►
From YouTube: Foundational Infrastructure Working Group [Oct 28, 2021]
Description
Call recording from the open source Cloud Foundry Foundational Infrastructure Working Group held on 28th Oct 2021.
Minutes:
Autobump bot PR don’t require reviews (just green ci)?
Changes to CI owned by a specific vendor does not require reviews
[RK] produce guidelines around which repo’s can auto merge after review and for which ones we need manual controls to spread the workload load of the VMware CI Pair.
A
I
co-prepared
this
from
somewhere,
so
we
we
do
new
faces.
Maybe,
as
everybody
tells
maybe
a
bit
about
themselves,
I
mean
we're
not
going
to
do
that.
Every
call,
but
I
mean
for
now
it's
useful.
So
I
will
start.
I
am
reuben
koster
I've
been
working
with
bosch,
for
I
don't
know
a
long
time,
seven
years
or
something
at
least
for
a
long
time
at
starcom
wayne
and
recently
for
vmware,
where
I'm
not
part
of
the
party
ecosystem
team
and
on
the
foundation
and
infrastructure
working
group.
I'm
the
tech
leads
so.
C
Yeah
hi,
I'm
behind
I'm
from
sap,
I'm
working
with
porsche.
I
think
since
2015
I
did
torture
at
that
time
in
san
francisco
and
I've
been
part
of
the
bosch
team
europe
so-called
and
since
then
I
am
working
with
bosch
and
at
the
moment
I
am
working
on
some
different
things
in
cloud
foundry
from
customer
perspective
and
I'm
the
technicality
from
sap
site
for
push.
D
Should
I
go
next,
so
I'm
diego
I'm
based
out
of
london
and
I
work
for
vmware,
I
joined
vmware
through
the
pivotal
acquisition.
I
joined
pivotal
back
in
2016
and
I
think
since
then
I
work
with
bosch.
I
worked
at
various
teams
at
pivotal.
I
worked
with
the
bosch
team
at
some
point
and,
and
I
think
that's
it
really
fernando.
Would
you
like
to
go
next.
E
F
G
Hi
folks,
my
name
is
ram
ayangar,
I'm
based
out
of
chennai
in
india.
I
work
with
the
cloud
foundry
foundation
as
a
developer,
advocate
and
yeah
I'm
here
to
help
folks
with
any
support
that
they
might
need
in
terms
of
organizing
the
working
groups
or
anything
else,
and
also
here
to
learn
as
much
as
I
can
from
everybody.
H
Chris
well
yeah,
I'm
chris
clark
operations
manager
for
the
cloud
foundry
foundation,
so
yeah
recently
relocated
from
the
bay
area
to
atlanta
georgia,
which
makes
these
calls
a
lot
easier
and
yeah.
You
know
for
anyone
that
needs
any
any
technical
support
over.
You
know
operational
things
like
need,
new
repositories
or
stuff.
You
can
reach
out
to
me,
obviously
very
soon,
hopefully,
once
once
all
the
working
groups
are
set
up
and-
and
things
are
really
in
motion-
there'll
be
other
mechanisms
for
that.
H
But
you
know
if
anyone
needs
anything,
you
know
docker
hub
or
github
or
slack
or
anything
related,
please
reach
out,
but
yeah
excited
that
we're
kicking
off
these
calls
and
that
the
working
groups
are
really
starting
to
take
place.
A
Okay,
with
everybody
next
item
on
the
agenda,
I
thought
might
make
sense
to
like
review
a
bit.
What
we've
been
doing.
We
did
like
a
soft
start
of
like
trying
to
so
one
of
the
main
objectives
we
have,
as
a
group,
I
think,
is
like
making
sure
that
the
requests
get
merged
and
for
that
we
need.
We
have
approvers
that
need
to
approve
for
requests.
A
A
Let
me
show
what
we
are
currently
doing,
so
we
have
a
project
board.
Also,
I
think
in
the
meeting
invites
I
don't
know
if
it's
there
at
least
so
yeah
incoming
things,
and
so
what
I'm
currently
doing
is
assigning
the
incoming
items
just
a
bit
at
random,
like
just
do
like
a
quick
estimate
of
like
who
would
be
a
great
a
good
fit
for
this
based
on
context,
but
that's
like
I
can
be
off
there.
A
So
maybe
I
don't
know
if
people
think
that's
like
a
a
useful
way
to
do
it
or
that
we
should
maybe
have
more
of
an
a
different
approach
where
we,
maybe
people
pick
it
up
themselves,
but
yeah.
There's
that
part
and
then
there's
the
reviews,
the
smaller
things
I
think
are
being
merged
pretty
quickly.
A
F
A
F
I
think,
because
it's
still
in
the
beginning,
I
think
we
can.
We
can
do
a
bit
of
both.
So
if
you
see
that
it's
already
there
for
a
week,
you
can
just
assign
people,
I
don't
know
just
once
a
week
you
can
just
say:
okay
still
hasn't
been
assigned
to
someone.
Still
someone
hasn't
picked
it
up.
Now
I
basically
force
you
to
review
the
damn
thing.
A
Maybe
that's
I
mean
force
is
a
big
word.
I
guess
you
have
also
like.
A
Yeah
someone
needs
to
pick
it
up
and
you're
the
one.
So
the
thing
is
like
with
a
week.
I
think
that's
too
long
as
in
currently
not
all
things
go
to
through
pull
requests.
I
mean
we
still
have
the
the
vmware
folks
who
just
commit
to
master
sometimes,
but
if
so,
I
ideally
we
iterate
on
this
process
and
get
things
merged
more
quickly
right,
because,
ideally,
everything
should
go
through
pull
requests,
but
we
before
we
can
start
doing
that
we
need
to
have
like
a
proper
throughput.
A
F
A
E
I
don't
know
if
it's
the
right
time,
but
I
noticed
that
we
didn't
diego
normie
presented.
Why
or
what
kind
of
work
we
are
doing
in
those
repositories
and
why
we
are
here
in
this
working
group.
So
maybe
that's
something
to
share
context
with
you,
because
we
don't
know
what
will
be
the
best
way
to
collaborate
and
to
work
in
those
repositories.
So
maybe
it's
not
ideal
that
we
have
merged
permissions
or
it
is
ideal,
but
with
some
conditions.
So
I
don't
know.
D
Cool,
I
can
do
that,
so
the
cryogenics
team,
at
vmware,
tries
to
take
some
components,
can
be
like
vmware
components
or
components
that
components
from
cloud
foundry.
The
platform
that
the
vmware
ships
within
it's
like
proprietary,
not
proprietary,
but
enterprise,
offering
of
cloud
foundry,
and
we
try
to
automate
as
much
as
we
can
in
terms
of
bumping
packages
making
sure
the
project
is
healthy,
automate,
minor
releases,
for
example,
releasing
patches
and
things
like
that.
D
Just
so
that,
like
we
can
reduce
the
burden
of
like
maintaining
those
projects
and
waiting
until
the
point,
someone
wants
to
come
and
add
any
features
to
that
project.
So
the
cryogenics
team
won't
be
actually
adding
features.
We're
just
trying
to
work
on
this
automation
to
make
sure
that
the
components
are
healthy
and
being
looked
after
and
being
bumped.
And
things
like
that.
D
So
I'm
saying
that,
because
we
invest
heavily
in
automation
and,
for
example,
the
components
we
look
after
we
try
to
build
a
system
where,
for
example,
in
this
case
we
use
dependable
every
time
there
is
a
new
dependency
for
a
component.
D
Dependable
will
spot
that
try
to
create
a
pull
request
and
then
the
pull
request
will
run
all
the
tests
and
everything
and
try
to
validate
that
new
version
against
the
test
that
we
have
and
if
it
works,
and
if
our
pipelines
are
green,
then
our
pipelines
will
auto
merge
these
bumps,
and
that
was
what
we've
been
doing
so
far
with
some
of
these
components.
D
That's
okay.
I
think
what
you'd
like
what
we
are
interested
on
is
discussing
a
model
where
we
can
still
sort
of
abide
by
the
rules,
or
you
know
the
processes
that
we're
all
designing
right
now,
but
also
something
that
we
could
leverage
and
keep
this
sort
of
an
automation
aspect
of
our
work,
and
you
know
like
for
changes
that
are
not
like
adding
features
or
like
changing.
The
design
of
the
code.
D
Just
like
bumping
depends
and
the
you
know,
small
things
like
maybe
ci
configuration
we'd
like
to
keep
those
flowing
in
a
sort
of
a
more
automated
way
and
automate
those
as
much
as
possible.
In
order
to
do
that,
we
need,
like
credentials
of
you
know
like
some
machine
user.
That
would
be
able
to
like
auto
merge
if
the
pipeline
is
green
and
things
like
that.
G
A
Was
clear,
yeah
that
was
clear
and
I
think
the
so
the
goal
is
to
have
like
there's
we're,
not
gonna
step
in
the
way
of
bots.
I
think
it's
just
the
the
situation
that
has
happened
with
the
backup
and
restore
repo,
I
think,
has
something
to
do
with
the
move
from
the
cloud
foundry
incubator
to
the
cloud
foundry
org
and
then
the
permissions
were
basically
reset,
so
we
just
have
to
figure
out
the
right
permissions
for
that
bot
user
to
resolve
those
issues.
A
But
you
you
do
bring
up
an
interesting
topic
which
is
like
what
do
we
do
with
like
pipeline
changes
like
things
that
are
not
functionally
like
we're,
not
functionally
changing
anything
we're?
Just
I
don't
know
fixing
pipelines
basically.
A
Are
those
things
we
want
to
review
right
or
how?
How
do
we
want
to
deal
with
those?
Because,
in
my
experience,
like
fixing
pipelines
and
having
to
go
to
vrs
or
having
to
go
to,
can
be
a
bit
painful
and
like
slow
you
down
a
bit
and
usually
when
you're
fixing
ci
you're
a
bit
in
a
hurry,
or
you
want
to
to
have
it
green?
But
that
would
be
maybe
something
to
bring
up
with
the
toc
in
general
right.
A
It's
not
something
I
think
we
can
decide,
but
maybe
having
an
exception
for
like
ci
directories,
right,
something
where
the
tasks
and
the
pipelines
live
could
be
something
potentially
interesting
to
like
for
for
speed
purposes.
I
mean
we
can
try
it
first
without
and
just
do
a
pr
based
model,
but
if,
if
that
starts
to
become
to
slow
us
down,
I
think
ci
would
be
a
potential
interesting
area
to
have
different
rules
for.
E
I
think
it
is
in
more
interest
to
cryogenics
and
vmware
in
general
to
for
those
pr
that
we
merge
to
provide
some
visibility
on
on
the
the
validation
we
have
performed.
I
I
think
it's
something
we
are
working
on,
so
that
may
helps
to
to
have
better
visibility
on
why
things
get
merged
automatically
and
so
on,
not
completely
opaque
about
the
processes
and
so
on.
C
E
D
Go
on
diego,
I
I
was
just
about
to
say
that
I
think,
like
they're
in
my
mind,
like
two
sort
of
overlapping
priorities,
so
like
I'm
coming
here
from
the
cryogenics
point
of
view
and
I'm
talking
about
the
repos
and
the
components
that
cryogenics
oversees
within
vmware,
which
is
like
not
clear
at
all
for
folks
in
the
cff
and
the
outside.
D
But
to
like
to
your
point
like
we
do
not
work
with
bosch
and
bosch
is
like
a
more
like
sort
of
a
much
bigger
component
and
project,
and
I
think,
like
there
are
much
more
people
and
companies
interested
on
it.
So
I
think
that
would
be
like
a
different
story
and
I
think
I
think
ruben
started
to
like.
D
Maybe
take
this
idea
that
cryogenics
wanted
to
like
fast-track
pr's
that
are
like
our
brs
bumping
dependencies
or
like
prs
just
you
know
changing
like
bits
of
ci
and
things
like
that,
and
maybe
differentiating
that
from
like
prs.
That
would,
like
you
know,
change
other
things
like.
I
think
this
conversation,
this
differentiation
hasn't
happened
for
bosch
yet,
and
I
think,
like
the
components
we've
been
talking
about
and
at
the
top
of
my
head,
that
we
look
after
and
which
are
in
the
this
working
group,
are
the
bbr
components.
I
think
only.
E
A
Yeah
I
mean
like
why
I'm
especially
talking
about
this
is
because,
for
example,
we
have
a
lot
of
repos
that
just
like
have
nci
dependency,
bumper
jobs
right,
that
would
bomb
dependencies,
run
some
tests
and
then
commit
and
push
so
those
don't
even
go
through
a
br
right
and
in
essence,
dependable,
is
the
same.
It's
just
creating
a
pr.
A
A
Those
types
of
pr's,
also
like
in
the
automation
for
for
generating
a
board.
I
have
explicitly
excluded
dependables
and
there's
a
few
other
bots
that
do
similar
things.
I've
excluded
those
because
they
just
generate
noise,
and
I
don't
think
it
makes
sense
for
this
whole
review
process
to
go
over
the
exchange
so
yeah.
Those
should
just
be
like
free
to
merge,
basically
for
the
team.
Maintaining
that
repo.
A
But
I
I
would
even
go
further,
maybe
right
with
like
maybe
saying
we
don't
need
reviews
on
like
the
ci
directory
as
in
like
pipelines
and
tasks
and
those
types
of
things,
but
I'm
not
sure
about
that.
Yeah
right,
that's
more
something
to
bring
up,
maybe
with
the
dlc.
If
we
have
like
a
bit
more
mileage
with
this
process,
and
we
noticed
that
it's
becoming
cumbersome
that
I
think
would
be
a
good
candidate
to
also
have
differentiation.
D
I
think
because
companies
have
sort
of
taken
ownership
of
specific
components
so
like,
for
example,
vmware
is
sort
of
more
involved
with
maybe
bbr
than
other
companies
and
then
like
vmware
has
set
up
ci
for
that,
and
it's
been
okay
so
far,
but
then
because
of
that,
I
think
vmware
teams
allow
themselves
to
sort
of
just
like
change,
ci
and
push
because,
like
there's
already
a
lot
of
vmware
stuff
in
there
anyways,
if
in
the
future,
we
feel
the
need
that
ci
should
be
more
like
vendor
agnostic
and
more
like
open
source
friendly,
so
that
maybe
anyone
who
would
clone
a
project
would
take
that
ci
folder
in
the
the
repo
and
be
able
to
set
up
their
own
ci
on
their
company
or
you
know,
within
their
development
environment.
D
D
D
I
think,
as
as
long
as
it's
10,
something
like
bender,
provided,
I
feel
like
teams,
would
just
keep
like
changing
ci
and
like
just
try
to
roll
them
like
to
push
them
through,
I'm
just
bringing
that
up,
for
example,
because
like
in
some
of
the
pipelines
we
have,
we
have
like
specific
vmware
bits
and,
for
example,
if
I
ask
someone
who
is
not
vmware
to
review
that
with
me,
it
would
be
like
I.
I
don't
see
a
meaning
in
doing
that.
A
F
F
F
A
A
So
I
think
that
would
be
a
good
candidate
to
to
have
that
like
where
we
would
review
ci
changes
right,
because
there
you
are
collaborating
on
the
whole
thing
for
the
rest
for
most
of
the
other
repos.
For
now,
I
think
it's
better
to
focus
on
like
functional
changes
or
like
things
that
are
actually
changing
the
behavior
of
the
software
and
not
so
much
the
ci
but
yeah.
F
Yeah,
so
we're
all
in
agreement
that
outdoor
bots
and
they
they
don't,
have
need
to
be
prs
and
we
just
ignore
them
and
automatic.
Because.
F
Right
reuben
is
frozen
again.
D
That
was
my
understanding
as
well
back
to
your
point,
ramon
and
also,
I
think,
for
components
that
follow
in
this
case,
where
ci
is
provided
by
a
company.
I
think
that
companies
should
have
more
flexibility
in
terms
of
you
know:
yeah,
maybe
merging
their
own
cia,
their
own
pr's,
or
you
know
like
just
pushing
to
master
because
like
they
would
have
all
the
context
on
their
ci
right,
and
it
would
be
weird
for
other
companies
to
like
review
pull
requests
with
them.
D
I
think
these
are
the
two
cases
that
come
to
mind.
F
Yeah,
I
agree
as
well,
but
yeah
the
the
problem.
That
is
all
this
that
yeah
there
are
hundreds
of
repositories,
no
one
is
an
owner
and
I
think
we
try
to
solve
it
all.
Just
we
don't
get
pr's
that
are
there
for
two
years.
A
A
Yeah
I
wanted
to
capture
what
we
discussed.
I
think
vermont
was
doing
that,
but
let's
put
that
in
the
notes,
so
auto
bumping
but
autobot
vrs
don't
require
reviews.
A
B
A
Yeah
and
and
bots
in
general,
we
still
want
bots,
I
mean
they
are,
but
I
don't
think
we
need
to
have
a
rule
or
something
about
that.
It's
just.
We
need
to
fix
the
cryogenics
thing
for
that
one
specific
repo,
but
I
don't
think
that
it
was
just
caused
by
something
else
right:
the
move
of
the
incubator
stuff.
A
We
would,
I
think,
it's
best
to
set
something
up
fernando
you
want
me,
maybe
or
maybe
diego
to
just
bear
on
getting
that
resolved.
A
Okay,
let's
do
that
later.
Okay,
shall
we
review
some
blocked
items
right
because
that's
the
whole
point
we
want
velocity.
So
if
things
are
blocked,
that's
not
lost
that's
country
too
this
one's
helping.
So,
let's
so
beyond,
I
think
you
put
this
one
here.
She
wanted
to
discuss
more
broadly.
C
F
Does
that
mean
that
for
it,
because
I
think
there
are
multiple
code
styles
now
between
every
repository,
everybody
did
his
own
thing.
A
Yeah,
so
I
think
the
idea
with
wash
is
that
you
have
robocop
set
up
and
there's
like
a
pre-commit
hook
so
before
you,
so
that
the
lines
that
you
touch,
you
have
to
fix
the
code
style
so
that
I
think
that's
the
rule.
There's
now
something
to
enforce
the
gold
style
and
then
yeah.
C
C
Integration
to
execute
the
integration
test
is
most
probably
not
possible
for
most
of
the
contribute
for
the
contributors
and
so
a
little
bit
more
clarity
on
for
people
contributing.
D
Can
I
add
something
yeah?
This
is
maybe
something
that
cryogenics
can
contribute
with,
because
in
many
of
the
components
we
own
right
now
we
have
configured
our
pipelines
to
run
tests
all
the
tasks
we
have
against
every
pull
request
and
we
update
the
pull
request
commit
status
on
github
following
the
results
of
our
ci
so
like.
If
you
submit
a
pull
request,
the
pipeline.
G
D
Kick
in
and
they
will
like
share
feedback
with
you,
and
this
is
something
that
we've
done
for
our
components
and
the
like
pbi
for
this
working
group.
But
maybe
this
is
something
we
can.
You
know
share
some
context.
A
And
that's
just
working
with
you're
doing
the
upload
concourse
or
get
your
actions.
D
F
D
I
think
in
a
world
where
there
is
a
ci
like
a
shared
ci
being
run
by
the
cloud
foundry
foundation,
I
think
like
this
would
not
be
a
problem
anymore.
No,
I
was
like
people
would
have
access
to
the
logs,
but
right
now,
as
the
ci
is
on
our
end,
and
it's
not
publicly
available,
the
only
feedback
people
would
have
is
the
pull
request
saying.
Oh,
these
tests
have
failed
and
like
without
details,
actually.
E
Yeah
currently,
we
have
been
eluding
or
working
around
that
problem
because
we
don't
test
or
don't
go
through
ci
for
fork,
prs
ps,
coming
from
a
fork
that
this
is
exactly
what
we
are
speaking
now.
I
guess
so.
A
Yes,
yeah.
There
needs
to
be
like
a
level
of
trust
right,
so
if
we
can
limit
it
to
people
that
have
can
create
branches,
so
we
limit
it
to
br
from
foundation
members.
I
think
that
would
be
already
a
different
thing
and
I
mean
it's
not
so
much
a
problem
if
you're
running
unit
tests
right
because
that's
cheap,
but
in
lots
of
cases
we're
running
against
like
shared
infrastructure
that
keeps
like
it's
more
long
running
and
it's
creating
actual,
is
resources
and
costing
real
money.
A
So
that's
yeah,
I
think
you
know.
Quite
traditionally,
it
has
been
a
bit
of
an
difficult
topic
for
at
least
things
like
porsche
core
to
do.
Vr
run
ci
on
every
pr.
A
But
I
mean
to
get
back
to
this
this
topic-
I
I'm
sorry
I
mean
so
we
could
say
something
like
running
tests
just
run
the
unit
test
right
and
then
we
at
least
we
do.
The
review.
You've
done
your
best
with
unit
tests
and
we
can
always
revert
the
pr.
Maybe
if,
like
something
big
breaks
down
the
road,
but
then
at
least
people
have
already
looked
at
it.
I'm
verified
that
it's
okay,
ish.
A
C
A
Unit
tests
for
unit
as
we
we
could
have
pr
or
unit
tests
on
hbr
right
at
some
point-
we're
not
there
yet,
but
that
would
be
something
I
I
would
think
would
be
useful
to
have
at
least
on
the
the
repos
that
have
the
most
traffic
or
see
the
most
prs.
A
But
that's
I
don't
think
we
don't
have
like
really
an
authority,
I
think
to
like
schedule,
work
right.
It's
it's
up
to
the
vendor's
own
backlog,
on
what
work
to
prioritize
we're
more
here
for
like
making
sure
that
the
work
that
has
been
done
gets
approved
and
gets
merged.
A
It's
not
so
much,
there's
not
so
much
road
map
work.
I
mean
we
can
collaborate
with
the
toc
on
things
like
the
the
jj
stem
cell
and
stuff,
but
for
it's
not
like,
we
can
decide
in
this
group
to
say
yeah,
we're
gonna
do
this
yeah,
we
can
say
we're
gonna
do
this,
but
then
still
someone
needs
to
do
the
work
so
and
that's
up
to
like
a
volunteer.
A
So
first,
my
first
priority
there
is
making
sure
that
we
have
like
a
good
visibility
into
what's
open
and
how
track
progress,
but
the
next
step
would
be
like
make
sure
that
we
have
better
feedback
cycles,
make
make
reviewing
easier
right
and
one
of
those
things
would
be
br
ci
status,
things
for
for
the
place
where
it's
doable,
but
in
the
interest
of
time,
do
we
is
there
an
action
item
we
want
to
take
here.
C
C
It's
also
for
me,
so
at
the
moment
we
are
just
refactoring
agents,
the
properties
for
the
pop
store,
and
we
have
many
ps
which
are
somehow.
C
Yeah-
and
I
don't
know
which
cpi
still
needs
to
remove
the
dependence
to
the
property,
I
think
on
openstack
cpi
at
least.
A
Yeah,
I
think
we've
merged
the
pr
for
the.
I
think
cp.
C
If
we
say,
let's
do
our
certain
stack
and
then
just
say
you
can
you're
safe
to
remove
the
there's.
A
Control
like
the
cloud.
A
A
F
We
can
also
make
like,
in
those
cpi
reap
repositories.
We
can
also
make
the
suggestion
at
least
an
issue
from
a
this,
is
what
we're
going
to
change
and
then
maybe
down
the
road.
We
can
remove
it
at
the
end,
but.
C
Yeah,
okay,
then,
I
think
yeah,
let's
go
with
dummy
values
in
the
ops
file
which
we
have
so
for
the
properties
I
mean
it.
A
It
ever
it
gives
you
the
same
end
result
right
or
as
in
like
you,
don't
have
a
valid
credit
or
a
credential
that
can
still
be
captured
and
used
right.
It's
a
credential
that
doesn't
work.
So
I
think
that.
A
C
C
A
Of
view,
it
should
satisfy
the
requirements,
yeah
yeah,
and
it's
also-
and
it's
not
a
breaking
change,
which
I
like.
So
because
the
alternative
is
like
really
hard
to
coordinate
right.
Then
we
need
to
make
sure
that
all
the
vr's
get
merged
and
released
and
that
all
those
things
get
bumped
into
bosch
deployment
before
we
can
merge
that
vr
and
that
that
would
potentially
mean
that
we
have
something
locked
on
the
board
for
a
month
or
more.
A
C
A
No,
not
us,
no,
no
also
about
merging
prs.
I
think
I've
been
doing
that
mostly
myself
as
in
app,
but
maybe
so
once
the
changes
are
approved,
then
I
just
go
over
that
changes
or
and
merge
them.
But
that
means
that
I'm,
like
a
bottleneck
in
the
system,
I
don't
think
that's
ideal.
F
A
Yeah,
so
I'm
still
on
this
just
discussing
that
with
the
other
working
groups
like
to
to
have
a
bit
of
alignment
around
the
different
repos
or
different
yeah
projects
within
cloud
foundry.
So
I
think
we're
now
aiming
at
two
reviewers,
but
maybe
that's
a
bit
too
much.
A
F
So
yeah
those
things
are
still
being
formalized
so,
basically
for
now,
until
that's
formalized,
we
say
two
reviewers.
So
if
two,
if
there
are
two
approved
previews
any
of
any
of
any
of
the
reviewer
who
done
the
last
review,
can
just
basically
do
a
merge.
A
Yeah
in
theory,
but
in
practice
the
the
thing
is,
we
also
have
like
ci
that
can
be
read
and
stuff.
That's
not
visible,
so
I
think
for
most
of
the
repos.
It
makes
sense
that,
like
one
of
the
vmware
folks,
who
also
deals
with
ci,
would
do
the
final
merge
right,
because
we
we
want
to
like
have
some
control
of
that
like
not
all
ci
is
red
all
the
time
right.
We
want
to
have
a
bit
of
coordination.
Like
is
ci
green
enough.
Can
we
merge
this?
A
Can
we
take
this
thing
in
but
like?
No?
That's
not
true
for
all
repos
right,
some
repos
are,
but
I
yeah
we
probably
should
set
up
something
there
to
make
that
more
clear,
which
things
can
be
like,
basically
auto,
merge
or
like
just
if
there's
approvers,
we
can
merge
and
some
things
where
we
want
more
manual
control.