Yocto Technical Team Minutes, Engineering Sync, for August 3 2021
archive: https://docs.google.com/document/d/1ly8nyhO14kDNnFcW2QskANXW3ZT7QwKC5wWVDg9dDH4/edit

== disclaimer ==
Best efforts are made to ensure the below is accurate and valid. However,
errors sometimes happen. If any errors or omissions are found, please feel
free to reply to this email with any corrections.

== attendees ==
Trevor Woerner, Stephen Jolley, Richard Elberger, Saul Wold, Scott Murray,
Steve Sakoman, Tim Orling, Trevor Gamblin, Armin Kuster, Richard Purdie,
Joshua Watt, Michael Halstead, Jan-Simon Möller, Rob Woolley, Denys
Dmytriyenko, Tony Tascioglu, Randy MacLeod, Michael Opdenacker, Philip
Balister

== project status ==
- 3.4 m2 (honister) released
- 3.3.2 (hardknott) through QA, pending release approval
- 3.1.10 (dunfell) in QA
- new syntax change is in, all layers need to update to new syntax
- older releases of bitbake support the new syntax too (includes dunfell)
- layer compatibility updated to honister
- fix to bitbake’s datastore pending
- prserv rewrite still ongoing
- some AB-INT issues
- multiconfig still not 100%

== discussion ==
RP: big change to override syntax. fallout period ongoing

RP: subtle datastore bug in bitbake fixed (identified by Ross last year).
there might be some fallout. if you have a variable that isn’t assigned
a default value (or anything) but does have a conditional override (e.g.
_append) then it changes the value from “” to None. one such case was
found in all of oe-core (DRIDRIVERS in mesa), but others might exist in
other layers

JPEW: so now only the new override syntax is valid in master?
RP: yes. i added some checks in bitbake so it will now error if it finds
_append, _prepend, _remove. this doesn’t catch all instances of old
syntax, obviously, but it’s a good basic check
ScottM: will the syntax changes be backported in oe-core?
RP: there are no plans to do it in oe-core
ScottM: it would help with cherry-picking
RP: there are other incompatible changes that would make it hard/impossible to
backport the override syntax change to older branches
TrevorW: i’ve noticed a couple examples of overrides that don’t
seem to work when updated to the new syntax. in conf/layer.conf
BBFILE_PATTERN_<layer> aren’t overrides?
RP: no. the things in conf/layer.conf aren’t overrides. also IMAGE_TYPEDEP
isn’t (…yet). there’s a chance we might reconsider some
TrevorW: and SRCREV_<name>?
RP: yes. SRCREV_<name> is not, but SRCREV_pn-${PN} is an override
Elberger: is there a list?
RP: i have a list… in my head, there’s things in the manual, but we should
probably add more notes. we should add it to the migration guide
Elberger: it would be nice to make it consistent
RP: this is the first step, i think there will be follow-on steps. we changed
tune variables into overrides for example, but they’re not always
overrides
TrevorW: these inconsistencies are a great reason to do the syntax change
TimO: this change provided a nice to opportunity learn more deeply about
overrides and override syntax
MO: nice that the project isn’t mired in the past, it is able to make such
changes

RP: i’ve added a MAINTAINERS.md file in oe-core. it doesn’t give exclusive
ownership over an area of the project; i will, for example, apply patches
that i think are okay if maintainer feedback isn’t forthcoming. we’re
basically trying to document tribal knowledge; wanting to document the
list of people who are knowledgeable in various areas. we’re also
trying to encourage people to take ownership in various areas. this list
shouldn’t describe any changes to what people already know.

ScottM: (re: prserv) am now able to reproduce some of the bugs that have been
reported by others and that we’ve seen on the AB. i have ruled out a
bunch of stuff. the issues seem to start at python 3.8 (rather than 3.9 as
we were expecting)
JPEW: is it still related to exiting?
ScottM: that still happens, but this is something else. after the
prexport.bbclass cases run other things are stuck waiting. it looks like a
message might be getting dropped
JPEW: how large is the message?
ScottM: that’s what i thought too. but they’re not very large. maybe 100
bytes? hash + prnumber + packaging. there’s probably more than 1 bug at
play. looking through Paul’s stuff there’s also a case where something
gets shut down, but other things are left waiting for it to shut down

RP: (to the OE board) were there any discussions to promote OE membership (the
oe tsc election is coming up)
Denys: we discussed this at our last meeting
Philip: we need to figure out how to recruit members, we’re looking for
people active in the community. we would like to get feedback on how to
attract more members.
RP: it’s not obvious that OE has a membership, or what its benefits might
be. i don’t think people who are new to the project know about OE, what
it is, why to join, how to join
Denys: we used to have developer meetings (OEDEM/OEDAM) but we don’t have
that now, those would include general assemblies.
Philip: if you want to become a member send an email including a bio and
reason why you want to join to board@... and we can vote on
it
RP: maybe a reminder to various mailing lists would be good too

ScottM: do we have a booth at ELC?
TrevorW: check with David Reyna, i think he’s working on getting one

MO: is it okay to update the docs with the new override updates?
RP: yes. i had said “hold off” a week ago, but now that’s it’s in we
should go ahead with the changes to the main yocto docs