►
From YouTube: WebAssembly Filters Meeting (Aug 30th, 2021)
Description
WebAssembly Filters Meeting - Aug 30th, 2021
Join the community at https://layer5.io/community
Find Layer5 on:
GitHub: https://github.com/layer5io
Twitter: https://twitter.com/layer5
LinkedIn: https://www.linkedin.com/company/layer5
Docker Hub: https://hub.docker.com/u/layer5/
A
All
right:
hey
thanks
everybody
for
coming!
It's
the
web
assembly
filters
meeting
for
august
30th
monday
a
heck
of
a
meeting
to
kick
off
the
week.
So
there's
a
number
of
you
that
are
on
the
phone.
There
are
most
of
you,
many
of
you
represented
in
the
meeting
minutes.
If
you
don't
see
your
name,
please
drop
it
in.
A
And
we've
got
a
couple
items
that
are
queued
up
from
last
week,
so
it's
nice
we
last
week
we
had
unbelievably,
we
had
the
foresight
to
create
what
would
be
this
week's
agenda.
That
seems
like
a
wise
thing
to
do
so,
we'll
put
a
next
week
down
here,
a
couple
of
things
and
by
the
way
now
is
the
time
to
enter
in
a
couple
of
your
topics
if
they
aren't
represented,
so
so
drop
them
in,
assuming
that
they
are,
though,
we'll
we'll
get
started
in
no
particular
order.
B
So
I
believe
you
guys
can
see
screen
recording
that
I'm
sharing.
So
this
is
with
regards
to
editing
configurations
on
mesh
map,
be
it
a
workload
or
any
filter.
That's
dropped
on
a
workload
so
currently,
on
my
system,
I
have
is
the
adapter
running,
so
I
can
see
all
these
two
components.
So
if
I
open
I,
I
would
be
able
to
see
all
of
these
two
components
that
are
given
by
the
hd
adapter.
B
Now
I
drop
an
xtr
mesh
and,
as
you
can
see,
I
get
a
tooltip
which
asks
for
some
settings
with
which
are
which
would
be
applied
to
the
workload
that's
dropped
over
here
and
now
after
I
close
this,
these
settings
are
basically
saved
in
mesh
map
and
you
can
any
any
time
recall
these
settings
and
if
you
ever
want
to
maybe
edit
maybe
change
something
you
can
always
just
hold
on
to
the
node
and
then
you'll
see
the
context
menu.
B
We
we
call
it
context,
menu
this
thing
appearing
and
if
you
just
go
over
there
and
leave
your
mouse,
you
would
see
the
settings
menu
open
again
and
you
can
just
edit
anything
that
you
want
and
save
it
again.
That's
the
first
part.
The
second
part
would
be
yeah,
so
now
I
would
be
searching
for
on
a
filter.
I
would
be
dropping
this
onboard
filter
onto
the
istio
mesh
and,
as
you
can
see,
another
pop-up
opened
for
the
onboard
filter
and
on
the
filter
has
its
own
set
of
settings
that
you
can
basically
specify.
B
So
once
I
add
a
few
fields
as
you
can
see
and
close
the
whole
button
now,
if
I
want
to
edit
any
of
those
settings,
what
I
can
do
is
I
can
just
click
on
the
filter
that
I
just
dropped
and
the
settings
menu
will
again
open
and
then
you
can
edit
any
of
those
fields
again
like.
I
would
be
showing
you
and
yeah
the
same
thing.
If
you
close
it,
they
are
safe.
Again,
that's
the
update
for
mind.
A
There's
you
know
some
some
questions
to
get
the
discussion
going.
Is
this
intuitive.
A
B
D
B
A
I
think
I'm
I
good,
then
I
am
intending
for
that
to
be
a
soft
suggestion.
As
for
food
for
thought,
like
yeah,
I,
and
then
that's
I'll,
just
speak
for
myself,
and
that
is
that,
like
I
do,
I
enjoy
the
subtle
nature
of
something
like
a
long
click
like
that
that
the
interaction
can
be
so
sophisticated
that
it
takes
into
account
well
like
things
that
aren't
weren't
conventionally
true
of
older
computers
or
older
style
computing.
Was
that
like
it,
couldn't
tell
the
difference
between
holding
something
down
versus,
or
that
generally
wasn't.
A
The
case
is
that
that's
kind
of
that's
nice?
It
also
does
mean
that
it's.
I
think
this
is
a
fact
or
like
would
be
true,
that
it
would
be
less
discoverable
or
you
know,
fewer
people
would
happen
upon
it,
because,
because
it's
just
not
as
common
of
a
style
of
interaction,
and
so
it
would
it
be
bad
if
that
was
available
through
right,
clicking
and
through
long
clicking.
A
Now,
I
think
part
of
the
thing
to
maybe
state,
as
we
think
about
that
is,
is
that
is
whether
or
not
one
of
the
goals
is
to
have
this
mesh
map
available
through
mobile
into
the,
because
that
has
an
effect
on
how
people
you
know
they're,
not
using
a
mouse
or
using
your
finger.
So
then,
actually
it
is
much
more
common
to
have
a
long
click
on
a
mobile
interface.
E
B
Yeah
like,
if
you
take
a
look
at
like
we
haven't,
exploited
double
cheap
by
the
way,
but
in
my
opinion,
it's
better
if
we
have
options
and
right
click,
like
that's
general
convention,
if
you're
right,
you
see
options
but
yeah.
Double
click
also
works.
E
Oh
another
small
ui
thing
more
of
us
think
so
once
we
open
the
edit
option
like
we
are
just
getting
a
close
option
and
there's
another
option:
yeah
delete
and
close
right
so
yeah,
instead
of
so
it's
getting
automatically
saved.
B
Like
to
imply
that
it's
being
saved
yeah,
actually
the
intention
is
that,
like
what
we
want,
the
users
to
think
is
that
you
don't
need
to
go
over
here
to
save,
but
that
stuff,
like
that
feature,
is
not
yet
implemented.
So
in
future,
like
like
probably
next
week,
because
this
this
thing
that
you
just
suggested
is
actually
being
tracked,
and
probably
next
week
or
by
friday.
What
we'll
have
is
once
you
click
anywhere
outside
of
this
tooltip.
E
E
Discard
all
discards,
the
changes
that
are
made,
so
maybe
user
is
doing.
He
made
some
couple
of
changes,
but
he
feels
like
maybe
let's
keep
it
how
it
was
before
then
that
kind
of
thing
will
be
handy
right.
So
if
he
clicks
outside
then
it'll
get
discarded,
but
if
he
actually
wants
it
wants
to
save
it.
He'll
just
click
on
the
save
button.
B
E
B
That's
that
is
one
part
of
the
bigger
question
that
we
haven't
answered
yet
is
versions
of
having
patterns
itself.
Probably,
for
example,
if
you
take
a
look
at
google
docs,
google
docs
has
versions
on
every
edit
that
you
make
on
it
right.
So
that
is
probably
something
on
the
lines
of
that
thing.
But
yes,
it
would
be
good
if
we
can
have
something
like
that
right
now,
too,.
B
A
Another
another
item
of
feedback
to
consider
is
well
at
one
point:
when
we
were
when
the
tool
tip
was
up.
There
was
a
short
description
of
the
component
itself
and
which
is
nice
because
you
know
because,
because
that
that
type
of
description
can
be
helpful,
that
people
understand
what's
going
on
are.
Is
there?
Is
there
a
backlog
item
for
dealing
with
the
description.
B
A
Yeah
something
to
this,
I
don't
know
that
this
is
the
best
suggestion.
But
something
to
consider
is
that,
in
the
title
of
the
tooltip
on
voice
filter
we
got
to.
We
can
close
this
out.
We
can
delete
the
remove
the
component
if
there
was
a
small,
a
question
mark
with
a
small
circle
or
like
an
info
button,
yeah
yeah
that
or
that
you
went
on
upon
hover
or
something
that
would
then
show
the
description
that
might
be
a
way
of
martian
yeah.
B
A
Yeah
yeah,
I
don't
know,
and
that
might
be
a
great
you
know.
I
think
that
that's
an
improvement
upon
what
we
have
currently
it
might
not
be.
A
It
might
be
that
we
end
up
arriving
at
a
different
place
and
so
part
of
megan
has
questioned
about
like
well
hey
what
happens
if
you
or
do
we
have
control
over
double
click
and,
and
you
know
how
we
should
that-
do
something.
It's
like
yeah.
It's
a
it's
a
good
question.
Yes,
we
do
it's
like
okay.
Well,
actually,
so
we've
got
long
clicks,
right,
clicks,
single
clicks,
double
clicks
and
we've
got
click
and
drag
and
then
and
then
we've
got
right.
A
A
To
invoke
edit
like
invoke
one
of
those
specific
context,
menu
actions
that
you
would
otherwise
take
two
clicks
to
get
to,
and
so
it
yeah
just
just
I'm
just
think-
I'm
not
suggesting
it
or
just
sort
of
thinking
having
a
thought
process
out
loud.
E
A
Yeah
right
yeah
because
yeah
when
you
think
about
it,
yeah
it's
like.
Sometimes
you
use
a
canvas
like
this
or
an
application
like
this,
and
if
it's
really
about
navigating
and
viewing
and
sort
of
looking
at
the
shape
and
size
of
things,
you
sometimes
have
used
some
programs
where
you
would
double
click,
and
it
would
interpret
that
as
like
a
as
a
jump
to
zoom
like
instead
of
like
smoothly
scrolling
and
zooming
in
it
was
more
like
no.
No.
A
A
This
is
a
designer
and
it's
it's
more
about
characterizing
objects
and
interconnecting
them
and
describing
them
and
configuring
them,
and
so
like
a
double
click
to
auto
zoom
doesn't
make
sense,
but
but
to
megan's
point
like
what
is
the
core
activity
that
someone
is
doing
core
activity
is
configured
like
specifying
configuration
of
you
know
one
or
more
elements,
and
so
like
okay.
Well,
what's
the
activity
that
specifies
configuration?
Well,
it's
it's
the
edit.
I
think,
and
so
yeah
yeah
good.
This
is
good.
These
I
like
the.
A
So
all
the
feedback
that
your
pen
can
handle
this
work
today,
that's
good!
B
A
Yeah,
all
right,
that's
even
worse!
It's
like
it's
like
ship
time,
anyway,
good
cool,
any
other
feedback
for
someone.
B
F
Yep,
so
I
have
already
filled
some
details.
Let
me
share
my
screen.
F
So
I
actually
suppose
that
we
should
be
saving
more
more
metadata
about
wars
and
filters.
So,
regarding
the
same,
I
actually
created
an
issue.
It
was
issue
number
3900
in
there.
I
was
supposed
to
propose
that
okay,
I
asked
to
show
more
metadata,
but
what
that
metadata
should
look
like
so
the
I
have
I,
I
actually
created
an
issue
about
a
week
or
two
ago,
but
there
were
not
much
details
that
were
added
to
it.
F
Now
I
have
actually
added
quite
some
details,
though
that's
what
I'd
be
sharing,
if
I
can
just
find
it
yeah,
so
I
am
proposing
here
to
store
more
metadata
associated
with
wars
and
filters
and
the
material
that
I'm
requesting
to
be
added
when
we
store,
basically
when
we,
when
versus
these
walls
of
filters
on
with
providers,
is
the
first
thing
is
that
the
workload
selection
metadata
so
the
reason
I'm
proposing
to
save
these
is.
Let's
say
we
want
to
do
something
like
this.
F
That
is
all
what
you
have
to
do
is
just
add
this
particular
label
to
your
as
a
user
or
maybe
as
application.
You
just
have
to
add
this
particular
thing
and
because
this
was
associated
with
was
a
filter.
Everything
that
had
this
table
is
has
basically
the
internet
filter
enabled
on
them.
So
this
is
one
of
the
things
one
of
the
things
that
I'm
proposing
to
see,
that
is
workload
selector.
It
shouldn't
be
like
it's
not
that
I'm
proposing
to
do
it
just
this
way.
F
That
is
pretty
much
outfitted
out
machine
I
o
or
something
it
could
be,
just
up
cool
and
something
here
name
of
the
application.
This
is
something
I'm
proposing
to
say.
The
second
is
actually
operation
the
type
of
operation,
although
I'm
referring
to
I'm
referring
here,
a
url
from
basically
istio's
configuration,
but
I'm
also
proposing
that,
although
this
particular
thing
is
being
so,
if
you
would
just
go
to
this
particular
endpoint,
you
would
see
that
there
are
certain
operations
that
actually
istio
supports.
F
Istio's
formula
filter
configuration
supports
that
doesn't
mean
that
every
service
mesh
whose
data
plane
is
on
the
filter,
is
going
to
support
the
same.
But,
having
said
that,
because
we
as
measuring
will
be
developing
a
constant,
a
super
constant
call
on
the
filter
and
we'll
be
at
the
end
implementing
that,
so
we
can
try
to
bridge
those
gaps
which
basically
bridge
those
inconsistencies
that
service
meshes
have
in
their
basically
onward
filter
implementation.
F
So
these
are
the
these
are
the
these
are
also
the
things
that
I'm
basically
suggesting
to
persist.
The
reason
to
persist
them
is,
for
example,
we
can
take
example
of
rate
limit
filter.
We
know
that
where
limit
filter
should
be
added,
we
can
just
basically
we
can
just
drop
it
at
the
end
and
it
doesn't
matter
because
at
the
end
it's
just
straight
limiting
filter.
We
don't
have
to.
F
Basically
we
don't
have
to
be
first
or
we
don't
have
to
be
in
middle
of
some
other
filter.
We
can
be
at
any
position,
so
we
can
specify
r
and
as
a
user,
if
you
are
using
this
video
filter,
you
don't
have
to
actually
specify
that
explicitly
because
that
metadata
was
already
associated
with
the
filter.
This
is
the
second
thing
that
I
was
proposing
to
include
when
we
persist,
and
the
third
is
this.
This
particular
object
and
that
is
applied
to
enlarge
applied,
is
basically
just
a
water.
F
What
particular
section
of
on
where
you
would
want
to
attach
it
to
so
so,
let's
say
your
filter
is
not
http
filter,
so
you
don't
want
to
go
to
http
manager,
and
you
don't
want
to
go
to
that
chain,
so
you
can
basically
specify
that
okay,
I
am
interested
in
this
particular
section
and
not
not
something
else
so,
and
this
is
also
the
reason
to
propose
this
is
because,
as
a
as
official
the
official
version
filter
that
would
be
coming,
they
will
have,
they
can
have
these
annotations
or
basically,
they
can
have
these
metadata
already
in
place.
F
So
as
a
user,
you
should
be
able
to
just
apply
it,
because
we
know
that
what
what
the
best
place
for
a
particular
filter
was
so
the
things
if
someone
want
to
look
at
what
what
these
are.
This
is
basically
this
and
again,
as
I
was
mentioning
that
something
could
be
that
it
could
be,
it
might
be
just
being
implemented
by
istio
as
service
mesh
and
may
not
be
available
in
other
service
machines,
which
are
which
are
using
onwards
directly.
In
that
case,
we
will
try
to
bridge
the
gap.
F
Actually,
I
checked
multiple
service
meshes
and
how
what
configuration
they
actually
passed
to
the
password
that
they
took
there
are
inconsistencies,
so
we
will
definitely
have
to
push
that
down
on
sto
is
like
istio
exposes
the.
Basically,
it
exposes
a
lot
of
configuration
for
conversion.
That
is
the
reason
we
are
referring
to
istio's
configuration
and
the
third
object
is
just
it's:
it's
part
of
the
same
object,
and
that
is
match,
and
this
is
more
towards
that
how
conflict
should
look
like.
I
can
actually
point
to
this.
Also.
F
So
yep
this
is
so.
The
thing
is
that
it
could
be
that
we
as
we
as
a
machine.
We
know
that
official
version
filter
doesn't
actually
works
on
x
version
of
onward
filter,
so
you
as
a
user.
Probably
if
you
are
just
clicking
on
the
ui,
you
don't
probably
know
if
this
was
if
it
is
going
to
work
with
certain
on
the
filter
or
not.
So
we,
if
we
attach
that
editor,
a
quality
with
version
filters
and
other
material
also,
then,
basically,
we
can
have
checks
on
the
server
side.
F
That
okay,
is
this
condition
being
met,
that
is
on
the
filter
version
1.9.
Is
it
greater
than
1.9.1
or
something
because
this
particular
wasn't
file
version
filter
can
be
only
applied
to
that
particular
version.
So
these
are
also
other
messages
that
I'm
proposing
to
save.
Therefore,
there
are,
there
can
be
multiple
other
things
that
we
can
also
say,
but
that
would
be
sort
of
that
will
either
get
too
steep
specific
or
it
would
be
way
harder
for
us
to
bridge
the
gap.
So
there
are
things
that
we
can
point
to
the
cluster
cluster
info.
F
Also,
that
is
this
is
how
the
configuration
should
be
dropped
in,
but
those
bridges
are
too
far
like
it
would
be
extremely
hard
to
fill
those
kind
of
bridges.
So
these
are
the
things
that
we
can't
persist
while,
while
trying
our
best
to
bridge
gaps,
if
if
there
is
inconsistent
consistencies
between
service
switches,
at
the
same
time,
it
will
also
allow
a
better
user
experience,
because
a
washer
filter
is
already
aware.
Basically,
it's
coming
with
all
of
this
associated
metadata.
F
That
will
help
mystery
server
to
exactly
pinpoint
that
how
to
deal
with
this
thing,
where
to
put
it,
how
to
put
it
inward
in
the
basically
in
the
inbound
network
or
outbound
network,
and
what
are
the
workloads
that
you
are
interested
in
so
yeah.
This
is
the
overall
material
that
I
was
actually
proposing
to
store
other
than
obviously
wasn't
bindery
and
filtering.
B
F
In
terms
of
yeah,
although
this
workload
is
more
of
it's,
it's
not
the
workload,
it's
not
the
workloads
that
we
have
it's
more
of
just
it.
It's
just
a
name
that
I
chose
worker,
etcetera.
So
basically,
yes,
it
was
sort
of
yeah
to
your
point.
If
you're
talking
about
that,
if
you
drop,
you
know
on
the
filter
to
a
certain
application,
so
in
that
case
yes
or
what
what
we?
B
Yeah
like,
for
example,
right
now
to
persist
filters
like
mesh
map
is
persisting
filters.
I
don't
remember
showing
in
print
on
demo,
but
if
you
load
another,
if
I
had
saved-
or
probably
I
didn't
show
it,
but
if
you
save
the
file
that
I
just
showed
in
the
demo
and
if
you
load
it
as
a
patent
file,
the
filter
would
be
persisted,
but
right
now
what
I'm
saving
is
the
just
the
id.
B
Actually,
as
the
label
like,
I
am
probably
storing
an
array
of
labels
or
probably
just
a
single
label,
so
we
would
need
to
change
that
to
storing
the
label
instead
of
the
id.
F
No,
so
I
id
so
so
applications
that
are
being
deployed
with
mission,
so
consequently
with
mishma
as
application,
they
always
get
ideas
labeled.
So
oh
yeah.
A
Yeah
well,
this
is
the
end.
It's
a
little
bit
of
a
tough
one
where
to
figure
out
how
to
figure
out
where
to
draw
the
line
between
measuries
reach
and
the
control
planes
reach.
A
Primarily,
the
difficulty
here
lie
in
the
fact
that
it's
really
only
istio
that
has
well-
or
I
guess
I
should
ask
this
rude
crash.
Has
you
looked
at
some
of
the
other
implementations?
Did
you
see
I
mean
like
istio?
Is
the
only
one
that
comes
to
mind
that
prominently
supports
envoy
filters.
A
And
okay,
and
so
as
such
yeah,
so
we
we
need
to
step
gingerly.
I
would
say
because
well
because
what
we
don't
want
to
do
is
collide
with
the
control
plane
capabilities
that
those
control
planes
are
inevitably
going
to
come.
A
You
know
add
this
particular
functionality
is
you
know
we
want
to
leverage
what
they
have
and
not
not
reinvent
it,
and
that
that's
in
no
way
a
comment
toward
utkarsh
and
your
design
you're
trying
to
fill
in
the
gap,
and
I
think
what
we're
recognizing
is
like
for
most
of
the
control
planes.
It's
an
acknowledged
gap
that
they're
they
will
seek
to
overcome
at
some
point,
and
so
how
do
we
yeah
now?
A
And
so
your
analysis
here
is
oriented
to
the
goal
of
seeking
out
lowest
common
denominator
or
minimal
configuration
that
must
be
captured,
irrespective
of
what
service
mesh
is
being
used
is
like,
if
you're
going
to
insert
a
filter.
A
You
know
like
if,
if,
if
it's
the
same
proxy,
that's
running
in
the
data
plane
envoy
well
universally,
you
know
you
can,
it
can
be
stated
the
envoy.
Will
it
needs
this?
This
amount
of
info,
or
this
certain
set
of
criteria
to
to
to
run
such
a
filter,
you
know
intelligently
or
effectively
or
or
like
it,
requires
this
stuff
anyway.
So
the
things
you're
seeing
exposed
in
the
studio's
control
plane
is
really
just
a
manifest.
A
You
know
if
some
of
those
configurations
are
required.
Things
like
what
direction
did
you
want
to?
You
know
apply
this
filter
traffic
coming
in
or
traffic
going
out
like
how
could
that
not
be
universal
like?
How
could
you
not
and
so
yeah?
I?
I
don't
know
that,
like
the
things
that
I'm
saying
are
probably
reinforcing
your
thought
process,
which
is
I'll,
make
a
quick
analogy
to
a
very
related
concept,
and
that
is
one
of
firewall
rules,
I'm
very
related
to
filtering.
A
In
this
case,
it's
like,
if
you
look
over
to
it's
just
like
we
talk
about
physical
network
switches,
which
your
firewalls
can
be
implemented
on.
Any
of
your
host
machines,
any
of
the
host
operating
systems
that
you're
running
can
implement
firewall
rules
or
other
pieces
of
software
or
physical
equipment.
If
you
go
to
any
of
them,
it's
pretty
consistent.
Some
of
them,
you
know,
function
slightly
differently
than
the
next,
but
it's
pretty
consistent.
A
The
fact
that
you're
like
required
to
fill
in
at
least
like
four
things:
what's
the
source
ip
address,
what's
the
source
port
number,
what's
the
destination
ip
address,
what's
the
destination
port
number,
if
you
don't
have
those
four
things
you
can't
really
make
a
firewall
rule
and,
and
then
any
number
of
those
manufacturers
or
those
systems
might
expose
any
number
more
metadata
around
that
to
make
it.
You
know
to
enhance
the
experience
and
differentiate.
A
So
that's
part
of
the
challenge
here
is
like
okay.
Well,
what
what
things
are
universal
and
must
be
captured?
If
you
don't
identify
an
inbound
and
outbound,
then
the
assumption
is
it's
just
both,
I
guess
but
but
you
still
might
need
to
tell
envoy
that
and
to
the
extent
that
envoy
needs
to
be
configured
in
that
way
and
that's
a
universal
on-way
consideration.
A
A
A
A
Now
the
what
helps
make
us
or
make
me
anywhere
make
the
rest
of
us
feel
good
about
leaning
in
to
specify
some
things
that
might
be
in
conflict
with
how
a
different
control
plane
would
deal
with.
This
is
if
we're
able
to
make
a
statement
that
mescheri
is
able
to
like
whether
or
not
a
service
mesh
supports
envoy
filters
itself
by
virtue
of
the
fact
that
it's
running
envoy
in
its
data
plane,
like
console
for
example,
console
as
a
service
management
does
not
support.
A
If
you're
running
the
right
version
of
envoy,
you
don't
necessarily
need
to
ask
console
for
permission
to
insert
a
filter.
Okay,
then
that
makes
me
feel
better
about.
You
know:
defining
something
universal
in
measuring
itself,
not
not
in
like
in
measuring
but
specific
to
istio
or
specific
to
this
mesh.
A
Part
of
the
question
then
becomes
is
the
the
application
binary
interface,
the
interface
between
envoy
proper
and
these
filters
that
it
will
let
you
specify
if
we're
you
know
that
interface,
if
we're
exposing
some
of
those
configuration
items,
that's
probably
a
fairly
safe
bet
that
those
are
going
to
be
universal
across
every
mesh
that
uses
envoy
as
a
data
plane.
F
So
actually,
the
things
that
we
are
exposing
here
is
either
mostly
is
so
there
there
is,
there
could
be
one
concern,
and
that
is
there
are
some
configurations
which
is
automatically
enforced
by
sto,
but
if
a
service
mesh
is
not
actually
basically
is
not
even
capable
of
doing
that
thing
automatically,
then
we
will
have
to.
We
can
actually
go
ahead
and
attempt
to
do
that
thing
on
our
own,
for
example,
let's
say
there
is
this
configuration
we
are
taking
in
proxy
match.
F
In
this
case,
we
can
just
quickly
check
if,
if
certain
onward
version
is
running
or
not,
and
if
it
is,
then
we
can
drop
in
and
or
we
can
skip,
but
this
is
something
that
is
automatically
does
so.
Basically,
a
few
configurations
are
there
where
we
we,
according
to
this
proposal,
we
will
have
to
do
some
some
of
the
jobs
that
ideally
the
controller
should
have
been
doing,
but
because
it's
not
then
we'd
be
doing
so.
F
That
could
be
one
of
the
concern,
while
the
rest
is
actually
fairly
generic.
That
is,
these
are
capabilities
that
are
exposed
by
on
on
onward.
So,
for
example,
capturing
that
hey
where
you
want
to
drop
the
filter,
or
is
it
in
inbound
or
bound
or
basically
those
kind
of
things
or
in
fact,
in
fact,
it's
the
pilot,
also
just
it's
devised.
It
also
just
generates
those
protos
from
this
configuration
and
then
obviously,
after
some
changes,
it
sends
to
onward
filter
so
onward
proxy.
A
Some
of
this
stuff
will
take
us
a
little
while
to
get
really
comfortable
with,
but
yeah.
I
don't
yeah
I'll.
Take
a
for
my
part.
I
would
crush
I'll.
Take
a
well
a
second
look.
This
is
the
first
time
I
was
looking
it.
I
think
one
of
the
examples
I
think
in
the
example
did
you
use
rate
limiting
as
an
example.
A
E
Yes,
so
futures
are
closed
regarding
like
patterns
and
applications
and
filters,
the
basic
one
is
added,
and
I
completely
forgot
about
the
special
one
I'll
be
working
on
it
now,
I'm
so
sorry
driving
on
for
so
long
I'll
work
on
it.
I
took
up
other
issues.
I
was
working
on
that
so
and
I'm
sorry
about
that
I'll
work
on
it
and
update
you
on
that.
A
A
Good
cool,
okay!
Well,
then
good,
so
that
should
tell
you
that
you
didn't
miss
an
expectation.
A
Okay
good,
but
to
to
to
conclude
on
your
iterative
delivery,
which
is
fantastic.
It's
just
a
reminder
is
that
if,
when
when
people
are
in
either
applications
filters
or
patterns
the
these,
these
three
things
are
types
of
content.
If
you
will,
because
users
can
bring
their
own
or
they
or
meshrey
can
can
supply
them
seed
content.
A
If
users
want
to
bring
their
own
there's,
this
upload,
url
and
people
can
okay,
so
good,
there's
small
iterative
things
we
can
do
here
as
well.
There's
there's
something
there's
something
nice
being
explored
here
in
megan's
ui,
and
that
is
you
know,
small
outline
in
a
in
a
blue
and
the
font
is
a
little
bit
there's
some
small
items
that
are
different,
that
that
are
worth
discussing
on
a
different
call
or,
like
you
know,
separately
there's.
A
Maybe
you
know
that
we
begin
to
highlight
the
fact
that
there's
a
need
for
some
common
components
and
some
of
that
common
component
tree.
If
you
will
is
a
small
info
icon
or
like
the
small
little
question
mark
icon,
it's
like
import
url
here
or
import
using
url,
great
okay,
and
I
was
going
to
import
an
application
because
that's
the
page
that
I'm
on
okay,
so
like
part
of
the
questions
I
might
have
as
a
user
is
like,
is
what,
if
I,
what,
if
it's
behind
what
if
it's
protected,
can
it
can?
A
I
provide
credentials
and
no,
since
I
don't
see
that,
but
if
I
give
it
just
a
directory
like
I
can,
can
I
import
multiple
at
the
same
time?
Can
I
do
a
bulk
import
or
like,
or
does
it
will
it
refresh
the
import
every
so
often
or
like
I
don't
know
just
some
of
these
are
silly
questions
that
I'm
coming
up
with.
A
But
the
point
being
is
that
oh
at
some
point
it
would
be
ideal
if
we
take
this
modal
component
and
we
begin
to
you
know,
build
out
common
functionality
in
the
modals
that
we
reuse,
such
that
something
like
a
like
in
this
modal
there's
no
x
to
close
it
and
that's
because
you
can
click
to
just
dismiss
it.
I
don't
know
if
that's
universally
the
same,
and
I
don't
know
you
know,
to
the
extent
that
we
would
use
an
info
or
a
question
mark.
A
E
A
Yeah,
if,
if
I
I
can't
even
remember
if
the
close
button
is
there
for
the
but
but
the
point
being
is
like
whether
it
is
or
isn't
like
yeah
you
driving
and
collaborating
with
nipish
karthik
is
probably
the
best
that
is
driving
commonality
across
yeah.
The
interface.
D
A
G
A
A
A
A
By
the
way,
small
verbiage-
I
don't
know,
if
I
don't
know,
if
you're
tracking
this
as
a
bulk
delete
or
not,
but
if
you
are,
then
we
would
just
change
this
to
these,
but
actually
what's
an
interesting
takeaway
as
well.
Is
that
the
check
marks?
I
don't
see
them
in
the
background,
so
it's
like.
Do
I
want
to
delete
these
well,
which,
which
ones
was
that
again?
No
I'm
not
so
this
is
one
of
ten
delete
and
then
it's
like
they
get.
They
disappear
from
the
table
immediately.
A
G
Okay,
I
got
it
you're
saying
that
getting
the
pattern
name
in
that
delete
modulation,
yeah.
A
That
would
be
really
nice
a
confidence
inspiring
and
like
yeah.
You
could
even
arguably
eliminate
the
need
to
show
them
as
checked
in
the
background,
because
you're,
because
that's
what
the
user,
because
because
the
only
reason
that
it
bugs
me
that
they're
not
in
the
background
is
because
I'm
like,
oh
well,
I
thought
I
did,
but
I'm
not
sure
like
this
is
you're
asking
me,
because
this
is
a
serious
thing,
so
I
want
to
seriously
double
check
it's
like.
A
Well,
I
don't
see
them
checked,
okay
and
then
you're
not
telling
me
what
I'm
going
to
delete.
So
it's
like
no,
I'm
not
comfortable,
okay,
okay,
but
if
I
do
just
to
one,
I
think
it's
just
you
know.
Okay,
it
still
asks
me.
Okay,
you
want
to
delete
this
one.
Well
again,
oh
yeah
same
behavior
right
so.
A
Okay,
any
feedback
for
garf.
A
Before
we
head
into
this
and
kind
of
get
lost
for
five
minutes,
talking
about
this
it'd
be
good
to
tee
up
what
we're
gonna
look
at
next
week,
gaurav
if
you're
in
here,
I
think
you
might
have
a
couple
items
to
follow
up
on.
A
G
And
one
thing
for
the
yeah:
you
can
do
it
yeah.
I
was
asking
for
those
similar
kind
of
thing
functionality
for
configuration
of
filters
that
we
have
in
itself.
So
can
we
have
a
similar
thing
that
we
currently
having
in
the
patents
for
configuration,
because
we
obviously
need
that
thing
in
the
filters
as
well.
A
Yep,
so
this
is
works
for
patterns,
okay,
good,
but
then
yeah
having
it
for
applications.
A
A
Yeah,
I
don't
know,
and
then
that's
because
like
I
just
I
don't
know.
A
A
So
sohom
or
wood
cars
for
you
guys
that
are,
I
guess
it
would
crash.
F
I
I'm
also
not
sure,
because
I
think
right
now,
we
do
generate
pattern,
but
it's
not
us
that
it
does
in
case
of
application
or
in
case
of
obviously
parents.
So
if
you
do
allow
them
to
I
mean
what
would
they
actually
be?
Would
they
be
just
getting
a
form,
a
single
form,
because
they
can
cannot
be
more
than
one
increase
if
it's
just
top
filter,
because
it's
just
a
filter,
so
I'm
not
sure
actually.
G
A
E
Yeah
sure
and
there's
one
more
thing:
I
actually
need
some
url
examples
for
adding
the
filters
and
patterns,
because
initially
what
happened
was
I
used
a
random
url
validator
that
I
found
online
and
I
checked
it
with
that.
Then
I
changed
it
to
the
one
that
we
already
have
in
the
direct
like
in
our
source
code.
So
I
didn't
get
a
chance
to
check
after
that.
D
A
B
E
Not
sure
exactly
should
I
just
should
I
just
face
or
try
pasting
one
of
the
url
from
here
itself,
like
github
like
one
second.
A
D
A
Something
else
okay,
so
we
only
have
a
couple
of
minutes
left.
We
won't
go
through
the
remainder
of
this
spec.
What
what
I
will?
What
I'll
say
about
this
is
that
so
last
time
the
venue
had
taken
us
through
a
draft
of
a
specification
for
how
to
bundle
pattern
like
useful
filters
patterns
and
applications
into
mastery
itself,
so
that
when
people
deploy
measuring,
it's
not
just
a
bunch
of
empty
tables
with
like
with
them
asking
the
same
questions
that
megan
is
asking
which
is
like
okay
great.
A
Where
do
I
get
some
of
these
examples
like
where?
Where
can
I
you
know,
it'd
be
much
more
helpful
if
measuring
just
had
this
stuff
built
in,
and
so
this
spec
has
been
rewritten
with
that
goal
in
mind
of
identifying
the
fact
that
there
are
that
measure
is
going
to
want
to
come
pre-bundled
out
of
the
box
with
some
content,
some
value,
you
know
some
value
inside
of
and
that
content
comes
by
way
of
filters
patterns
applications.
A
All
of
this
has
been
rewritten
is
good
good,
to
go
good
to
comment
on
where
we,
where
the
rubber
meets
the
road
in
terms
of
beginning
to
execute
on
bundling
of
these
things,.
A
That
would
also
fetch
patterns
and
filters
from
like
the
urls
that
we
were
just
describing
and
and
pull
in
those
artifacts
into
the
in
this
case,
into
the
local
provider,
and
that
there
would
be
a
collection
of
those
that
are
just
available
out
of
the
box
outside
of
just
the
fact.
The
literal
effect
of
like
retrieving
that
content
and
pulling
it
into
the
workflow
is
that
the
workflow
or
mesherie
itself.
It
needs
to
re
like
self-register.
A
So
when
meshre
server
spins
up
it
does
a
few
initialization
routines
and
one
of
those
can
be
looking
to
its
local
docker
image
like
measuring
servers
running
in
a
docker
container
and
within
that
same
image,
is
you
know,
a
file
system
that
has
stuff
like
seed
content
and
and
so
it
can
just
have
this
initialization
routine,
that
when
it
when
it
starts,
it
can
look
to
its
local
file
system
if
it
finds
any
seed
content
to
just
go
ahead
and
invoke
its
own
apis
to
register.
A
A
A
And
yeah
so
think
about
this.
If
you
would,
I
guess
the
thing
that
we'll
put
for
next
week
next
week's
topic
is.
A
Like
I
guess
I'll
put
this
here
as
the
goal
of
having
converted
this
to
an
issue
and
having
folks
having
spoken
with
a
few
people
to
make
sure
that
there's
enough
detail
in
there
that
we
can
start
to
iterate
on
that,
it's
probably
multiple
people
right.
A
Somebody
may
be
working
on
a
workflow
to
check
out
content,
bring
it
into
the
build
time,
someone
else
maybe
to
work
on
measuring
server
to
say,
hey
when
it
starts
up.
The
service
server
starts
that
it
would
have
this
initialization
routine
yeah,
there's
already
an
api
in
mesheri
to
import
by
uri
or
to
import
by
file
system
path,
that's
already
supported,
and
so
megana.
That's
actually
a
really
good
thing
to
just
take
a
quick
look
at
when
we're
importing
filters
by
url
megane.
Do
you
is
this
also
capable?
A
A
Yeah
yeah
that
in
the
same
model
like
if,
if,
if
what's
what's
really
being
done,
is
like
they're
they're
importing,
filter
import
filter
and
they
can
either
paste
in
a
url
or
they
can
click
browse
to
identify
and
upload.
So
it's
like.
A
Oh,
you
know
what
I'm
sorry
I
totally
yeah.
I
I
forgot
that
this
is
right
here.
Yeah,
I'm
sorry!
I
don't
you
know.
Let
me
let
me
retract
the
stuff
I
just
said
like
I
maybe
yeah.
Maybe
there's
I
don't.
Maybe
it's
better.
If
they're
combined,
I
don't
know,
but
but
it's
not
worth
doing
work
on
it
until
it's
better
thought
through
like
yeah
right
now,
we
do
have
both
okay
got.
It.