►
From YouTube: Meshery Development Meeting (August 12th, 2020)
Description
Welcome @Neilzblaze007!
A
A
A
A
B
A
Also
goes
by
neil
on
in
slack
which
I've
given
him
a
really
hard
time
about,
and
I
like
for
the
life
of
me,
I'm
not
sure
why
I
gave
him
a
hard
time.
It's.
I
guess
I
guess
that's
that's
just
what
I
do
cool
okay,
all
right,
great,
let's
get
rolling
and
we've
got.
I
think
a
number
of
things
to
chat
about.
I
turned
on
the
recordings.
What
we'll
post
our
meeting
like
we
always
do.
A
A
A
A
A
As
of
yesterday,
I'm
getting
a
double
l,
so
I
just
want.
I
want
to
go
on
record.
I
want
to
be
clear,
actually
see.
This
is
an
example
of
like
holiday,
okay,
okay,
good
now
that
everyone
knows
that
that
it's
not
me,
that's
misspelling!
It's
it's
my
keyboard,
but
can
we
maybe
talk
about
this?
This
first
item
here,
smi
conformance?
A
C
A
As
a
matter
of
fact,
like
this
conversation
is
probably
good
to
have
abhishek
on
as
well
in
part,
because
he's
been
focusing
on
adapters
and
so
as
mike
as
mikhail
goes
to
put,
you
know,
potentially
make
waves
on
the
console.
Adapter
it'll
be
good
to
have
a
shack
here
to
walk
through
so
wow
we've
got
the
entire
agenda
backwards.
A
C
Yes
sure
should
be
just
so
current.
Actually
I
don't
have
my
imagery
running,
so
I
will
just
give
it
a
walk
through
by
explaining
the
problem
with
it.
So
whenever
we
run
our
performance
test
in
measuring,
we
get
the
results
in
the
results.
Menu
sub
menu
in
the
nav
bar
on
the
left
hand
side.
So
the
results
get
populated
as
a
list
in
the
table
so
generally,
whenever
we
are
doing
any
performance
test
based
on
the
any
of
the
service
meshes.
C
The
issue
is
with
is
that
gen?
If
we
are
using
a
kubernetes
conflict
or
anything,
then
the
results
are
showing
appropriately
as
it
needs
to
but
based
on,
but
when
we
are
using
a
service
mesh
service
mesh
to
check
the
performance
test
and
show
the
results,
but
the
model
which
shows
the
performance
of
the
mesh
doesn't
gets
populated
with
the
data.
C
A
A
The
the
verbal
description
that
nikhil
just
gave.
You
can
kind
of
see
it
in
action
in
this
recording.
That's
that
as
a
user,
you
go
in
to
look
at
performance,
results
and
part
of
what
you
can
do
in
measuring.
Is
you
can
see
a
histogram
of
the
performance
metrics
or
the
the
statistical
analysis
of
the
the
results?
A
Then
you
can
also
see
that
that
horizontal
ellipsis
that
those
horizontal
three
dots
you
can
click
on
any
of
the
rows
to
get
a
little
bit
of
a
tabular
view
of
the
same
data.
It's
both
a
tabular
view
of
that
same
data,
and
it's
also
some
data
that
isn't
graphed
and
yeah
and
so
what's
being
shown
here.
A
Is
that,
like,
on
occasion,
I'll,
go
into
look
at
that
data
and
it
won't
display-
and
my
I
you
know
for
the
my
thinking,
yeah
and
there's
a
couple
of
different
scenarios
in
which
this
potentially
occurs
so
part
of
the
data
that
is
retrieved
during
a
performance
test
are
metrics
about
the
kubernetes
cluster,
the
number
of
nodes,
how
many
resources
they
have.
That
kind
of
a
thing
and
some
of
the
other
potential
data
is,
I
think,
dependent
upon
it
shouldn't
be
dependent
upon
whether
or
not
you
have
grafana
or
prometheus
connected.
A
C
The
guess
is
that
it's
because
of
the
control
in
the
code
that
we
have
for
the
service
mesh
section,
we
have
three
sections
for
the
in
general.
Our
results
model
have
will
have
three
part,
one
for
the
communities,
one
for
the
service
mesh
and
one
for
the
general
performance
system
to
which
we
learned.
So
the
service
mesh
part
of
the
table
is
facing
some
issue.
There
is
some
code
issue
with
that
part
of
the
code
so,
and
that
was
with
the.
C
This
is
a
thing
that,
when
we
pass
a
member
of
the
object
like
a
data
member
of
the
object,
so
when
in
react,
we
can't
render
the
object
directly,
we
will
have
to
pass
it
the
basic
prototype,
the
basic
data
types
that
are
available.
That
is
the
number
string
another.
So,
instead
of
passing
the
data
members,
the
object
itself
was
passed
to
the
function.
C
So
that's
the
result.
I
converted
past
use
the
object.entries
formation
inbuilt
form.
That's
inbuilt
function
to
convert
the
objects
to
a
pair
of
array
array
with
two
keys
and
object
values
in
which
we'll
read
the
values
and
pass
it
to
the
function.
Instead
of
us
passing
the
object
by
d
referencing
it
from
the
function,
so
that
there
is
no
issue
with
the
data
type
of
the
value
which
has
been
crossed.
A
Okay
and
then
okay
and
then,
as
I
was
verifying
the
the
fix,
I
seem
to
still
experience
the
behavior.
So
I'm
just
I've
just
checked
out
this.
This
branch,
this
pr
and
I'm
compiling
and
was
gonna,
take
a
look.
It's
like
it's
one
of
those
things
that
I
think
we
have
to
talk
through,
because
I
I
still
see
the
issue
it
manifests
even
with
this
change
and
which
yeah.
C
So
yeah,
I
the
issue
of
what
I
think
the
issue
is
with
this
function
is
particularly
because
it
calls
another
function
but
inside
that
part
randomized
info
and
that
the
same
function
is
being
called
by
other
functions
also.
But
if
you,
when
it
is
called
by
other
functions,
it
runs
appropriately
inside
as
we
wanted
to.
But
when
it's
called
from
this
function,
then
it
shows
that.
A
I've
got
I'm
not
running
any
adapters
because
because
I've
just
built
locally
and
I've
only
built
mesh
reservoir,
I
haven't
built
the
adapters
and
so
they're
not
in
the
environment,
but
if
I
go
over
to
some
results
and
in
the
recording
there's
a
quick,
quick,
wrk
2
test,
that
was
a
quick
wr
tape,
wrk2
test
again
that
I
was
kind
of
using
as
the
as
the
issues.
So
so,
if
I
come
here,
we
search
for
wrk2
all
right.
A
So
I
think
these
are
the
two
right
here,
so
wrk2
test
and
a
wrk2
test
again,
okay,
so
in
this
environment,
if
yeah-
I
don't
know-
I
don't
recall
quite
what
was
in
the
environment
at
the
time
and
that's
actually
that's
funny,
because
that's
actually
the
point
of
this
additional
data
is
what
you're
seeing
right
here
is
like.
This
is
some
additional
info
about
the
performance
analysis,
which
can
be
helpful
to
have
in
a
tabular
format.
A
A
I
mean
that's
because
that
data
wasn't
collected
at
the
time
and-
and
this
is
why
I'm
saying
like
I
think
that,
in
order
to
collect
that
data
we
might
have
had,
I
probably
would
have
had
to
have
had
prometheus
connected
and
I
might
not
have
connected
it.
So
that's
why
I'm
saying
like
well,
sometimes
the
bug
might
manifest
in
differently
in
different
situations.
C
B
A
A
So
the
results
ran
they're
here
you
can
download
those
see.
You
know
we're
able
to
download
some
of
that
some
of
the
data
not
all
of
it,
I'm
here
and
then
we're
able
to
go
back
to
results.
We've
got
our
tests.
We've
got
our
click
here.
This
shows
now.
I
think.
Okay,
that's
good,
while
the
other
test
that
we
looked
at
was
performed
using
wrk2.
A
A
C
C
C
I
won't
be
able
to
do
it
again,
but
it
worked
for
me
I'll
give
it
another
try,
and
I
guess
I
will
have
to
try
with
prometheus,
but
as
I
can
see
the
conversation
between
you
and
ayush,
I
guess
are,
you
sure,
is
working
on
it.
Previously.
He
tried
with
q
when
it.
Oh
sorry
if
it's
prometheus-
and
it
was
working
for
me-
I
don't
know-
means
if
anyone
else
also
has
prometheus
set
up
or
the
final
setup
in
his
system
can.
A
F
A
A
C
A
F
Yeah
yeah,
we
can
do
that
and
maybe
just
two
sentences
background
so
I
chose,
I
think
I
thought
that
was
a
good
place
to
start,
because
right
now,
I'm
actively
working
with
console,
exploring
it
in
a
poc
and
so
the
first.
When
I
tried
to
use
measuring
to
look
at
console,
I
noticed
that
version
was
outdated
and
there
is
an
issue
for
that.
So
I
thought
that
could
start
with
that
one
and
then
there's
other
issues
like
that.
F
It
should
be
able
to
improve
support
for
running
on
onenote,
which
is
typical
if
you
run
it
locally
and
then
I
saw
a
refactoring
issue.
You
know
that
be
using
helm
to
generate
the
config
that
is
deployed,
which
of
course,
is
also
very
good
idea,
and
then
it
became
a
tiny
bit
too
big
as
a
first
step
into
this.
But
then
I
thought,
then
you
gave
me
this
task
that
I
can
do
about
the
release
drafter.
F
F
If
that's
at
all
worth
doing,
and
then
I
wondered,
do
we
have
some
guidelines
how
the
default
configuration
of
the
service
meshes
that
are
included
in
the
adapters
should
be,
you
know,
should
it
be
like
production
ready
with
all
the
security
on?
Should
it
be
close
to
the
default,
usually
sort
of
like
tutorial,
like
configuration
given
by
the
vendor,
so
that
it's
sort
of
like
easy
to
get
started
with
it?
F
You
know
whether
you
have
tls
on
or
not
and
stuff
like
this,
it
probably
make
will
make
a
difference,
but
it
will
be
important
to
you
know
that
configuration
is
as
similar
as
possible
across
all
the
meshes,
if
that's
possible
at
all,
but
you
know
back
to
the
console
you
know.
Should
I
just
sort
of
like
you
know
not
really
worry
about
the
the
original
configuration,
but
just
create
a
one
note,
so
that
it
works
locally
and
sort
of
tutorials
that
you
can
start
exploring
and
playing
locally
without
having
to
worry
too
much
yeah.
F
A
It's
it
yeah
yeah
we're
well
yeah,
it's
a
good!
It's
a
great
highlight,
like
the
in
general
mesherie's
abilities.
There
are
woefully
underpowered
the
initial,
like
we
kind
of
accomplished
the
initial
goal
of
or
some
time
ago
we
accomplished
in
the
initial
goal
of
doing
what
you're
saying
around
providing
a
little
bit
of
a
playground.
A
You
can
kind
of
evaluate
performance
to
the
extent
that
you're
comparing
apples
to
apples
but
the
lack
of
controls,
but
but
that's
a
inju
like
as
we
so
not
that
we
should
discuss
much
about
the
refactoring
of
the
adapters
but
part
of
thinking
about
taking
on
empowering
the
user.
With
a
bit
of
choice
like
we
want
to
like
hey
the
the
deployment
model
or
even
before
that
is
really
like.
Hey,
did
you
want
to
deploy
the
latest
version
of
console
or
1.7
or
1.7?
A
A
That
that
does
that's,
that's
really
agnostic
to
any
tech
that
any
given
mesh
has
it's
just
identify
version
numbers
and
so
yeah,
but
but
yeah
to
get
to
right
side,
the
adapter
so
to
speak,
because
it's
kind
of
belly
up,
if
you
will,
which-
which
I
think
we
were.
I
was
mentioning
there's
a
few
hashicorp
folks
that
are
in
the
that
are
involved,
they're
a
partner
of
layer,
five
they're
there
and
I'm
here
in
the
slack
they're
very
much
so
engaged
they've
been.
A
I
can't
tell
you
how
many
meetings
that
we've
had
where
they're
like
hey.
So
you
know
when
when's
that
adapter
gonna
go
from
you
know:
code.
That's
like
almost
two
years
old
now,
like
1.5,
was
quite
so
long
so
yeah
I
mean
so
they're
they're
stoked
at
the
notion
that
that
someone
is
sniffing
around
the
adapter
that
yeah,
sorry,
so
sorry
so
kind
of
long
long
story.
I
think
you
hit
it
totally
on
the
head
that
a
hey,
hey,
hey!
A
We
there
is
a
there,
isn't
an
adapter
dock,
yeah,
okay,
but
it's
kind
of
looked
at
that
time.
But
it's
got
nothing
to
do
with,
like
the
actual
things
that
you
need
here
and
now,
which
is,
as
you
know,
what's
the
current
structure
of
the
code
as
a
developer.
How
do
I
you
know?
What
am
I
so
we
we
need,
some
of
that
I'm
we're
right
in
the
crosshairs
of
like.
A
Generally,
it's
a
natural
thing
and
it
that
goal
very
well
aligned
with
the
goal
of
providing
interoperability
with
a
bunch
of
meshes
very
well
aligned
with
measuring
the
approach-
or,
I
would
say
meshrey's
goal-
is
to
in
a
first-class
way
ensure
that
each
of
the
meshes
are
able
to
expose
their
proprietary
value.
Our
proprietaries
may
be
the
wrong,
but
their
special
sauce
value.
Over
and
above,
what's
you
know
the
watermark,
if
you
will
of
common
functionality?
A
Hence
the
fact.
Hence
why
we
don't
have
one
adapter.
We
have
an
adapter
for
each
one.
Now
that
said,
that
doesn't
mean
that
meshery
shouldn't
be
either
taking
advantage
of
sm.
Well
I
mean
it
does
I
mean,
but
taking
more
advantage
of
smi,
literally
for
some
of
the
below
the
watermark
things
so
yeah
anyway,
design
just
says
sort
of
talking
loud
about
a
design
principle
of
so
the
met.
The
istio
adapter
is
the
most
powerful
adapter.
Mastery
has
two
reasons:
one.
A
It
was
the
first
one,
and
so
it's
been
around
the
longest
two
is
the
most.
You
know
the
more
powerful
mesh,
or
at
least
originally
anyway,
and
we've
one
of
the
questions
and
michael,
is
you
go
and
get
familiar
with
this
like
a
good
way
like
a
way
of
making
a
mark
on
measuring
as
a
project
is
to
think
about.
A
A
Yeah,
this
is
probably
fine,
but
the
point
here
is
some
adapters
are
categorized
as
being
an
of
an
alpha
status,
a
beta
status
or
a
stable
status
and
early
on
in
the
project
for
lack
of
having
any.
You
know
much
engineering
rigor
around
this.
That's
you
know
it's
a
simple
way
to
just
sort
of
quickly
categorize
them
and
set
set
a
general
expectation
about
what
you
would
expect.
You
know
a
user
would
expect,
but
michaela
those
are
soft
lines
between
stable
beta
and
alpha
and
our
potential
things.
A
A
I
think
it
will
be
helpful
for
me
to
say
these
next
couple
things
and
that
is
of
the
so
you're
familiar
with
the
the
you
know,
this
high
level
architecture
you've
got
a
great
understanding
of
that
one
of
the
concepts,
one
of
the
things
that
there's
two
things
about
adapters
that
we
we
generally
try
to
do.
One
is
treat
them
as
a
stateless
thing
and
I
think
there's
a
couple
of
statements
to
be
made
around
the
notion
that
it's
stateless
and
that
is
that
any
give.
A
There's
another
component
being
worked
on
in
measuring
called
well,
it's
called
the
measuring
operator
and
exactly
yeah
it'll,
be
hopefully
you
know,
introspecting
the
kubernetes
environment,
identifying
you
know
discovering
service
meshes
and
their
config
and
as
it
itemizes
and
sort
of
indexes
you
know
what's
what's
going
on
inside
of
kubernetes
environment,
it
will
be
eventing,
it
will
be
sending
over
that
data.
Ideally,
it
will
be
sending
that
that
information
back
to
that
the
type
of
adapter.
A
So
if,
if
mesh
sync
discovers
that
there's
a
constant
console
in
the
environment,
it'll
be
gathering
up
info,
ideally
sending
it
back
through
the
console
adapter,
because
the
adapters,
to
the
extent
that
we,
that
we
can,
we
try
to
make
sure
that
they
have
only
mesh
specific
info
or
mesh
specific
capabilities.
Like
anything,
that's
a
general
capability
that
we
would
want
to
have
probably
an
imagery
server,
because
that's
just
you
know,
we
don't
want
to
repeat
ourselves
acro,
but
the
the
the
unique
considerations
of
a
given
mesh.
Ideally
that's
what
goes
into
the
adapters.
A
A
So
yeah,
I
don't
know
how
okay
so
then
here's
one
other
thing
about
them.
It
kind
of
part
of
their
philosophy
of
part
of
how
they
work
is
that
yeah
again
they
don't
really
do
anything
until
they're
acted
upon
one
of
the
ways
and
they
are
acted
upon
every
single
time
that
right
now.
I
believe
this
is
a
true
statement
like
it
like
every
single
time.
A
A
A
A
Most
likely
this
won't
well.
Actually.
Maybe
this
will
work
okay,
so
the
reason
that
this
so
that's
fine.
The
reason
that
this
worked
is
because,
in
my
if
this
didn't
work,
I
I
here,
let
me
let
me
get
myself
all
confused.
How
did
this
work?
Because
I
went
into
my
resolve.conf,
my
my
local
dns,
config
and
added.
A
A
A
A
All
right,
so
so,
actually
I
am
connected
to
our
adapter.
Now
and
now
I
can
demonstrate
the
fact
that
like
well
here,
I
guess
I
can't
really
do
it
quickly,
but
but
basically
ev
every
time
that
that
this
page
loads,
actually
you
can
see
first
see
I'm
clicking
refresh
and,
and
you
can
see
for
a
split
second
or
longer
than
you
should
see
that
what
is
initially
being
shown
is.
It
says:
hey
configure
settings
which
is
yeah
the
meshery
server
code
saying.
Well,
I
don't
you
know
I.
A
A
Page
a
couple
of
interesting
things
going
on
one:
you
know
it
basically
what
it
does.
What
I
was
saying,
which
is
basically
goes
over
retrieves
a
list.
It
retrieves
the
name
of
the
adapter,
it
retrieves
a
list
of
supported
operations,
those
operations
in
this
case
they
fall
into
three
categories,
but
you
know
which
is
like
you
know
installing
it
deleting
it
installing
a
couple
of
different
sample
apps,
and
so
you
will.
We
will
see
in
the
adapter
those
like
it's
kind
of
it's
kind
of.
A
I
guess
what
I'm
we're
trying
to
get
to
is
like
hey,
like
the
adapters
are
going
to
have
their
each
of
the
they're
each
going
to
have
their
own
capabilities.
Mesh
reserver
is
ignorant
of
what
those
are
and
so
it'll
just
ask
sort
of
real
time.
Hey
you
tell
me
what
you're,
capable
of
and
I'll
just
display
that
I'll
format
that
in
a
pretty
way
well
in
a
semi
pretty
way
to
the
user
and
then
I'll.
Let
them
invoke
whatever
operations
you're
telling
me
you're,
capable
of
so
yeah.
A
That
was
half
of
that
was
most
of
my
point.
I
think
the
second
part
of
this.
That's
interesting
is
like
well,
okay,
hey
also,
I
am
connected.
Mastery
server
is
connected
to
a
kubernetes
environment
and
you
can
see
it's
interrogating
that
interrogation,
it's
retrieving
a
bit
of
info
environment
info
and
and
nikhil
the
bug
that
you
were
working
on.
It's
this
kind
of
environment
info
that
I
was
that
that
we're
expecting
to
see.
G
So
we
I
have
a
suggestion
here,
yeah.
Well,
I'm
more
of
a
wondering.
So
if
we
are
planning
on
making
the
measuring
backend
server
to
be
a
stable
application,
then
there
is
a
little
bit
of
trade-off
here
as
of
maintaining
states
or
when
you'll
have
to
query
every
time.
You
refresh
that's
what
I
mean
and
if
not
in
the
other
case
we
can,
we
can
have
a
local
caching
or
some
kind
of
statefulness
for
mystery
back-end
to
maintain
this
or
to
sort
of
avoid
this
issue.
A
Yes,
yeah,
that's
good
this!
Let
me
let
me
pull
this
back
up,
which
is,
which
is
to
say
that
abhishek
you're
right
that
there's
a
bit
of
state
caching
that
goes
on
in
mesh
reservoir
and
that's
actually
cached
it
in
a
hidden
folder
under
the
user's
home
directory
that
that
function.
That
does
persist
and
kind
of
cache.
Some
things
today,
one
of
the
things
that
it
doesn't
cache
is
the
connection.
You
know
the
supported
operations
that
are
provided
by
each
adapter
right.
A
A
Adapter,
it's
been,
it's
been
I'm
sure
tiresome
for
hashicorp,
because
you
know
they
have
potential
customers
taking
measuring
and
trying
to
evaluate,
should
they
run,
console
and
and
they're
getting
a
bad
console
experience
or
an
old
console
experience
and
so
they're
excited
to
see
it
potentially
bump
up
in
version
number
it
is
yeah.
I
think
you
know
what
you
you'd
say
so
yeah.
A
It
really
doesn't
matter
what
the
okay
I'll
say
it
like
this,
when
we
were
writing
out
the
istio
adapter
istio,
you
know
sort
of
pop
prominently
would
display
here
if
you
want
to
install
istio
and
you're
familiar
with
this
michelle.
But
if
you
want
to
display
I'm
sorry
if
you
want
to
run
istio
in
a
demo
deployment,
here's
here's
a
set
of
configs.
If
you
want
to
run
it
in
a
minimal
deployment,
here's
a
set
of
config
you
want
to
run
it
securely
like
and
and
gen
in
general.
A
A
The
project
itself
has
had
bitten
off
attempting
to
do
that,
and
I
don't
and
as
a
project
I
would
say
we're
not
in
a
position
we're
not
in
a.
We
don't
have
enough
people
here
to
do
that.
Well,
and
so
we've
been
focusing
on
this
this
bit
about
provisioning
a
mesh
to
be.
You
know,
really,
you
know
fourth
direct,
I
would
say
well,
you
know
who
you
know.
A
I'd
say
something
to
the
effect
of
you
know
who
who
gives
a
rip
like
like
there's
no
specialness
about
that
like
I
don't
need
to
use
measuring
to
do
that
necessarily
so
so
why
am
I
going
to
use
the
tool?
Well
you're,
going
to
you're
going
to
use
a
tool
and
we
built
it
initially
after
some
just
some
some
first
use
cases,
but
we
very
quickly
want
to
get
it.
It's
like
people
ask
this
question,
which
one
should
I
use?
Well
hey.
A
This
is
a
quick
tool
to
provision
that
supports
more
than
any
other
tool
does
so
you
can
quickly
get
familiarity.
Here's
some
demo
and
some
sample
apps
to
do
that.
Some
things,
some
reasons
why
you're
going
to
come
back
to
the
tool
and
come
back
to
the
tool
is
for
performance
management
is
one
that's
a
very
helpful
thing
that
not
only
is
an
initial
adoption
question
but
ongoing
is
like
you
need
to
manage
that
performance,
make
sure
and
then
the
further
we
get
along
it'll.
A
A
A
A
A
A
G
G
A
A
G
E
Yeah,
so
I
just
wanted
to
point
out
that
I
still
have
issues
with
running
gtl
on
my
system,
so
I
wouldn't
be
for
the
meantime.
I
wouldn't
be
attending
to
any
detail
based
issue,
so
the
current
issue,
with
the
copy
to
clipboard-
I
don't
know
if
someone
can
take
it
off
yeah.
So,
oh,
so
I
don't
stay
so
long
on
it.
A
D
E
A
E
A
H
A
H
A
H
H
I
don't
know
what
I
don't
think
it
was
so
that
they
are
doing
we're
really
happy
that
I
think
the
kubecon
or
linux
event
they're
very,
very,
very,
very
expensive
and
people
have
very
criticized
about
that.
They
wanted
to
be
at
the
free,
but
I'm
really
happy
that
even
they
did
this
for
community
members
as
well
like
the
organizers
for
us
as.
A
C
If
the
protocol
is
not
present
in
the
url
right
after
submitting
the
performance
session,
it
will
on
the
so
after
they're
doing
this
one.
So
as
we
discussed
on
that
call
on
doing
it
in
the
client
side
means
on
the
chain
doing
it
while
changing
it.
So
I
just
went
through
this
component
that
they
have
a
textual
component
and
they
provide
a
prop
known
as
head
context,
which
checks
the
value
of
the
error.
Prop
it's
a
boolean
value.
C
C
So
yeah
all
right-
I
was
thinking
of
implementing
another
part
that,
as
we
are
having
this
issue,
that
the
performance
test
fails.
If
we
don't
provide
the
protocol,
so
we
will
be,
I
will.
I
was
thinking
that
we
should
disable
the
run
test
button
until
unless
all
the
errors
are
gone,
if
all
the
errors
are
gone,
then
only
we
should
enable
the
run
test
button.
Otherwise
it
would
be
a
bad
experience
if
the
user
doesn't
knows
submits
the
test,
and
then
situation
sees
a
error
over
here.
C
A
In
general,
this
is
an
improvement
over
over
the
you.
What
you're
doing
is
bringing
each
of
the
form
validations
down
to
the
individual
input
fields,
which
is
a
better
experience
than
having
a
form,
a
single
form,
level
validation
that
only
happens
on,
submit
right,
you're,
making
it
more
granular
you're
bringing
validation
on
a
per
field
basis.
C
F
A
C
A
C
A
C
A
C
H
A
All
right,
fair
enough,
we're
like
15
minutes
over
and
so
let's
I
want
to
you
know,
give
make
sure
we
don't
have
a
marathon
meeting
nice
good
michaela,
I'm
hopeful
that
that
was
somewhat
helpful.
I
my
for
my
part,
I
wrapped
up
final
draft
of
a
book
on
monday
night,
and
so
I
for
my
part,
I'm
much
more
free
to
be
more
helpful
than
I
have
been
so
as
you
go.
You
know
you
know,
raise
questions
as
you
go
because.