary of the 'splitting the tree' session:

Nobody seems to disagree with splitting the metadata tree into more
managable chunks. The expected goals are:

1. Making it easier to find things,
2. Being able to mask whole groups out for improved parsing
3. Identifying core code which needs to always work, hence improving the
observed quality of OE.

Several criteria for splitting were discussed and there was a consensus
for splitting by functional complex. A tentative list of directories to
be used is:

base (10% underlying recipes)
toolchain
x11
framebuffer
gtk (includes gnome apps)
qte (includes opie and kde)
efl

The attendendees were well aware of all the inherent problems with
taxonomy and its different meaning to anybody in the first place as well
as that some recipes will always be difficult to categorize since they
are very flexible to configure. This did not prevent the consensus that
splitting up will be necessary to improve quality.

It was noted that oestats should be enhanced to be able to report a
subset of stats to start developing a test matrix of things which should
always work.

More discussion shall be carried on on this list and after the TSC has
been elected, it will make a final decision.

Cheers,

:M: