►
From YouTube: Layer5 Community Meeting (Nov 13th, 2020)
Description
Conformance testing with @layer5 sample app and @smp_spec specifications.
C
I
miss
I
mean
I
missed
the
talk
from
mystery
from
yesterday
from
all
day
devops.
I
could
not
stay
much
more
online.
It
was
a
week
with
a
lot
of
conferences
this
week.
I
don't
know
what
is
happening,
but
there
are
so
many.
In
the
same
time,
it
was
the
digital
ocean,
one
the
one
from
if
influx
db,
something
from
tigera
cubecon
next
week
and
all
day,
the
box
also.
C
It
was
too
much
it's
like.
I
don't
know
it's
like
a
virus.
It's
too
much.
It's
too
many.
In
the
same
time,.
A
Yeah
there
is
yeah
totally
yep,
a
re
invent
is
coming
up
and
it's
it'll
be
three
weeks
long.
C
C
A
A
Very
good,
well,
if
you're,
so
by
the
way
we
should
so
talking
about
kubecon,
there's
going
to
be,
there
was
going
to
be
four
talks
about
measuring
smp
and
smi
conformance.
It's
just.
You
know
too
much,
I
think,
for
the
collection
of
us,
so
we
moved
it
back
to
three
talks,
but
there'll
be
three
talks
next
week
that
highlight
the
works
that
are
being
done
here
and
I
think
there
are
links.
A
I
think
we
there
are
probably
links
to
them
in
older
meeting
minutes,
but
you
should
all
you
don't
all
have
to
attend
or
have
to,
but
you
should
all
minimally
take
pride
in
the
fact
that
your
works
are
out
there
pretty
pretty
neat.
A
I
was
just
on
a
an
advisory
board
called
with
cisco
this
last
hour.
Hence
the
lack
of
prep
in
the
meeting
minutes,
but
I'll
say
that
it
isn't
without
it's
within
reason
that
things
like
s
p
in
particular,
are
and
measuring
actually
are
of
interest
to
cisco
and
in
that
in
those
conversations.
So
so
it's
good
who
has
a
call
for
topics
if
you've
got
topics
or
questions
or
things
toss
them
in.
A
If
you
don't
shredy
is,
I
think,
going
to
list
out
some
some
newcomers
in
the
community
that
joined
this
last
week.
We've
got
a
few
of
them,
and
actually
I
don't
know
that
we
have
any
newcomers
on
the
call
right
now.
A
A
So
so
there's
there's
some
things
to
chat
on
kush
you're.
Here
you
had
a
topic
la
this
last
wednesday
on
aks.
I
think
we
covered
it,
but.
E
A
So
curious
dhruv
did
your.
Did
your
name
go
up
on
the
schedule?
Are
we
are
you
a
kubecon
speaker
at
this
point.
A
Or
are
they
not
listing
their
names?
Okay,
they're,
not
listing
their
names,
but
all
right,
but
here's
one
of
the
here's.
Here's
one
of
the
places
where
we'll
be
discussing
the.
A
A
E
E
E
A
Sorry
to
interrupt,
but
actually
as
people
go
to
give
feedback
it
might
it
would
be
good.
If
can
you
do
a
pull
on
the?
Can
you
switch
to
master
because
I
think,
because
this
got
merged
and
I
tossed
in
a
few
updates,
as
I
did,
that
just
some
user
experience
updates,
if
you
you
can.
A
A
One
thing
that
I
didn't
do
is
remove
the
output
that
we're
giving
about.
I
think
the
the
azure
cli
version,
which,
for
my
part,
I
consider
unnecessary
that
hey.
If
we
discover
it
and
are
able
to
interact
with
it,
that's
good
and
the
user.
E
A
Actually,
yeah
a
couple
couple
of
things
on
this
because
actually
like
but
part
of
it,
I
don't
mean
to
interrupt
your
presentation,
but
just
to
I
want
to
make
sure
that
so
everyone
that's
on
this
call
today
is
is
a
measuring
contributor.
An
s
p
you
know
is,
are
contributing
and
contributing
code,
and
each
of
you
are
designing
the
experiences
and
going
through.
A
I
want
to
walk
through
this
a
little
bit
to
say
that
the
the
initial
that
initial
experience
of
kind
of
giving
people
output
that
that
there's,
you
know
that
json
that
we're
showing
would
be
if
at
some
point
we
ever
had
a
verbose
flag
for
measuring
ctl.
A
We
could
include,
you
know
debug
like
that,
but
we
otherwise
wouldn't
otherwise
we
would
otherwise
just
say
configuring,
meshri
access
to
aks
and
the
dot
dot
is
important
because
we're
indicating
to
them
that
this
is
a
a
process
and
a
procedure
and
that
the
call
the
colon
at
the
end
of
the
line
you
know
we're
giving
them
an
instruct
we're
not
asking
a
question.
So
we
don't
have
a
question
mark
but
we're
giving
them
an
instruction
in
this
case,
and
it's
you
know
it's
a
direct
we're
using
direct
language.
A
A
Yeah
for
people
who
who
are
savvy,
who
already
know
that
like
this
is
the
requirement
that
you
need
to
provide
arguments
that
have
this
for
people
that
aren't
like
that.
This
is
not.
This
is
pretty
pretty
good
like
it
is
the
same
experience
that
you'll
have
with
the
aws
cli
or
the
aks
cli
that
when
you
go
to
interact
with
those
for
the
first
time,
they'll
they'll
ask
you
some
questions
as
well,
and
so
it's
supporting
both
is
what
I'm
saying
is
like,
ideally
like.
A
One
thing
I
tossed
in
and,
and
I
don't
know
this
is
just
you
know-
this
was
a
thought
at
the
time
was
before.
If
you
were
not
to
enter
in
a
name,
it
would
exit
the
the
the
process
would
exit
and
it
would
give
you
the
standard
out
error
or
not
the
standard.
It
would
give
you
the
standard
error
which
was
which
I
removed
and
like
because
we
don't
need
to
to
tell
them
these
things.
So
so,
actually,
what
I
did
here
is
not
only
did
I
remove
the
well.
A
I
left
in
that
that
ugly
error
unexpected
new
line,
but
I
but
I
changed
it
to
let
them
try
it
again.
So
at
least
one
time
I
we
just
said:
error,
re
air
reading
the
cluster
name,
unexpected
new
line,
and
actually,
I
think,
that's
still
a
little
confusing.
But
what
we
say
is:
let's
try
again.
Please
enter
the
aks
and
in
this
case
like
it's
wrong,
there's
a
bug
it
should
say.
Please
enter
the
aks
resource
name,
which
is
the
first
question
that
you
know.
A
The
other
item
and
kush,
I
think,
you're
aligned
your
thinking,
is
aligned
with
this
is
that
that
we
would
ideally
not
create
a
temporary
entry
in
a
file
system,
but
rather
use
it
in
the
dot
mystery.
A
A
A
A
I
only
made
it
sort
of
halfway
through
all
of
what
this
is,
but
when
you
have
that
new
file
that
has
the
right
certificate
information
for
meshri
to
be
able
to
connect
to
aks,
were
you
using
the
dash
dash
the
global
flag,
dash
dash
config.
A
E
A
E
What
it
do
is
it
just
fetches
the
conflict
from
this
location
and
list
out
the
context
which
are
available
and
then
site
set
context
is
involved,
which
list
the
context
and
ask
you
which
context
you
want
to
set
for
machine
and
then
it
and
and
then
it
makes
a
put.
I
guess
it's
a
put
request
to
mystery
api
to
select
the
following
context.
A
Okay
yeah,
so
that
makes
sense
if,
if
we're
just
moving
the
location,
then
that
the
rest
of
the
experience
sounds
sounds
right.
A
A
F
B
A
E
A
As
opposed
to
asking
people
to
take
manifest
either
use
helm
or
take
manifest
to
go,
deploy
measuring
that
we
can
we'll
still
offer
that
as
an
option,
but
that
we
would
be
able
to
provide
a
better
experience
by
using
instead
of
measuring
ctl
systems,
start
that
assumes
that
you're
going
to
deploy
locally
to
docker.
There
would
be
another
flag
or
there
would
be
a
question.
That's
asked
after
that.
That
says:
oh
did
you
mean,
did
you
want
to
do
it
locally,
or
did
you
want
to
do
it
in
a
to
a
kubernetes
environment?.
E
A
Yeah,
it's
it's
actually
egregiously
missing.
It's
been
there's
been
an
open
issue
on
it
for
a
while,
a
few
people
brought
it
up
and
really
the
the.
I
think
that
the
primary
reason
why
that
hasn't
gone
forth
is
just
about
a
design
spec
that
gives
holistic
thought
to
what
we've
talked
about
in
the
past
to
profiles
and
contexts
like
bringing
forth
the
the
if
you're
familiar.
You
know
you
used
cubeconfig,
you
I'm
sorry.
A
A
We
need
to
bring
in
the
notion
of
a
context
and
then
that
will
help
one
it'll
be
more
powerful
to
be
able
to
manage
your
multiple
clusters
and
which,
where
you
want
mesry
to
connect
to,
but
two
it
will
also
avoid
long
command
line.
Arguments
of
saying,
dash,
dash,
kubernetes
but
like
which
kubernetes
did
you
mean
and
then
beyond
that
which
you
know
like
which
name
space?
Did
you
mean
and
stuff
like
that?
So.
D
A
That's
yeah:
this
is
exactly
kind
of
the
some
of
the
initial
thoughts
it
was.
This
was
back
when,
if
you
see
that
in
the
in
the
comment
it
says,
if
you
go
back
for
a
second
kush
in
the
comment
it
says
measure
ctl
start
dash
p
like
for
platform.
A
It's
in
the
middle
of.
B
A
G
H
I
was
working
on
the
mesh
commands.
I
didn't
get
into
context
because
by
the
time
I
decided
to
move
on
from
battery
ctl
context,
amplifiers
was
still
pretty
new.
Okay.
A
Okay,
anyway,
so
I'm
kind
of
digressing
as
to
a
whole
other
set
of
conversations,
but
but
yeah
so
provide
so
husseina
is
looking
at
supporting
mesh
resist
measuring
ctl
system,
config
eks,
which
is
good.
A
I
think
for
eks.
This
story
gets
a
lot
more
complex
when
you
deploy
meshrie
or
any
app
inside
of
eks.
My
understanding
from
abhishek
is
that,
if
you
want
to
integrate
with,
I
am.
A
A
So
if
anybody
has
experience
there
or
interest
there,
please
see
husseina
because
she's
working
on
that
kush.
Thank
you
for
sharing
that's
good.
Any
other
comments
for
cush.
A
Look
at
this
chris:
do
you
mind
I'm
adding
the
notes
about
parameters
or
arguments.
I
I
I
Oh,
can
you
see
my
screen
yep?
Okay,
so
for
the
people
who
are
not
don't
know
what
about
efforts
are
regarding
smi
sms,
basically
an
interface
with
certain
specifications
that
they
have
defined
for
to
generalize
the
the
capability
of
a
service
right,
so
for
seeing
whether
a
particular
service
muscle
is
compliant
or
not.
We
are
defining
certain
set
of
test
cases
for
each
of
the
spec
which
they
are
within,
and
these
are
currently
ongoing,
work
and
yeah.
I
I
So
that
is
how
we
define
whether
a
service
message
conform
and
do
that
particular
the
smi's
picture
and
yeah.
So
that
is
the
main
nutshell,
and
what
we
are
doing
right
now
is
currently,
there
are
two
things
which
we
want
to
do
regarding
sms
aspects.
I
One
would
be
to
run
your
test,
your
test
on
your
local
mesh
or
your
mesh
itself
to
see
whether
it
is
smi
compliant
or
conformant
or
not,
and
you
can
currently
do
that
using
adopters
and
measure
itself
like
I
don't
have
it
deployed
so
I'll
just
show
a
screen
shot
quickly,
so
yeah.
So
this
is
this
function
is
right
now
you
can
do
it
properly,
and
this
is
where
to
go.
If
you
find
any
bugs,
then
please
tell
us
and
except
that
we
also
wanted
to
update
the
service,
miss
landscape,
which
we
have
right.
I
This
is
basically
telling
whether
a
particular
service
message
smi
over
here
it
is
com,
smi,
compliant
or
not
right.
So
what
we
have
in
mind
is
each
salvage
mesh
provider
will
have
their
own
will
have?
They
will
have
integrated
measuring
cdl
in
their
ci
cd
pipeline,
in
which
they
will
run
this
test
and
those
particular
tests.
I
We
would
save
in
our
machine
cloud
and
we
will
use
them
to
show
a
more
detailed
table
data
of
how
much
compliance
at
all
is,
if
not
so
right
now
we
are
working
on
that
particular
capability
and
in
the
landscape.
This
is
the
current
layout
of
the
table,
which
we
are
thinking,
and
so
this
is
a
very.
This
is
in
a
prototype
stage,
don't
mind
the
alignment
and
small
things
like
that.
I
We
can
like
discuss
about
the
structure
of
the
tables
itself
that
currently,
if
we
click
on
a
particular
mesh,
we
would
see
the
previous
version
of
that
recommendation,
whether
they
were
into
that
particular
stake
or
not,
and
currently
this
is
just
the
data
which
we
have
filled
up
manually.
These
are
not
the
tests
which
we
had
done,
which
will
be
updated
and
yeah
like.
We
have
idea
that
the
tooltip
will
create
will
have
more
data
about
why
we
are
passing
or
why
they
are
not
processed
and
stuff
like
that,
so
yeah
any
comments
on
tutorial.
A
No,
no,
no!
I
I
was
going
to
say
something
so
there's
the
next
smi
meeting
smi
meetings
happen
twice
a
month.
The
next
one
we
weren't
going
to
have
another
one
this
year.
A
But
we
changed
that
and
put
one
in
for
not
this
coming
wednesday,
because
that's
kind
of
a
week
of
kubecon
but
the
following
wednesday.
So
I
think
it's
the
25th
on
that
wednesday.
That
meeting
is
dedicated
to
anyone.
Who's
participating
in
smi,
it's
dedicated
to
discussing
smi
conformance,
is
to
to
this
initiative.
F
A
Oh
now,
just
letting
everyone
letting
everyone
know
in
two
weeks
wednesday,
the
25th
we'll
have
an
smile
meeting
that
is
dedicated
to
smi,
conformance
and
so
good,
for
you
know
open
to
anyone
that
wants
to
attend,
but
good
for
us
to
get
this
table
up
to
help.
People
understand
what
part
of
the
goal
is.
A
If
all
of
that
level,
I
don't
know
what
we'll
see,
I'm
not
sure
if
all
of
that
level
of
detail
is
necessary
here
in
a
landscape
like
this,
I
it's
quite
expected.
A
I
would.
I
certainly
anticipate
that
these
results
will
be
shown
on
meshri.io
here
on
the
landscape,
but
they'll
probably
also
be
shown
on
sm
smi.
A
Spec.Io
and
so
yeah
as
it
turns
out,
we've
been
getting
more
and
more
into
smi,
so
the
main
the
meet
the
maintainers
is
something
that
that
I
had
organized
to
so
that
we
can
go.
Tell
people
about
s,
my
smi,
but
also
about
smi
conformance
and
so
there'll,
be
a
few
other
smi
maintainers.
That
will
join
that
session,
which
reminds
me
I
have
an
action
item
to
go,
create
a
few
slides
so
that
we
can
have
them
in
the
background
as
people
come
to
that
session.
A
A
So
anyway,
there's
a
curated
set
of
all.
All
of
them
are
svgs
and
all
of
them
yeah,
so
all
of
them
are
svgs
and
all
of
them
have
been
like
trimmed
like
you
were
saying,
there's
two
the
same
two
sets
of
those
there's
one
in
the
mesh
redox
for
the
adapters
and
then
there's
another
one
in
meshri.io,
so
you
can
go
grab
a
copy
of
those
from
the
other
one
of
those
two
robots.
A
Oh
one
yeah
yeah,
one
thing
too
is
gerald,
had
I
think
in
his
css
updates,
had
removed
the
padding
and
the
tables
up
above
and
that.
A
Cool
nice.
Thank
you
drew
really
important.
I
was
just
showing
this
to
a
principal
engineer
at
hpe
yesterday,.
A
Yeah,
no,
that's
fantastic!
As
a
matter
of
fact,
your
your
work
is
being
held
up
by
the
lack
of
persistence
of
the
test
results
to
a
remote
provider,
so
so
yeah
we
need
to
we.
A
I
want
to
get
past
this
particular
thing,
like
it's
really
good,
for
measuring
it's
helpful
for
people
to
recognize
meshri
as
like
a
source
of
authority
for
for
smi,
a
source
of
authority
for
smp,
but
in
terms
of
like
the
layman
or
or
the
core
persona
that
we're
targeting
with
meshary,
you
know
very,
very
few
of
them
are
going
to
care,
and
so
consequently
we
only
want
to
invest
or
we
want.
We
want
to
get
it
behind
us.
A
It's
going
to
it's
going
to
I.
Unfortunately,
it's
going
to
take
longer
than
I
had
hoped
it's
going
to
take
longer
because,
as
we
go
to
engage
with
each
individual
surface,
mesh
team
they're
going
to
it's
they're
going
to
want
to
define
the
t,
we
don't
need
to
be
in
a
position
to
define
all
the
tests
so
nginx.
A
I
So
they're
going
to
contribute
in
adding
the
test
cases.
Is
that
so.
A
A
So
bring
your
brass
knuckles.
J
Yes,
if
I
may
about
smi
my
comment
in
document,
so
maybe
it's
a
sort
of
like
a
in
the
smi
on
my
question
in
the
smi
document
that
we
looked
at
as
well.
J
Our
setup
is
sort
of
like
the
intentionals
of
smi
or
of
measuring
that
we
can
use
the
same,
learn
five
layer,
5
learn,
what's
called
application
and
deployment
descriptor
and
run
it
on
all
service
meshes
without
the,
without
the
possibility
to
to
do
some
service
mesh
specific
configurations
or
is
it
just
because
is.
J
A
In
order
for
like
as
part
of
this
smi
conformance
in
order
for
meshri
to
verify
that
a
given
service
mesh
adheres
to
the
specifications,
it
needs
to
run
tests
and
define
assertions
and
verify
them
and
in
the
process
of
doing
that,
measuring
needs
to
this
is
why
mesri
is
being
used
because
it
needs
to
stand
up
any
number
of
different
service
meshes.
Deploy
a
sample
application
generate
load,
verify
that
things
are
behaving
as
they
should
and
so
to.
A
What
michael
is
pointing
out
is
like
well,
when
you
deploy
a
sample
app
onto
one
mesh,
it
needs
it
needs
annotated
in
different
ways.
It
needs
labeled
in
different
ways.
You
know
the
one
onboarding
procedure
for
one
mesh
isn't
necessarily
the
same
for
another,
and
so
so
to
facilitate
part
of
the
this
to
facilitate
the
testing
of
smi
conformance.
A
new
sample
app
was
created
a
relatively
simple
one
that
was
instrumented
a
relatively
lightweight
and
simple
one
that
we
can
use.
A
That
is
also
instrumented
for
receiving
some
metrics
and
to
verify
things
with
the
traffic
metrics
spec,
and
things
like
that
in
process
of
creating
the
sample.
App
though
it
does
need
to
be
it's,
it's
it's
manifest,
it's
deployment
needs
to
be
compatible
with
each
and
every
mesh,
and
they,
you
know
they
do
it
differently.
So
michael's,
pointing
out
that
for
console
part
of
their
configuration
is
part
of
the
configuration
of
services
running
on
the
console
service.
Mesh
needs
to
have
particular
annotations.
A
The
right
file,
but
there's
a
there,
is
a
portion
of
what
we're
doing
here.
That
does
is
it's
a
case
statement.
There's
a
portion
of
what
we're
doing
that
takes
into
consideration
that
different
service
meshes.
A
J
I
think
the
I
think
the
labels
are
on
the
namespace
actually,
but
I
think
the
label
is
on
the
namespace
and
also
examples.
I
was
looking
at
the
open
service
mesh
in
the
code
in
the
validate
command
in
the
osm
adapter,
the
mesh
you
can
send
in
the
mesh
name
as
a
parameter
to
the
you
know
the
rom
test
function
and
it's
not
osm.
J
So
so
this
case
statement
isn't
isn't
picked
up
if
you
like,
so
I
I
was
also
a
bit
confused,
because
I
saw
this
case
statement
and
I
think
in
case
of
osm,
it's
actually
not
executed.
It's
not
going
through
this
case
statement
because.
A
There's
part
of
the
architecture
of
what
we're
doing
for
performing
this
testing
that
doesn't
sit
well
with,
doesn't
sit
well
in
general
and,
what's
going
on
here,
is
that
that
we
had
early
on
considered
that
that
we
would
like
to
use
kudo
an
open
source
project
to
help
us
perform
assertions
and
verify
assertions
in
a
kubernetes,
centric,
environment
and,
and
so
kudos
methodology
is
to.
A
Is
you
more
or
less
for
you
to
give
it
a
set
of
manifest
that
you'd
like
for
it
to
provision
in
a
separate
namespace?
And
so
that's
why
there's
a
little
bit
of
a
disconnect
between
what
the
adapter
so
the?
A
I
have
to
go
look
but
to
your
point
like
there's
a
you're
able
to
specify
annotations
and
service
mesh
specific
needs
in
different
areas.
A
J
J
That
was
any
way
that
was
my
understanding
and
that,
and
they
are
actually
defined
in
in
the
in
in
meshkit.
Aren't
they.
J
I
couldn't
say
it:
no,
no!
No,
but
you
know
the
the
sort
of
like.
Maybe
the
sort
of
the
general
question
is
the
the
learn
layer,
5
sample
application
in
the
context
of
smi
and
maybe
also
of
smp
if
the
service
mesh
specific
deployment,
and
that
can
be
a
variety
of
things.
I
imagine
you
know
service
accounts,
and
maybe
you
would
like
to
maybe
and
that's
another
question.
Would
you
like
to
test
with
different
configurations
as
well?
A
The
architecture
of
the
the
conformance
test
or
of
this
sample
app
kind
of
violates
the
the
architecture
that
we
have
for
all
of
the
other
sample
apps,
which
is
the
if
you
know
for
those
that
have
been
running
mesherry.
You
understand
that
you
can
provision
a
bunch
of
different
service
meshes
on
them.
Once
they're
provisioned,
you
can
provision
a
bunch
of
different
sample
applications
and
we
intentionally
support
provisioning.
A
The
same
sample
application
on
a
number
of
different
service
meshes
to
help
facilitate
learning
and
understanding
of
the
behavioral
differences
between
service
meshes
and
and
the
mesh
specific
annotations
and
considerations
like
that
that
there
are
individual
instances
and
descriptors
of
what
those
are
and
it's
stored
in
each
of
the
service
mesh
adapters.
That's
exactly
why
the
adapters
have
been
made
is
to
hold
the
service
mesh
specific
requirements
and-
and
that's
where
this
approach
kind
of
violates
that
that
rule
like
nowhere
else
in
or
generally
mostly
nowhere
else
in
measuring.
A
Do
you
see
a
case
statement
that
says,
if
linker
d
this?
If
rather
that
that's
the
adapter's
job
so
michael?
I
like,
as
you
familiarize
it,
may
be
the
case
that
we're
able
to
the
the
the
crux
of
the
of
the
issue,
the
reason
that
that
architecture
went
forth
here.
It
was
about
using
kudo
and
leveraging.
We
had
considered
that
it
would
be.
F
Sort
of
like
well,
we
can
get
rid
of
it,
so
it's
been
one
of
those
areas
where
you
intend
to
test
this
using.
A
A
Here's
here's,
the
thing
is
the
the
the
okay
is.
Do
we
need
kudo
to
provision
a
service
mesh
for
us?
No,
do
we
need
kudo
to
provision
a
sample
app
for
us?
No,
as
a
matter
of
fact,
it's
kind
of
getting
in
the
way,
okay?
Well,
what
do
we
need
kudo
for?
Are
we?
Are
we
well
what
what
the
framework?
What
its
framework
provides
is?
A
Is
this
when
you're
it
lets,
you
define
particular
it
lets.
You
define
a
set
of
tests
in
a
in
a
seek
in
a
series
that
you
want
to
be
verified
and
you
describe
those
in
and
so
okay,
so
that
that's
helpful,
so
we
can
go
through
and
it
sort
of
its
logic
is
it
just
follows
the
sequence
that
you
define?
You
know
zero.
J
A
Right
so
we
haven't
beaten
off
the
like
part
of
the
reason
that
so
part
of
the
reason
we
got
to
this
architecture
is
because
it's
like
well
hey.
This
is
a
good
thing.
We
should
do
this.
This
is
helpful,
but
it's
not
really
it's
like
off
to
the
side.
We
don't
want
to
invest
too
much,
and
so
that's
why
it
feels
slightly
off
to
the
side.
A
That's
not
why
it's
not
like
if
we
were
to
do
away
with
kudo
and
still
support,
smi
conformance,
which
at
this
point,
like
you
know,
we're
committed
to
what
we
would
ideally
pull
in
is
is
is
more
of
a
generic
like
workflow
engine
potenti
to
to
to
the
workflow
engine,
to
just
basically
run
the
assertions
in
is
you
know
a
serialized
and
occasionally
parallelized
like
like
here
that
the
0404
sequence
I
kind
of
control
that
have
the
output
of
whatever
that
workflow
was
returned
and
then
to
sort
of
generate
a
report
off
of
that
that
workflow
engine
will
be
leveraged
for
a
lot
of
other
things
that
we
want
to
do
in
in
measuring
the
project
itself.
A
So
it's
some
point.
It's
potentially
a
worthwhile
investment
to
displace
kudo
in
this
particular
example
of
the
issue
that
we're
facing
with
service
mesh
specific
annotations,
it
might
be
that
we're
able
to
use
customize
or
or
helm
or
a
bit
just
a
bit
of
service
mesh
specific,
go
to
overcome
that.
A
What
I
think
is
the
last
mile
you
know,
because
I
think
we're
you
know
we
because
I
think
we're
there
with
with
cuda.
I
know
that
we're
running
successfully
open
service,
mesh
tests,
we're
getting
results
back
and
and
it's
it
takes
a
while
kudo
takes
a
while,
and
but
it's
one
of
those
things
like
the
the
truest
value
of
it,
is
really
to
the
service
mesh
maintainers
that
they
would
build
this
into
their
pipeline.
A
To
have
mesh
re,
run
these
tests
and
validate
their
conformance
ongoing
because,
as
as
an
smi
maintainer
we'll
end
up
pushing
for
in
all
of
them
to
do
that
to
hey.
If
you
want
to
claim
that
you're
you're
conformant
and
you
want
to
carry
this
logo
on
your
site,
you've
got
to
be
passing
these
tests.
Hence
you
should
probably
come
participate
in
this
conversation
about
what
actual
tests
are
going
to
be
run
and
the
manner
in
which
they
run,
and
so
that
strong
arming
is
something
that
we
need
to
go.
Do.
A
F
J
The
related
order,
you
know:
where
should
we
I
ate?
It
was
my
understanding
of
how
this
is
set
up
and
works
correct
and
b.
If
we,
you
know,
want
to
make
say,
console
adapter
running,
be
able
to
run
these
tests.
Where
should
we
invest
the
time
to
like
get
these
specific
annotations
into
the
you
know
the
test
framework?
A
Deployments,
I
think
that's
about
right
because,
because
I
think
we're
kind
of
all
we're
essentially
there
already
we're
sort
of
essentially
the
one
thing
that
that
I
guess
I'm
confused
on
is
the
the
annotation
that
you
need
per
pod.
A
J
J
So
you
define
it
when
you,
when
you
register
your
services,
if
you
you
have
to
define
the
upstream
services
because
it's
not
transparent
so
that
apparently
comes
later
or
is
planned
so
like
in
istio,
you
don't
have
to
do
it.
It's
sort
of
like
magic,
yeah,
somewhere
yeah.
It.
F
A
Cool,
I'm
really
glad
that
we
have.
I
know
that
that
conversation
was
probably
really
boring
to
a
number
of
folks,
but
it's
pretty
important.
A
A
If
you're
interested
we're
going
to
get
organized
on
it
and
start
some
regular
meetings,
I'm
going
to
meet
with
a
professor
at
I'm
sorry
at
knit
k
tomorrow
with
a
couple
of
students
that
are
here
in
the
community
and
we're
going
to
discuss
some
so
we're
going
to
discuss
some
pretty
interesting
things
around.
A
Algorithms,
and
so
the
cfp
is
open
for
kubecon
eu,
which
is
it's
always
bothersome
to
me
that
before
the
first
before
the
the
current
kubecon
is
done
that
the
next
cfp
is
open.
It
makes
usually
I'm
trying
to
finish
up
slides
for
the
current
one,
but
so
shabbai.
A
Thank
you
for
pointing
that
out
I'll,
put
out
an
open
call
for
anyone
that,
if
anyone
wants
to,
if
anyone
wants
to
co-present
I'm
game
or
if
you
are
submitting-
and
you
need
somebody
to
help,
you
review
it
or
make
some
suggestions
on
what
you
might
talk
about.
A
K
K
Sharing
so
earlier,
like
I'll
quickly
showcase
so
earlier
we
had
that
sort
of
a
weird
issue
that
you
know
we
made
so
basically
what
I
did
was
that
now
the
transition
that
will
be
made
it
it
won't.
Actually
you
know
so
like
let's
say
now,
I'm
on
my
home
page
and
now
I
come
back
and
I
go
over
here.
So
it's
like
now
only
if
you
are
like
you
know
scrolling.
So
I
did
look
at
that
exact
video
lee
and
the
actually.
K
The
issue
you
know
was
that
with
strappy,
how
they
had
implemented
was
that
they
had
a
single
div
and
they
were
using
javascript
to
actually
you
know,
change
the
content
of
the
of
the
menu
itself.
So
since
we
have
basically
embedded
css
within
a
javascript
file
for
the
navigation,
so
I
just
thought
that,
okay,
for
now,
if
you
want
like
like
since
I
just
wanted
to
create
like
a
first
preview
in
case
this
is
not
looking
still
not
looking
good
enough.
K
Where
now
like,
at
least
you
don't
have,
that
kind
of
you
know
weird
effect
when
you're
changing
from
one
menu
to
another.
So
now
it's
only
like,
let's
say
only
when
you
hover
over
to
the
actual
name
like
let's
say
the
actual
menu
item
itself.
Only
then
it
will
open
the
navbar,
and
there
was
also
like
a
small
issue
with
respect
to
overlapping
of
the
menu
itself.
K
So
let's
say
if
earlier,
actually
what
was
happening
was
that
when
you
had
the
two
cards
like
the
cards
where
you're
showing,
I
thought
that,
at
least
for
the
mobile
view,
we
don't
really
need
to
show
it,
because
actually,
what
was
happening
with
that
it
was
overlapping.
K
So,
like
the
content,
it
was
actually
overflowing
from
the
card,
so
I
just
removed
it
for
now,
but
again
like
in
case
it
needs
to
be
restored
and
we
want
to
go
ahead
with
some
kind
of
other
design.
We
can
go
ahead
and
do
that
so
like
yeah.
K
These
were
like
the
update,
but
like
definitely
if
we
want
something
like
the
one
that
strappy
has,
but
in
that
case
we'll
be
using
javascript
to
actually
do
that,
since
it
would
be
like
a
single
div
element
and
it
would
render
the
contents
based
on
the
menu
item
that
we
have
right
now.
We
are
using
css
that
is
embedded
in
javascript.
A
Well:
okay,
to
do
three
points
of
feedback,
one,
the
mobile
view
from
my
from
my
perspective.
I
think
it's
fine
not
to
have
the
graphical.
You
know
the
other,
that
you
know,
there's
just
a
limited
lan.
You
know
space.
K
This
this
right,
this
was
like
one
of
the
bugs
that
you
were
mentioning,
but
actually
again.
I
think
that
since
the
I
would
say
since
okay,
the
hover
effect
that
actually
happens
now,
it's
now
only
limited
to
this
a
so
like,
basically,
the
href
or,
like
the
aaa
tag,
the
anchor
tag
earlier.
What
we,
what
we
were
doing
was
that
it
was
also
like
part
of
the
li
that
is
like,
since
these
are
all
part
of
the
right.
K
So
so
right
now
in
case
you
want
like,
I
could
probably
increase
the
padding
of
this
to
be
somewhere
like
around
here,
because
right
now,
it's
only
going
to
be
activating
the
hover
effect
only
on
when
you
are
like
navigating
on
top
of
the
icon,
like
on
top
of
the
nav
bar
element,.
A
J
A
The
that
flicker
do
it
again
if
you
would
like
to
go
back
and
forth
a
few
times
so.
K
Now
now
like
earlier,
what
was
happening
like
if
I
compare
it
with
the
current
version
of
the
website,
I'll
just
open
layer,
negative
I'll,
give
you
one.
Second,
it's
acting
weird.
You
know
this
one.
Oh
sorry,
okay,
so
currently
like
what
what
you're
seeing
is
that
in
case
you're
over
here,
you
see
that
weird
flicker
right
that
that
takes
place.
K
This
is
the
issue
that
was
there
earlier
so
now
in
case,
like
you
want
to
compare
with
it.
So
now
you
have
this
kind
of
smooth
animation
and
one
once
you
close
it
you
so
like.
That
is
the
difference
right
so
over
here
earlier,
when
you
were
doing
it,
it
was
like
it
was
causing
this
kind
of
a
weird
flicker
issue.
So
now
that
that's
gone.
A
So
we,
while
we
do
want
for
the
contents
of
the
menu
to
be
context
specific
to
you,
know
to
to
that
top
level.
Menu
item
yeah
when
you
just
take
your
go
across
the
go
across
the
row,
all
the
menu
items.
K
So
the
thing
I'm
trying
to
get
to
it
will
flicker
every
time
it
didn't
flicker
over
here
since
about
didn't.
Have
a
item
like
it
didn't
have
the
mega
menu?
Oh
okay,.
A
K
So
now
now
it's
like
now
in
case
you
close
it
only
when
you
hover
on
top
of
so
like.
That
is
why
I
just
wanted
to
probably
confirm
with
you
that
if
you
want
to
go
ahead
with
that
implementation
was
in
strappy,
they
are
having
javascript
written
in
which
they're
dynamically
just
changing
the
content
itself
like.
So
if
I
go
to
strappy.io
quickly,
strappy
dot
io.
K
So
if
I
quickly
just
go
over
there,
what
you
would
see
is
that
the
div
element
like
the
div
itself,
that
is
remaining
the
same,
so
okay,
I'm
just
training
for
it
to
load.
So
the
thing
that
you
see
over
here
is
that
this
div
element
itself
is
the
same,
but
it's
only
changing
the
content
dynamically
based
on
the
hovering
effect
for
us,
it
was
that
we
were
changing
like
it
was
creating
like
you.
K
A
K
A
A
I
will
shoot
myself,
probably
by
the
end
of
the
year,
if,
if
I
have
to
watch
that
thing
like
flicker
in
and
out
the
whole
time
as
I
and
then
the
second
reason
is
because
with
css
only
my
hunch
is
is
that
if
we
want
to
is
that
we
would
be
lacking
a
fair
bit
of
dynamicism
in
terms
of
what
the
freshness
of
each
of
the
menus
that
we
would
have
to
basically
do
a
release
of
the
the
site.
A
A
Yeah,
how
is
that
a
difficult
thing?
Did
you
expect
no.
K
No,
no,
I
I
don't
think
it's
difficult
so
but
yeah
like
the
only
thing
that
why
I
implemented
this
was
that
now,
at
least
you
don't
have
that
kind
of
a
flicker
effect
once
you
like,
you
know,
go
through
between
menus,
so
it's
like
you
would
see
like
at
least
it's
a
bit
more
probably
you
could
call
it
like
a
bit
more
smoother
in
terms
of
transitioning
that,
like
what
regular
effect
that
you
would
be
having
that
in
case,
you
lose
your
focus,
it
will
like
it
will
just
go
out.
K
So
at
least
now
like
this
is
like
the
version
that
I
put
like
put
the
pull
request
for
right
now,.
A
Nice,
okay
and-
and
I
realized
like
by
the
way
everyone
that's
on-
if
this
is
boring
or
you're
not
invested,
please
don't
hang
out,
it's
been
a
lovely
to
spend
the
friday
with
you
I'll
pause
to
let
every
to
let
people
jump
off,
because
it's
really
important
that,
like
people,
that
that
site
is
quite
important,
because
people
are
going
to
judge
much
of
the
work
that
people
do
here
by
its
cover-
and
you
know
by
that
book
by
that
that
and
and
so
being
anal
and
being
extraordinarily
meticulous
with
with
that.
A
So
so
I
have
one
more
thing,
then,
for
you
survival,
and
that
is
that
the
the
transition
effect
that
you
see
on
strappy
I
o,
is
something
of
like
a
roll
down.
It
looked
like
like,
whereas
the
one
that
we're
showing
yeah.
G
K
A
You
know,
there's
feels
there's
a
certain
elegance
and
beauty
to
it.
Can
you
go
to
stripe.io
to
human.
A
A
Normally,
cool,
okay,
I
really
yeah.
I
really
appreciate
you
working
on
this
shivaya.
It's
it's
just!
It's
really
important
when
you're
done
with
the
javascript
version,
your
instruction
for
everyone
about
the
way
in
which
the
god,
what
are
we
calling
those
those
highlights
highlights,
as
in
like
the
two
blog
posts.
K
A
Yeah,
when
you're
done,
then
yeah
you'll
you'll
be
in
a
position
to
educate
us
on
the
controls
that
we
have
there.
My
hope
is
that
those
controls
are
are
things
like?
Are
you
know
controlled
through
metadata
on
the
object?
Like
you
know,
the
mdx
like,
if
there's
an
mdx
file
and
the
front
matter,
it
has
a
certain
value
like
featured
or
not
that
it
would
be
of
a
certain
type.
It
would
be
like
a
type
project
featured
true
or
false,
or
whatever
the.
A
Okay,
we're
gonna
at
some
point,
give
a
pretty
hard
press
on
this
site
there's
and
as
a
matter
of
fact,
for
those
interested,
we
should
probably
commit
to
taking
a
look
at
the
google
analytics
that
have
been
the
traffic
that's
been
going
to
the
site
and
where
that's
been
coming
from
and
just
get
you
know,
I'm
gonna
be.
A
I
think
I
think
it
would
feel
good
to
to
each
of
you
to
see
who
it
is.
That's
checking
out
your
stuff,
oh
speaking,
of
checking
out
your
stuff,
maybe
I'll
just
I
will
try
to
get
an
email
out
or
something,
but
but
a
book
was
published
yesterday
and
it
talks
about
the
landscape.
A
It
talks
that
was
shown,
I'm
so
gerald
and
andrew
and
many
others
that
have
worked
on
it.
That
landscape
is
linked
in
that
book,
like
three
four
times
measuring
and
its
architecture
with
inclusion
of
the
measuring
operator
and
some
of
the
patterns
that
are
being
worked
on.
That
diagram
is
included
in
the
book
as
well,
and
so,
and
so
is
image
hub.
A
Imagehub
is
in
the
book,
and
so
anyway,
we'll
make
sure
that
we're
going
to
stuff
your
inboxes
with
a
pdf
copy
of
it
at
some
point
when
I
get
my
hands
on
it,
but
I
but
I
really
want.
I
hope
that
everyone
takes
it
as
a
point
of
pride
because,
because
there's
a
lot
of
people
who
are
going
to
be
looking
at
your
stuff,
so
yeah
the
name
of
the
book,
it's
the
name
of
the
book
is
long.
It
is.
A
Whoops,
it
is
the
link
I
just
sent
without
the
3os,
it's
the
second
edition
of
the
enterprise
path
to
service
mesh
architectures,
and
it's
the
second
edition
which,
if
you
have
an
o'reilly
subscription,
I
think
you
can
go
see
it
now.
If
you
don't,
I
think,
in
about
a
week,
I
would
expect
probably
timed
with
kubecon.
Maybe
nginx
will
be
hawking
it
any
which
way
guaranteed.
I
will
probably
send
everyone
a
link
to
it
19
times
over
the
next
few
months,
but.
L
Ali
that
link
doesn't
exist
like
yeah.
I
misspelled
books,
you'd
think
if
I
could,
if
I
could,
author.
A
A
There
are
links
on
that
page,
so
there's
an
open
source
contributor
that
created
the
the
second
edition
page
on
layer,
five
io
and
at
the
time,
and
even
now
I
don't
believe
that
there
is
a
link
to
get
a
copy
of
the
pdf.
I
don't
I
don't
have
one
myself
I
just
I
just
have
the
original
you
know.
So
I
think
this
will
take
you
to
like
the
old,
the
current
version,
but
but
yeah
so,
okay
yeah,
I
I
sort
of
maybe
spoke
about
this
slightly
prematurely.
I'm
just
excited
that.
A
A
A
Well,
hey,
I
please
do
anticipate
with
q
con
next
week.
We
might
end
up
canceling
one
or
two
of
the
meetings
we'll
try
to
do
that
today,
based
on
who
who's
gonna
be
available,
but
it's
just
a
it's
a
super
busy
week.
It's
always
busy
cool.
A
Thank
you
all
happy
diwali
see
you
guys.
Next
time
see
you
michael.