►
From YouTube: 20190722 sig testing commons office hours
Description
No description was provided for this meeting.
If this is YOUR meeting, an easy way to fix this is to add a description to your video, wherever mtngs.io found it (probably YouTube).
A
A
Alright,
so
just
a
reminder:
here's
kind
of
the
current
set
of
priorities
we
have
for
the
1:16
cycle.
There's
not
much
new
to
this
p0
is
basically
refactoring
the
framework,
so
it's
more
easily
importable
to
the
outside
world.
This
is
primarily
useful
for
everyone,
who's,
making,
C
or
D
based
anything
as
well
as
the
cloud
providers
and
all
the
other
stuff
returns
move
on
a
tree.
A
Second,
one
was
the
RP
one
was
the
user
experience
stuff
that
is
been
kind
of
a
far
and
distant?
Second
I
have
not
created
a
an
issue
to
federate
the
work
for
what
I
would
consider
to
be
a
cleaner
user
experience,
especially
from
the
component
config
perspective,
although
this
is
important,
it's
just
not
been
a
high
priority
for
me
at
the
time.
A
B
I
put
it:
okay,
yes,
okay,
so
that
one
is
just
a
small,
a
small
thing
that
we
just
don't
and
I
came
across
while
working
on
one
of
the
yards,
a
there's
been
some
work
on
trying
to
clean
up
called
in
terrors
and
also
another
type,
a
PR.
The
merge
done
this
create
the
script
essentially
runs
against
AC.
I
blocking
job
a
walking
job,
some
kubernetes
any
prevents
anyone
from
using
Omega,
explain,
go
mega,
expect
or
any
of
the
any
of
experiences.
B
As
you
can
see
in
the
screen
just
going
for
a
message
of
use
framework,
they
expect
no
error
expect
nor
been
wrapper
for
Omega
with
top
set
equal
to
one.
So
that
means
at
the
whatever
ever
it
happens,
is
gonna
be
sent
to
the
color
of
the
function
that
fail,
not
another
function
itself.
So
it's
just
wanted
to
throw
it
out
there.
B
If
I
never
saw
this
go
through
any
drone
in
venture
in
any
in
any
of
the
channels
and
sig
testing
code
organization-
and
it
seems
a
little
bit
arbitrary
to
just
bloke
and
take
omega
expected,
given
that
people
actually
use
a
lot
more
Omega
throw
the
test.
Also
a
dimension.
A
the
motivation
for
pushing
down
to
see
is
there's
another
issue
that
I
linked
in
the
notes.
Were
he
says
we?
The
battle
is
destroyed.
B
The
co
mega
expect
batter,
but
the
actual
the
actual
item
that
people
were
looking
to
get
out
of
that
one
is
to
just
an
update,
all
the
all
the
test.
So
it
is
so
whenever
something
fell,
they
didn't
look
like
I,
think
where
he
just
says
expect
no,
ever
and
and
nothing
whatsoever,
so
yeah,
maybe
thoughts/opinions.
B
Yeah
also,
there
were
also
the
other
thing
and
I
guess.
One
of
the
reasons
why
it
matters
a
little
bit
more
to
us
is,
for
example,
within
the
framework.
The
function
that
that
it
mentions
is
that
we
have
to
use
this
framework
expect,
nor
which
is
a
wrapper
for
a
comic
I,
expect,
with
up
set
equal
to
one.
There
is
no
wrapper
for
just
a
plain
Omega
expect,
so,
if
we
actually
want
to
so,
let's
say
that
we
have
a
let's
say
that
we
do
some
refactoring
were
a
function.
B
It
returns
an
error
instead
of
instead
of
actually
filling
a
filling
a
within
that
function.
So
in
order
to
keep
the
behavior
as
expected,
we
need
to
keep
they
upset
people
they
equal
to
zero,
a
from
a
from
the
color,
and
the
only
way
that
we
can
keep
the
behaviour
is
so
it
would
be
to
actually
call
behavior
a
from
the
framework.
A
called
the
I
forgot,
what
it's
called
with
the
function
that
expects
big,
nor
actually
calls
which
is
not
a
big
deal,
but
it
looks
a
little
bit
again.
A
I
would
expect
that
we
have
the
ability
to
call
a
line.
Number
emitting
error
function
from
the
top
level,
but
one
of
the
things
that
I
don't
like
is
that
we
call
the
escape
hatch
for
air
conditions
within
certain
areas.
That
are
weird
right
like
there
are
doing
it
within
the
test
seems
fine
because
it
ejects
from
the
test.
But
there
are
preconditions
where
we
inject
earlier,
like
that
one
rapper
that
Titan
tried
to
make
changes
to
and
that
one
is
really
weird
because
you're
not
it's
in
the
precondition
before
everything.
A
I,
don't
have
strong
feelings
here
other
than
like
what
I
think
it
might
make
sense
for
us
to
start
to
document
what
we
consider
to
be
best
practices
for
how
to
omit
errors
and
your
tests,
and
maybe
that's
a
more
general
thing
that
we
should
be
doing
like
when
is
the
best
place
and
when
she
use
certain
types
of
tools
within
the
framework,
because
right
now
idiot
like
I
mentioned
earlier.
It
is
not
consistent
in
that.
B
A
B
A
B
A
I
personally
think
that
this
is
totally
an
unneeded
unwarranted.
You
don't
need
to
do
this.
I
think
this
is
us
right
like
we
have
so
many
verify
in
checks
that,
like
I,
don't
there's
no
benefit
to
this
right
now
it
is
just
like
you
should
be
using
a
best
practice,
but
if
anything,
then
we
should
add
a
full
lint
that
basically
says
here
all
the
practices,
and
here
the
things
we're
not
doing
right.
This
seems
pretty
arbitrary
and
it
doesn't
seem
like
it
was
prioritized
by
this
group.
B
A
Like
he's
not,
there
are
so
many
more
high-value
targets
that
amici
and
company
could
be
focusing
on
that.
That
could
be
aligned
with
the
priorities
that
we
have
set
here
and
I.
Think
if
somebody
sees
something
like
this
to
put
a
hold
on
the
issue
and
decide
it
to
me
because
I
know,
I
didn't
green,
like
this
thing.
B
Yeah
yeah
actually,
when
today
went
to
the
history
and
that's
another
reason
why
I
wanted
to
bring
her
up
for
discussion.
It
really
didn't
went
through
much
of
a
review.
It
was
just
a.
It
was
actually
a
a
part.
It
was
a
derivation
of
a
clean
up
that
people
were
doing
in
some
other
in
some
other
e2e.
Just
me,
you
know
just
one
day
just
wanted
to
keep
in
there.
So
it's
a
little
bit
arbitrary,
but
it's
a
worth
it
removing
at
door.
A
A
C
A
E
Think
we're
just
kind
of
going
through
them.
Slowly
like
not
much
to
update
mastico,
then
I
saw
gee,
Tom's
PR
go
through
and
there's
a
few
other
in
progress
that
are
just
doing
yeah,
like
the
expect.
No
error
that,
but
we're
moving
along
I
think
one
thing
that
came
that
it's
coming
up
often
that
seems
to
be
blocking
a
lot
of
people
is
a
lot
of
like
is
master
checks
or
is
control
playing
checks.
Things
like
that,
there's
still
a
lot
of
contention
around
like.
When
do
we
at
what
point?
A
I'm
gonna
kind
of
rely
on
there
is
a
separate
feature
request
from
Clayton
in
the
enhancements,
repository
and
I'm
gonna
kind
of
rely
on
them
to
actually
fix
both
Clayton
and
Jordan
to
actually
fully
flush
out
the
details.
I
do
know
that
there's
a
lot
of
feedback
from
Justin
Santa
Barbara
on
this.
With
that
says,
like
you're
you're
saying
this
is
not
good
and
we
shouldn't
be
doing
it,
but
you're
not
giving
us
what
we
should
be
doing
so
as
a
result,
we
don't
actually
have
the
proper.
A
F
G
G
Right
now
we
have
a
lot
of
the
coils
in
the
echo
system,
depending
on
the
master
label,
including
cop
scream,
ADM
and
so
on.
Historically,
justin
explained
that
the
deployers
agreed
to
this
label
and
but
again
clayton
saying
that
it's
not
official,
it's
not
part
of
the
core,
so
maybe
our
sing
commence.
The
requires
can
employ
a
consensus
on
this
label
outside
of
core.
Like
you
know,
the
committee
of
the
deployers
agree
mama
label
that
makes.
A
A
G
But
the
question
remains,
however:
how
are
we
going
to
handle
any
arbitrary
question
with
the
framework
and
basically
the
suggestion
there
was.
I
already
walked
a
ticket
for
this.
The
suggestion
here
was
to
employ
a
granularity
on
the
demands
that
the
framework
has
from
a
particular
node.
For
instance,
can
we
schedule
workloads
on
this
node
instead
of
asking
for
a
particular
label
and
somebody
sides
to
work
on
this?
G
However,
I
think
he
was
scared
after
he
realized
that
it's
not
so
simple,
because
the
is
master
call
is
present
in
the
number
of
areas,
including
tests,
so
I'm
going
to
get
back
to
this
ticket
after
some
time-
and
maybe
you
know
it's
already-
has
the
camp
wanted,
but
I
didn't
really
want
to
jump
on
it
and
certainly
leave
someone
else.
They
want
to
work.
This.
A
G
Yes,
I
I
do
agree
with
them.
That's
it's
simply
the
addiction
there
is
sorta
law
in
us
to
do
this,
but
I
wonder
if
if
the
label
should
be
something
that
we
are
checking
off
regardless,
given
the
label
is
something
that
is
supported
in
the
ecosystem
by
so
many
deployers.
So
you
know
if
the
label
does
not
exist,
maybe
we
can
have
a
fallback
to
check
for
that
actual
granularity
that
you
want,
or
maybe
it
should
be
the
other
way
or
out.
G
G
A
G
Just
seconds
I
can
share
the
ticket
in
chat
that
ticket
originally
oh
yeah.
Somebody
hundred
said
it
basically
in
the
there
is
a
special
linked
interest
inside
this
ticket.
It's
all
original
PR
that
attempted
to
go
change
on
his
function.
Basically,
the
Disco's
all
the
tests
in
the
framework
and
in
the
suite
that
use
this
function,
basically
I'm
seeing
stuff
like
checking
whether
there
is
a
controller
manager
on
these
notes,
so
it
might
be
tricky.
It's
a
very
generic
assumption
that
should
be
made
modified.
A
G
B
G
Think
the
trickiest
part
of
the
gifs
that
basically
shows
the
usage
of
is
masternode
is
inside
the
the
list
of
get
muscle
and
worker
nodes
or
die.
That's
basically
the
generic
assumption
where
he
enumerates
what
nodes
we
have
to.
So
it's
quite
tricky.
This
means
that
we
can
no
longer
have
this
function.
We
can
no
longer
say
that
we
have
a
list
of
master
nodes.
E
G
A
D
A
That's
the
reason
why
the
granularity
was
set
there
in
the
beginning,
so
there's
only
a
finite
set
of
tests
that
use
a
certain
set
of
utilities,
so
only
import
that
utility
library
for
the
test
then
is
needed.
Otherwise,
the
framework
pretty
much
depends
upon
the
ability
to
call
expect
error
or
expect
no
error.
G
Quickly
to
the
previous
point,
I
have
to
say
that
I
really
like
that
we
are
wrapping,
go
make
up
with
the
you
know,
function
we
have
for
expect
or
error
I
really
like
this,
because
it's
allows
us
to
swap
or
maybe
add,
more
logic
inside
the
function
like
we
can
have
a
stack
trace
and
things
like
that
correctly
correctly.
If
you
look
at
the
failing
tests,
you
just
say
it
just
says
something
like
around
waiting
for
the
condition
and
you
don't
have
any
any
way
to
navigate.
A
G
Added
this
next
item,
because
the
ipv6
slash
dual
stack:
people
are
correctly
trying
to
implement
some
new
flags
in
the
framework
and
I.
My
immediate
response
was
that
we
should
not
do
this
anymore
and
instead,
the
proposal
originally
was
to
implement
some
sort
of
auto
detection.
Whether
a
particular
test
depends
on
to
all-stock,
ipv6
or
IP
before
and
but
the
comments
were
in
the
lines
of
no,
it's
not
easy
for
us
to
do
this.
Let's
add
the
flag.
A
Here's
my
problem
with
this
and
I
said:
I
allowed
Patrick
to
get
away
with
a
bunch
of
stuff
like
that
to
is
if
we,
this
was
the
same
problem
that
comedian
was
suffering
from
and
then
we
just
basically
said
no
and
we
stopped
it
because
we
understand
the
bleeding
the
answer.
I
think
here
is
the
same
answer
that
we
did
with
comedian
like
we
cannot
plumb
through
all
the
different
options
that
exists
and
keep
on
doing
this
over
time,
because
it's
not
sustainable.
We
are
we're
already
in
this.
A
A
Debt
I
think
we're
just
gonna,
be
in
service
of
debt
for
the
next,
like
six
months
to
a
year
with
everyone
who's
working
on
this
stuff,
cuz,
it's
part-time
effort
at
best,
so
I
would
say
no
I,
think
I
think
we
can
do
better
and
I
think
we
should
do
better
if
there
are
feature
gate
flags.
You
know
like
if
there
are
feature
gate
flags
that
are
enabled
for
a
component.
Those
things
should
be
part
of
the
component
and
easily
detectable,
as
part
of
their
component
configuration
right.
A
So
if
you
have
a
feature
gate
flag,
you
should
be
able
to
detect
that
as
part
of
the
component
configuration
for
a
feature
or
for
for
a
component
or
a
feature
right
and
that
should
be
stored
on
cluster.
It's
in
the
config
map.
It
shouldn't
be
specify
another
parameter.
You
should
be
able
to
dynamically
Auto,
detect.
G
Yeah
the
thing
that
I'm
most
scared
about
is
that
somebody's
going
to
implement
these
flags
for
the
framework,
and
then
we
are
going
to
pull
a
lot
of
end-to-end
testing
jobs
that
use
these
flags
and
we're
going
to
couple
this
with
testing
front.
Then
these
people
were
to
disappear.
We
have
this
legacy
flag
that
nobody
noticed
and
the
slow
longer
temporary
flag
is
actually
quite
coupled
between
different
requesters
and
I
am
going
to
continue
the
Rasputia
description
here.
G
We
need,
in
my
opinion,
meant
to
have
auto
detection
for
this
stuff,
because
the
component
coffee
is
not
there
yet.
Maybe
some
cases
are
going
to
not
defend
from
the
edge.
You
know
we
have
to
have
a
way
to
be
capable
to
try
to
use
ipv6
or
if
all
my
pieces
are
provided.
Maybe
one
of
the
IP
so
in
the
guess
is
that
before
the
other
one
I
said
six,
then
he
goes
back
waiting.
Is
the
test,
so
I'm
really
I'm
trying
to
ask
this
question,
but
this
question
but
I
also
Maxim
understanding.
Why,
like?
G
A
G
If
we
enable
component
config
for
the
framework,
we
have
to
get
rid
of
all
the
flags
that
already
exists,
and
it's
so
there's
so
many
flags
and
like
so
many
people
probably
already
depend
on
this.
That
I
was
at
some
point
proposing
that
maybe
we
should
do
you
know
framework
thoughts
up
to
pretty
much
create
a
new
version
already
well.
A
We
talked
we
talked
about
this
in
the
conformance
working
group
and
I,
don't
mind
soliciting
the
idea
of
of
having
it
listed
on
the
requirements
for
a
framework
me
too.
What
I
don't
want
to
see
happen
is
us
to
try
and
do
this
weird
human-animal
hybrid,
where
we
start
implementing
the
ideas
for
a
new
implementation.
A
G
We
should
definitely
make
this
like
a
long-term
plan
for
the
framework,
and
we
have
to
deal
with
what
we
have
right
now
as
a
comparison
chip
test.
We
do
started
but
I'm
not
convinced
that
the
architecture
of
campus
as
a
whole
is
a
very
clear
and
right
now
implementing
something
that
is
cubed
is
v2,
but
maybe
there
are
underlying
problems
that
have
to
be
resolved
first
and
it's
just
an
m4
assigned
to
somebody
to
work
on,
but
it's
not
very
clear
so.
A
A
I
do
think
that
it's
worthwhile
to
entertain
the
conversation
of
starting
a
specification
of
identifying
what
are
all
the
key
features
and
issues
that
we
see
and
I
don't
think
it's
a
bad
idea
for
someone
to
start
curating
this
that
we
talk
about
it
every
week
at
least
a
little
bit
right
and
maybe
just
make
incremental
still
progress.
At
least.
If
we
were
to
do
this,
we
would
have
at
the
end
of
the
day
a
document.
G
A
Yeah
I,
don't
care
about
framework,
so
I
want
to
make
certain
that
we
do.
We
talk
about
that
in
detail
like
I
care
about
the
requirements
before
frameworks,
so
I
looked
at
the
Gherkin
idea
and
proposal
and
document
and
I
don't
want
to
create
another
hybrid
I
want
to
make
sure
that
we
outlined
requirements
in
current
problem
space
that
we
see
right
like,
for
example,
like
our
parallelization
of
the
current
framework
and
their
ability
to
execute
on
some
sweets
doesn't
exist.
A
G
It's
just
to
be
like
one
of
the
initiators
of
the
work
was
some
eaching,
but
we
don't
know
sure
I
really
assigned
it
to
him
because
he's
not
present
the
meeting.
Maybe
we
should
ask
him
if
he
wants
to
take
point
with
he
plan
this,
because
he
is
already
very
knowledgeable
on
the
framework
and
person.
I
have
modem
no
understanding
of
what
demands
individual
six
have
their
test
from
the
framework.
It's
going
to
be
very
hard
to
obtain
this
information.
G
A
Needs
to
drive
this
someone
with
experience
and
I
can
help
Shepherd,
but
I'm
not
gonna,
sign
up
to
drive
I'm
already
overloaded,
as
it
is
this
today,
especially
with
cluster
api-based
work.
Getting
that
group
on
task,
so
I
can't
I
cannot
honestly
commit
to
being
the
main
driver.
I
can
commit
to
helping
Shepherd
main
driver
and
it
be
a
great
experience
to
develop
like
sort
of
soup-to-nuts
what
it
means
to
kind
of
reflect
our
good
portion
of
the
code,
which
is
great
if
you're
a
new
contributor.
A
H
B
I
will
also
like
to
help
on
the
a
on
that
point.
Okay,.
A
A
H
G
A
Had
Friday
Friday
spec
down
as
the
time,
but
we
didn't
work
out
well
for
West
Coast
and
we
did
a
did
a
quick
vote
last
time
about
who,
what
time
worked.
I
don't
know
if
you
have
suggestions
for
maybe
earlier
time
on,
Monday,
which
it
actually
is
not
bad
for
me,
but
might
be
bad
for
West
Coast.
Maybe
you
look
at
me
really
can
poke
Zeitung
to
figure
out
a
happy
path
between
you,
two
and,
if
they're,
like
maybe
an
hour
or
two
earlier,
it
works
better
than
that's.
Okay,.
G
A
H
I
had
a
quick
question
say:
testing
is
on
the
calendar
to
get
an
update
at
the
community
meeting
this
week
and
I
wanted
to
make
sure
that
I
included
updates
from
all
of
our
sub
project
testing
Commons
being
one
of
them,
I
was
just
gonna.
Go
off
of
sort
of
what
you
have
in
the
meeting
notes
is
the
sort
of
triage
priority
when
I
wasn't
sure
if
there
were
specific
accomplishments
that
she
wanted
to
talk
about
or
specific,
like
calls
to
action
that
you
wanted
I
think.
A
Okay,
so
the
call
to
action
would
be
like
hey
we've
had
this
we've
had
this
documented
approach
for
how
we
want
to
configure
components,
some
kind
of
set
of
kubernetes,
so
we
exposed
features
and
knobs
to
the
wild
and
we've
had
it
documented
for
what
years
us
I
don't
know
at
least
ten
years
done
and
we're
still
like
not
very
far
along
with
some
of
it
right
and
please
go
talk
to
Michael,
toughen
and
Lucas
in
the
component,
config
working
group,
okay,
otherwise,
steady
progress,
son
refactoring
the
framework
as
we
get
part-timer
is
kind
of
flowing
through
this,
but
it's
been
I
would
say
it's
been
positive.
A
A
We
should
add
part
of
this
part
of
the
action
ends,
for
this
list
was
that
we
should
have
a
guideline
stock,
that
George
said
he
was
going
to
work
on,
and
he
said
to
poke
me
first
about
where
the
location
should
be,
which
and
I
said
should
be
when
next
I'll
but
other
stuff,
which
includes
the
tag
references.
But
we
should
put
a
guideline
stock
up
and
see
if
we
can
get
that
in
place.
Okay,.
H
A
I
I
I
I
A
H
You
know,
communities
itself,
sort
of
has
like
three
layers
of
testing
unit
integration
and
to
end
and
some
links
to
sort
of
Doc's
that
describe
that
stuff.
There
are
all
sorts
of
other
aspirational
Doc's
about
like
good
practices
and
ways
to
write
tests
that
I,
suspect
are
out
of
date,
but
I
feel
like
this
is
a
good
landing
page
to
start
and
to
get
to
get
you
to
figure
out
like
what
are
the
next
questions
that
you
would
like
that.
I
Ok,
yeah
I'm
very
familiar
with
the
way
that
documents
can.
I
Can
vary
in
their
quality
and
accuracy
and
up-to-date
Ness,
so
that's
not
a
problem
at
all
really
is
I
think
this
is
probably
what
I
was
looking
to
get
is
really
just
a
way
to
start
to
understand
how
the
project
works,
and
my
objective
is
eventually
to
start
to
get
a
feel
for
how
much
we
could
actually
trust
the
kubernetes
platform
to
remain
stable.
If
we
put
large
scale
production,
those
on
to
it,
I
mean
that's.
H
Fair,
so
you're
in,
what's
called
a
sub
project
that
the
kubernetes
project
is
broken
up
into
a
bunch
of
special
interest
groups,
special
interest
groups,
sort
of
our
parents
of
sub
projects
which
can
home
one
or
more
repositories
or
pieces
of
code.
This
is
the
testing
common
sub
project,
which
is
all
about
sort
of
the
actual
test
frameworks,
and
then
the
writing
of
the
tests.
Sic
testing
sort
of
as
an
umbrella,
is
about
okay.
We
don't
actually
write
the
tests
for
people
right.
H
We
just
want
to
enable
the
best
practices
and
have
the
tools
available
for
people
to
write
tests
and
then
make
sure
that
they
are
run
and
their
results
are
collected
and
used
and
displayed
in
an
effective
manner
and
I'm
actually
supposed
to
be
summarizing.
What
it
is
that
we
have
done
what
it
is
you're
planning
on
doing
over
the
next
quarter,
so
sig
testing
has
meeting
tomorrow
at
1:30,
Pacific
and
they'll
be
giving
an
update
of
the
community.
H
I
Gonna
make
sure
so
I
have
sick
testing
4:00
to
4:30
East
Coast
time
tomorrow,
yeah,
you
said
there
was
something
on
Thursday
as
well.
Yeah.
H
A
H
Yeah
well,
the
court
having
you
seeing
where
you
can
contribute
that's
sort
of
how
this
special
interest
group
started
in
the
first
place.
Was
we
needed
to
make
sure
that
project
is
well
tested
in
a
sane
and
effective
manner
and
that
contributors
from
all
across
the
project
could
actually
contribute
in
meaningful
ways?
Be
that
code
tools,
test
results,
opinions,
tests,
things
of
that
nature,
yeah.
I
A
Would
love
help
patches
welcome,
as
you
heard
from
the
earlier
conversation
piece
about
starting
a
document
to
overhaul
what
we
would
like
I?
Think
if
you're
interested
and
have
resources
and
time
to
devote
towards
this
plus
experience,
I
think
that
would
be
a
great
way
to
have
feedback
come
in,
especially
if
you
have
years
of
experience
to
outline
the
good,
the
bad,
the
ugly,
and
we
can
talk
about
the
good,
the
bad,
the
ugly
that
we
see
in
this
communities.
Sure.
I
A
Great
well
met
by
next
meeting.
Hopefully
George
and
hippy
will
have
the
formation
of
a
dock
that
we
can
start
to
discuss
in
the
meantime,
there's
a
there's
the
existing
backlog,
which
has
actually
been
prioritized
at
the
top
of
this
document,
which
outlines
umbrella
issues
that
of
the
current
work
items
that
we're
trying
to
do.
I
Would
you
be
able
to
just
put
that
dock
into
the
chat?
Some
things
like
one
of
the
one
of
the
problems
I'm
having
is
that
kubernetes
documentation
is
both
voluminous
and
sparse
at
the
same
time
and
trying
to
find
the
right
documentation
with
the
right
content
is
a
you
know:
it's
a
bit
of
a
challenge:
you're,
first
starting
up
yeah.