►
From YouTube: 20210406 SIG Arch Conformance
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
Hello,
everybody
and
welcome
to
the
april
sixth
or
seventh,
the
big
architecture,
conformance
subgroup,
meeting
and
clayton
and
rion
and
steven,
and
I
are
all
know
each
other.
We
know
the
code
of
conduct
that
we
all
adhere
to
and
we
always
treat
each
other
kindly.
If
not
let
let
me
know
or
let
somebody
and
the
chairs
of
the
city
architecture.
A
We
have
an
agenda
today
and
I'll
pull
that
up
via
our
screen
share.
That
is
the
one
that
says
start
page.
A
Got
load
shedding
going
on
right
now,
which
means
they
turn
the
power
off
for
periods
of
time
and
so
he's
only
on
via
phone
and
can't
leave
the
meeting
today.
But
our
first
item
up
on
the
agenda
is
the
discussion
on
blocking
conformance
changes.
We
had
some
patches
get
applied
to
some
tests
that
changed
conformance
metadata,
which
is
something
after
a
release.
We
never
wanted
to
happen
because
it
we
have
defined
that
release
once
it's
out
the
door
conformance
wise
and
then
we
generated
some
really
really,
not
so
lovely
things.
A
So
the
initial
pr.
This
is
the
commit
that
went
in
and
here's
where
we
are.
If
we
go
back
to
see
the
initial
pr
that
the
slack
discussion,
if
you
want
to
go,
we
click
on
this
link
here
to
go
to
that
slack
discussion.
But
as
far
as
our
discussion
points,
it
looked
like
the
right
thing
to
do,
but
if
we
go
back
and
look
in
it
all
it
did
was
add
these
reposts
and
what
was
great
was
nikita,
not
only
updated
the
configuration
here.
A
She
also
updated
the
blockade
plug-in
so
that
it
would
support
branch
regular
expressions.
So
any
branch
or
that
has
released
star
and
it
would
not
work.
However,
we
were
hitting
a
nil
pointer
and
our
hook
blocking
pre-submits.
So
I
wonder
if
we
need
to
wait
until
the
next
release
is
done
before
we
pick
it
up.
A
I
just
wanted
to
make
sure
we
didn't
forget
this,
because
it
did
cause
an
unknown
number
of
people
to
fail
conformance
for
a
while
as
a
tooling,
because
the
tool
chain
takes
a
bit
to
update
all
the
way
through
to
when
our
friends
over
at
the
vmware
boy
picks
up,
because
it
compiles
in
the
binary
from
the
edd
thing
from
the
release
which
doesn't
match
the
metadata
that
we
pull
currently
to
validate
them
from
so
like.
A
We
have
austrian,
and
you
may
not
have
any
feedback
on
this
claim,
but
it's
yeah
so
we'll
just
go
ahead
and
skip
that.
If
you
see
something,
that's
not
of
interest
to
you,
we
can
skip
where
you
are
our
soul
audience.
Today.
A
A
Now
this
is
a
it's
a
longer
test,
so
it
may
take
a
bit
of
time
to
walk
through,
but
that's
one
of
the
reasons
we're
here
to
go
ahead
and
catch
these
as
we
go
from
issue
describing
what
we're
going
to
do
to
this
final
portion,
where
we're
submitting
it
as
a
pr
is
this
an
area
that
you
have
any
authority
or
expertise
in
clayton.
E
E
E
B
Yeah
this-
I
don't
see
anything
concerning
in
this
for
sure,
so
I
mean
really
we're
just
waiting
for
approval.
Was
there
an
open
question?
I'd
missed
sorry.
E
No,
it
was
more
around
that
this
was
just
the
style
that
I've
sort
of
been
doing
for
a
lot
of
the
status
in
points.
So
it's
more
when
we
look
at
another
pr
later
that
I've
currently
having
issues
to
do
with
patching.
That
would
be
good
to
get
some
feedback
on
that.
One.
B
Yeah
I
mean
the
when
we
I
saw
this
in
the
other
one
that
we
talked
about
so
last
time.
So
I
mean
this.
Is
I
don't
see
anything
wrong
with
the
style
of
the
approach?
I
think
we
should
discuss
it
elsewhere.
B
A
A
So
the
next
one
is
a
damon
set
next
week.
Okay,
so
damon
set
list
and
delete
connection
test
this
next
one
open
link.
A
This
one
is
interesting
in
that
we
can
see
these
two
extra
endpoints
that
we're
hoping
to
hit
and
we
go
through
our
approach,
but
someone
else
wrote
a
pr
for
it,
which
is
great.
We
like
it
when
the
stuff
gets
written.
I
think
they
did
a
different
approach,
and
so
we
kind
of
want
to
look
between
these
two
and
decide
which
approach
do
we
want
between
them?
However,
rion
and
steven
ran
this
and
as
far
as
when
we
run
this
test,
we
don't
see
it
hitting
these
extra
endpoints,
even
though
it
says.
E
Around
at
the
moment,
when
it
tries
to
do
the
delete
part
of
the
test,
it's
actually
got
some
a
bit
of
a
logic
issue
and
it's
failing
the
test
it
times
out
after
five
minutes.
E
So
it's
it's
he's,
had
some
feedback
from
emoji
already
on
the
test
and
some
of
the
guys
from
sick
absolutely
done
a
little
bit
already
on
the
teeth.
Yeah,
so
rihanna's
already
just
posted
about
a
bike.
A
And
we
let
him
know
the
meetings
that
he
could
have
depending
on
his
time.
Yes,
in
china,
the
the
person
who's
working,
these
tvs,
we
might
try
to
reach
out
and
see
if
he
wants
some
help
running
the
tooling,
so
he
can
actually,
within
his
dev
environment,
ensure
that
he
knows
how
to
hit
those
end
points
that
they
show
up.
A
A
E
E
E
What
would
the
logic
be,
but
the
other
thing
too
is
that
I
thought
that
that's
actually
just
waiting
for
it
and
the
delete
should
happen.
But
when
I
ran
out
of
my
cluster,
it
was
not
actually
doing
the
delete,
even
though
it's
got
the
labels
selected
there.
E
B
E
E
I
don't
see
how
the
test
is
failing,
but
basically
the
test
times
out,
because
the
initial
call
for
it's
got
a
timeout
of
five
minutes
and
it
never
actually
gets
to
return
true.
B
Yeah,
I
wonder
if
there's
a
yeah,
I
mean
it's
possible,
that
label
selection
david
said
it's
deleted.
That'd
be
a
great
bug
find,
but
I
don't
know:
we'd
have
to
see
someone
test
it.
I
might
just
want
to
test
it
directly
see
if
that
works.
A
A
A
C
This
just
doesn't
have
a
pr
for
it
yet,
but
this
is
the
I'll
drop
that
into
this
is
just
on
the
branch
to
make
use
of
the
rest
of
the
things
that's
inside
clayton,
I
dropped
a
link
in
the
chat.
E
Yeah,
so
both
me
and
ryan
ryan
started
the
initial
part,
and
I
came
along
and
just
did
a
few
extra
checks
and
it
is
both
hitting
list
and
also
it
goes
through
the
process
of
actually
doing
the
delete.
Based
on
how
we've
been
doing
our
watch.
E
B
B
E
I
I
had
a
test
about
a
year
ago
that
was
looking
at
deleting
a
whole
of
pods
and
jordan
basically
decided
or
came
back
with
some
feedback
around
waiting
for
the
default.
30
seconds
wasn't
really
appropriate
and
he,
since
then,
I've
been
actually
changing
the
grace
period.
B
B
When
you
run
an
image,
you
can
set
a
shorter
grace
period
on
the
pod,
that's
what
we
should
do
and
you
should
not
be
specifying
grace
period
zero
in
these
tests
like
grace
period.
Zero
is
like
should
never
be
used.
We
should
have
tests
for
it,
it's
not
a
conformance
thing
and
it's
not
testing
the
normal
behavior.
It's
testing
weird
behavior,
grace
period.
Zero
is
like
like
especially
on
pods.
So
what
I
would
probably
say
is
in
this
example
that
you
guys
send
online
98
a
98.
B
You
should
set
termination
grace
period
to
one
second
here
on
the
pod,
which
means
that
the
calls
will
all
finish
in
a
second
or
two
and
that's
right
and.
B
B
In
some
cases
that
might
be
okay,
but
it's
abnormal.
It's
not
our
normal
behavior.
I
don't
want
to
see
conformance
tests
using
zero.
They
should
just
have
containers
that
have
short
termination
periods
and
when
you
set
that
on
the
pod,
then-
and
you
say,
delete
you
know,
it'll
take
a
second
or
two.
It's
much
faster.
E
Okay,
it
was
just
yeah.
I
wish
I.
B
Caught
that
when
jordan
said
it,
because
I
definitely
would
have
gone
the
other
way,
then
I
would
have
saved
some
time
later
yeah.
It
was
the
other
place
that
it
set.
I
commented
on
8
29,
but
then
you
said
another
one
earlier
where
we
let
me
I'm
going
to
comment.
Oh,
I
can't
comment
and
I
mean
let
me
place
this
link
back
so
inside
your
nude
well
hold
on
all
right.
B
At
that
link,
oh
yeah,
just
you,
you
would
just
add
inside
the
pod.
Spec
set
termination
grace
period
seconds
one
and
when
you
do
that,
basically,
it's
like
setting
a
very
short
grace
period
but
happens
on
the
pod.
B
The
cuba
does
all
the
work
for
you,
but
you're
going
through
the
normal
deletion
process
and
you're
not
triggering
that
at
normal,
because
yeah
and
it
would
probably
just
be
great,
like
just
go
review
all
the
places
tag
me
on
something
that's
doing
grace
period
zero
and
we'll,
like
all
of
them,
that
you're
creating
pods
this
approach.
Once
you
get
it
like,
once
you
have
the
right
syntax
and
all
that
this
approach
solves
it
for
everybody.
We
do
that
in
other
tests.
B
E
That's
cool,
let's
just
again
works
for
seeing
a
a
technique
used
in
one
part,
to
solve
particular
issues
around
the
test,
timeout
the
length
of
a
test
being
taken
and
just
trying
to
pick
up
other
techniques
that
have
been
provided
so
yep.
Thank.
A
Thank
you
open
link
for
this
other
one.
We
are
yeah.
This
is
the
one
you're
saying
that
we're
trying
the
same
thing
we've
done
before
and
now
we're
failing.
D
A
E
Basically,
when
I
try
to
do
for
replica
set
what
I've
been
doing
for
all
the
other
status
endpoints,
it
works
for
get
and
for
replace,
but
as
soon
as
I
try,
if
we
just
scroll
up
a
little
bit
for
that,
so
as
the
test
goes
through,
it
works
okay
for
doing
get
and
replace,
but
as
soon
as
it
tries
to
do
the
patch.
E
E
E
If
we
just
scroll
down
it's
they
after
the
watch
yeah
down,
the
very
bottom
is
where
I'm
just
patching
the
condition
and
type
and
the
status
by
leaving
those
two
exactly
the
same
as
what
I've
done
on
the
other
test,
they've
all
passed,
and
yet
it's
either.
There
is
something
that
the
replica
set
is
either
tripping
up
on
when
I'm
changing
the
patch
or
what
is
getting
tripped
up
here
should
be
happening
elsewhere.
I
don't.
B
Know
you
should
have
you
printed
out
what
the
patch
that's
being
sent
is
so
like
online
like
after
this
chunk,
so
where
you're
patching
the
status.
Are
you
printing
out
the
actual
patch
that
you're
sending
to
the
server.
F
B
Is
I
bet
you
you're
marshaling
a
struct
here
and
I
bet
you.
The
zero
values
of
all
of
the
fields
that
are
under
replica
set
status
that
are
not
conditioned
are
also
getting
set.
So
even
though
you
didn't
intend
to
change
those
fields,
I
bet
when
you
look
at
what
payload,
if
you
print
out
payload
it's
going
to
show
you.
You
said
it's
going
to
say
like
replica
set
status
and
then
inside
there's
gonna
be
a
whole
bunch
of
ready,
replicas
zero,
blah
blah
blah
right
zero.
B
If
you
write
the
json
directly
since
it's
a
very
static
patch
and
it's
you
know,
it's
going
to
look
like
the
json
struct
of
whatever
this
is,
but
the
difference
is
is
when
you
serialize
a
go
struct
you
depend
on
the
default
behavior
of
the
other
fields.
So
if
they
don't
have
omit
impression,
they're
going
to
be
included,
it's
just
easier
to
usually
write
the
patch
another
option,
and
so
this
is
the
other
one
which
is.
There
is
a
function
that
we
have
in
a
couple
of
places.
B
I
don't
know
if
it's
in
cube
control,
it
might
be
in
the
test.
Libraries.
You
can
generate
two
structs,
so
you
do
one
struct,
that's
just
empty
replica
set
status,
and
then
you
do
a
second
struct,
which
is
has
your
condition
in
it,
and
then
you
can
get
it
to
calculate
the
diff.
We
use
that
in
cube
control
in
a
couple
places,
but
it
may
not
be
shared
code,
so
you
may
not
be
able
to
access
it.
That
code
actually
goes
and
says.
B
E
Awesome
tried,
I
tried
to
hunt
it
and
I
found
a
reference
about
the
valid.
I
tracked
the
error
to
validation
check
but
yeah.
I
was
just
going
around
in
circles
a
bit.
A
A
A
B
F
B
B
F
B
F
Oh,
I
don't
know
why
it's
not
pasting
all
right.
Okay,
let's
do
what
channel
are
we
in
I'm.
B
B
Roughly
you're
going
to
be
calling
doing
something
like
that.
You
guys
see
this
yes,
yeah,
okay,
so
the
first
one
just
create
like
the
empty
structs.
So
like
the
first
one,
let
me
update
these
in
line
so
json.marshall.
B
So
the
first
one
you're
gonna,
do
something
like
this.
B
So
the
first
one
you're,
basically
marshaling
the
empty
one.
That
would
be
your
original
js
and
then
you
take
your
patch
here
and
marshal
it
and
that's
your
changed
js
and
then
called
the
merge
patch
from
all
the
new.
What
that'll
do
is
all
the
equal
things
get
stripped
out
so
that'll
handle
this,
and
then
you
can
use
that
elsewhere,
which
is
all
the
places
you're
calculating
patches.
B
You
could
do
this,
you
could-
and
I'm
not
saying
this
is
like
what
you
should
do
but
like
there
probably
should
be
a
helper
for
creating
merge
patches
like
this
in
the
test
code,
at
least
so
you
could
just
say
given
to
interfaces
like
a
function,
takes
two
interfaces
and
gives
you
a
merge
pad,
and
then
you
give
it
the
base.
You
just
pass
in
an
object.
Json
marshall.
It
pass
in
a
second
object:
json
marshall.
It
call
that
method
gives
you
back
a
merge
patch.
E
Oh
probably
just
going
to
the
atoa
framework
yeah.
B
I
created
new,
I
create
a
new
package
for
it
probably
like
don't
like,
I
would
say,
don't
stick
stuff
in
general.
Packages
should
be
like
test
util
patch
or
something.
If
there
is
there
probably
is
a
patch
folder,
let's
say
test.