►
From YouTube: WebAssembly Filters Meeting (Aug 23rd, 2021)
Description
WebAssembly Filters Meeting - Aug 23rd, 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
A
Good,
I
hope
your
monday
is
going
better
than
mine
all
right
so
for
those
that
are
on
the
call,
please
do
drop
your
name
into
the
meeting
minutes.
You
will
be
that'll,
be
at
least
five
ten
minutes
of
internet
fame.
When
you
do
also
speaking
of
internet
fame,
we
end
up
publishing
our
youtube.
You
know
our
meeting
our
meeting
recordings
on
youtube,
so
so
this
will
be
up
there
soon.
A
If
now
is
the
right
time
to
call
for
topics,
so
we've
got
one
topic
today,
which
is
great.
It's
a
it's
a
meaty
topic,
but
what
else
do
other
people
have
items
to
go
over?
I
didn't
copy
from
down
below.
Since
last
we
met.
I
know
we
had
some
spillover
from
last
time.
We
met
so
take
a
moment
now
go
and
review
so
utkarsh.
We
had
a
topic
with
you
last
time.
I
don't
see
anything
for
this
week,
though
so
I'm
hoping
there's
progress,
megana.
A
A
So
anyway,
now
is
the
time
to
get
the
topics
listed.
So
go
ahead
and
take
a
moment
get
your
topics
there.
I'm
gonna
go
ping
a
couple
of
people
to
join
while
everyone
works
on
that.
B
Nope
there
is
no
update
on
actually
I've
created
the
issue,
but
there
are
no
further
deals
that
are
presented
in
the
issue.
So
no
update
on
last
week's
topic.
A
A
A
All
right,
good
and
ashish
is
with
us
good.
Okay,
all
right,
I'm
getting
some
topics
down
the
first
one,
so
angara
of
no
update
this
week,
then,
okay,
so
still
no
bulk
delete
on
filters.
I
assume.
A
Okay,
all
right!
Well,
let's
oops!
That
was
the
wrong
button.
Let's
get
going
so
the
venue
you've
got
the
first
one.
Do
you
want
to
you
want
to
take
it
away?
You
want
to
take
people
through
the
design,
spec.
E
These
are
currently
available
through
measuring
as
well
as
the
provider.
So
what
this
design
spec
aims
to
deliver
is
to
use
measuring
as
an
artifact
repository,
so
when
a
user,
so
the
use
case
is
basically
that
when
a
user
first
installs
measuring,
they
should
have
access
to
a
set
of
sample
filters,
sample
patterns
and
sls
sample
applications.
E
So
we
we
already
have
sample
applications
in
the
current
version
of
measuring
that
I,
that
is,
the
users,
are
able
to
already
comes
with
some
sample
applications
like
the
booking
for
application,
the
bookstore
application
emoji
water
application,
so
these
applications
come
pre-packaged
with
measuring,
and
users
can
directly
deploy
or
onboard
these
applications
onto
their
service
mesh.
E
So
what
this
spec
aims
to
do
is
do
something
similar
for
patterns
awesome
filters,
as
well
as
applications
as
a
separate
construct
and.
E
So
there
are
some
constraints
with
this
design.
That
is,
we
should
so.
The
the
main
the
main
point
is
here
is
that
we
will
be
packaging.
These
sample
patterns
filters
and
applications
into
measuring.
E
So
when
the
measuring
server
is
built,
these
sample
files
will
be
fetched
and
it
will
be,
it
will
be
part
of
the
actual
built
and
so
that,
when
a
user
installs
measuring
it
can
automatically
it
is
automatically
with
them,
so
they
don't
have
to
download
anything
or
they
don't
have
to
import
anything
from
from
from
a
url
or
something
so
that
would
also
enable
air
gap
support
and
it
it
should
also
be
that
the
user
can
control
what
sample
applications
or
what
sample
artifacts
they
have.
E
E
So
the
the
overall
goal
here
is
to
can
be
defined
in
two
steps.
E
Basically,
the
first
step
is
to
files,
basically
fetch
these
files
and
bundle
them
with
the
measuring
server
builds
and
when
measuring
first
starts,
we
should
import
these
files
into
measuring
so
that
the
user
can
actually
use
this
use
these
artifacts
right
out
of
the
box.
E
So
there
are
the
two
steps
that
we
have
and
the
other
constraint
is,
as
I
mentioned,
the
aircap
support.
So
basically,
if
we
are
bundling
these
files
with
the
measuring
server
build,
it
should
automatically
be
supporting
aircap
deployments,
so
users
would
not
have
to
connect
to
internet
to
get
this
working.
E
Yeah
right
and
the
other
constraint
I
mentioned
with
the
persisting
changes
they
make
to
sample
buttons.
So
if
you
are
bundling
this,
if
you
are
bundling
this
with
measuring
measuring
image,
what
what
could
happen
is
every
time
the
user
updates
measuring
or
research
measuring
it.
These
sample
applications
could
be
updated,
but
the
user
could
have
modified
these
files
as
well.
E
So
this
is
a
something
that
we
might
need
to
look
into
in
the
future,
so
we
should
also
provide
the
user,
the
ability
to
make
changes
to
the
sample
applications
or
the
sample
filters
or
the
sample
patterns
as
well.
E
That's
pretty
much
it.
This
pack
is
in
draft
seats
there.
It
still
needs
some
iteration
to
figure
out
how
exactly
we
will
implement
it.
But
I
see
I
see
some
changes,
some
edits
already
in
this
pack,
but
yeah.
If
you
have
any
questions,
please
ask
them
or
if
you,
if
you,
if
you
need
any
changes,
please
arrest
them
now.
A
So
root
rocks
and
utkarsh.
The
two
of
you
might
be
closest
to
this.
A
Okay,
very
good,
all
right,
fair
enough.
Let
me
let
me
give
you
guys
some
things
to
think
about
some
stuff
to
think
about,
so
this
particular
design
spec
could
be,
could
be.
Alternatively,
named
potentially
like
in
part
what
this
begins.
The
design
spec
starts
to
speak
to
and
sort
of
unravels
into
is
content
management
within
measures,
so
mesherie's
an
extensible
platform
mesherie
allows
users
to
bring
in
their
own
quote
content.
A
Okay,
the
part
of
the
prologue
here
begins
to
define,
what's
meant
by
content
so,
and
it
needs
to
define
a
couple
of
definitions
for
those
that
aren't
familiar.
So
most
almost
all
of
you,
or
most
of
you
have
signed
in
into
mescheri
before
you've
deployed
meshri.
The
first
interface
web-based
interface,
that
you
see
is
one
of
a
choice
of
provider
like
do
you
want
to
which
provider
would
you
like
to
use?
There's
a
built-in
provider
called
none
or
called
the
local
provider.
A
A
Common
things
you
get
with
a
provider
are
identity,
persistence
or
storage,
and
so,
just
like
you
might
yeah
yeah.
So,
okay
well
part
of
that
storage.
Then,
as
you
connect
to
a
provider
whether
it's
the
local
provider
or
a
remote
provider,
they
may
that
provider
might
have
see
it
might
have
content,
it
might
have
like
your
user
preferences.
A
It
might
have
filters
that
you've
created
patterns
that
you've
made
applications
that
you've
uploaded
your
your
kubernetes
apps
or
to
facilitate
quicktime
to
value
so
that
people
who
are
just
learning
meshrey,
are
just
learning
service
meshes
or
don't
really
know
what
the
power
of
the
tool
is,
that
they
would.
There
would
be
some
content
resident
or
seated
inside
of
out
of
the
box,
so
to
speak.
Now
the
quote-unquote
out-of-the-box
experience
is,
is
both
a
combination
of
measuring
server
itself
like
measure
the
bits
that
are
released.
A
The
remote
providers
that
they're
remote
they're
written
separately
and
run
separately
and
they
may
offer
up
different
types
of
content,
but
both
to
facilitate
quick
time
to
value
and
an
understanding
of
what
mesh
re
does
and
how
to
manage
service
meshes
this.
This
spec
is
proposing
that
there
should
be
any
number
of
filters
service,
mesh
patterns,
applications.
A
Maybe
performance
profiles
like
common
benchmarks,
like
any
amount
of
these,
that
are
predefined
and
readily
available
for
users
to
use.
Now
the
word
sample
comes
in
here
when
we
talk
about
this
type
of
content,
like
all
of
those
types
of
things,
can
be
actually
the
venue.
If
you
scroll
up
to
the
design
prologue,
there
is
the
definition,
a
distinction
and
a
definition
of
like
user-produced
content
like
what
is
content
first
of
all
and
what
is
user-produced
content.
What
is
provider-produced,
they
overlap
like
right
now.
A
The
way
that
they
can
distinguish
is
that
when
you're,
a
user
you're
using
measure
you
generate,
you
run
some
performance
tests.
We
have.
You
have
results
for
those
performance
tests,
those
you
produced
those
results.
As
a
user,
the
provider
didn't
say,
here's
some
tests
that
have
been
run
and
here's
their
results
like
that
might
be
interesting
to
you
as
a
benchmark,
but
that's
generally
not
what's
going
on,
like
you
are
running
the
test,
you're,
generating
the
results
for.
A
But
and
then
oh
yeah
and
you
might
also
be
bringing
your
own
filters,
your
own
patterns,
your
own
applications,
your
own
content,
which
is
you
know
very
much
intentionally
you
know
you
should
be
measuring,
is
designed
it's
for
you
to
bring
that.
If,
when
you
go
to
run
mesh,
read
there's
nothing
in
there.
There's
no
filters.
There's
no
patterns,
there's
no
applications.
A
You
it
takes
the
user
a
while
longer
to
grok
what
this
functionality
is
and
to
also
it
takes
longer
to
get
to
the
value
of
what
measuring
can
deliver.
So
there
are,
you
know,
as
a
community
we've
been
working
on,
I
think
we've
got
like
six
or
seven
different
filters
that
are
published.
Good,
hey.
Why
don't
we
make
those
readily
available
like,
irrespective
of
what
provider
someone
chooses
to
use
and
why
don't
we
make
those
readily
a
bit?
Why
don't
we
seed
some
content?
A
Why
don't
we
have
some
example
pattern
we've
been
working
on
patterns
they're
publicly
published.
Why
are
those
not
readily
available?
Initially
I
mean
they
are
readily
available
to
meshuri.
The
user
can
go
in
and
import
them,
but
they
have
to
take
that
step.
So
why
don't
we
just
have
those
available
out
of
the
box
and
that's
the
proposal
that
the
vendor
is
writing
here
is
is
to
basically
say,
let's
facilitate
quick
time
to
value,
provide
users
with
quote-unquote
sample
content.
A
I
would
say
that
it's
seed
content,
because
the
filters
that
are
provided
the
patterns
that
are
provided
they're
not
exactly
they're,
not
samples
to
the
extent
that
they're,
not
production,
worthy
that
they're
not
highly
useful.
That's
that's
so
we
have
to
be
careful
with
that
potential
misnomer,
while
they
are
sample
filters
that
are
provided
as
a
sample
for
you
to
use
like,
I
think
the
use
of
that
word
samples,
probably
probably
doesn't
do
them
justice.
A
Rather,
these
are
the
out-of-the-box
functionality
out
of
the
box
content,
the
seed
content
to
get
you
going,
and
maybe
it's
all
that
you
need
good.
So
so
that's
that's
a
large
part.
What
the
design
is
about
part
of
what
happens
is
there's
kind
of.
So
if
we
go
to
design
goals,
that's
the
goal.
Number
one
is
like
facilitate
short
time
to
value,
offer
seed
content
goal.
A
Number
two
is
to
well
begin
to
consider
not
begin
to
like
consider
air
gap
deployments
and
air
gap
deployments
have
something
in
in
common
with
this
design
goal
of
seeding
content,
and
that
is
if
there
isn't
content,
that's
seated.
If
there
aren't
out
of
the
box
filters
and
patterns
and
applications,
then
well,
then
you've
got
to
import
them
into
meshri,
and
so
you
either
need
to
bring
your
usb
stick
and
walk
up
to
measuring
and
plug
it
into
the
measuring
and
load
up
the
you
know
your
filters
in
there
or
you
need
to
connect.
A
You
know,
use
the
network
to
connect,
and
so
in
air
gap
deployments
you
bring
your
usb
stick
and
it
would
be
nice
if
people
didn't
have
to
walk
around
with
the
usb
stick
to
facilitate
the
air
gap.
If
you're
not
familiar
with
an
air
gap
deployment.
Is
that
means
that
you're
deploying
meshi
into
an
environment
that
does
not
have
an
internet
connection?
There
is
no
internet
connection,
so
the
thought
of
like
dynamically
retrieving
a
filter
and
loading
it
in
from
a
url.
A
Well
that
url
has
to
be
local
like
local
to
that
mastery
deployment.
That's
what
air
gap
haircut
means:
okay,
understanding
that
there
are
any
number
of
users
that
run
infrastructure
and
want
to
run
mesher
in
that
type
of
environment.
A
Seeding
content
will
help
with
overcoming
some
of
those
air
gap
challenges
if
the
some
example
or
not
example,
but
yeah.
If
I'm
out
of
the
box
content,
is
available
and
is
right
there
in
the
box,
then
people
don't
have
to
go,
get
their
usb
stick.
They
have
to
go
figure
something
out.
So
that's
good,
so
we
should
think
about
that
air
gap
deployments
and
the
considerations
there.
The
third
thing
that
this
spec
touches
on
is
honoring
user
customizations,
which
is
to
say
that
as
a
user,
you
might
have
uploaded.
A
You
might
have
taken
some
of
the
like
the
patterns
that
are
provided
out
of
the
box
like
as
we
go
to
do
this
as
you
can
implement
this
spec
and
provide
people
with
any
number
of
filters.
Any
number
of
patterns
there's
they
want
to
change
the
configuration
of
those
change.
The
configuration
of
the
filter
upload
their
own
filters,
maybe
delete
the
existing
filters
because
they
don't
like
them,
they're
not
going
to
use
them.
A
A
A
Meshrey
will
have
this
behavior,
whatever
that
behavior
is
and
it
needs
to
be
described
in
this
spec
could
be
different
things.
It
could
be
a
big
fat
warning.
Sign,
comes
up,
saying,
you're
about
to
overwrite
your
customizations
or
no
need
for
a
warning
sign
it
just
doesn't
overwrite
customizations
like
so
so
there's
there's
different
ways
to
handle
it.
A
That's
the
overarching
objectives
that
are
defined
here
and
in
the
way
that
the
reason
that
we're
couching
that
we're
defining
a
couple
of
things
we're
defining
the
fact
that
there's
content,
which
builds
upon
the
fact
that
mesherie
is
a
platform,
it's
an
extensible
platform.
A
You
can
bring
content
content
like
like
that
word
or
not,
like
that's
a
generic
term
for
describing
the
the
variety
of
things
that
we've
just
been
talking
about
specifically
filters,
and
that's
this
design
spec
is
in
cons
in
is
written
in
context
of
filters,
because
the
question
is
just
that,
like
there
are
no
like
many
of
you
that
want
to
go,
use
the
filters
functionality
and
lay
down
a
custom
filter
in
your
service
mesh
deployment.
You
might
be
saying:
where
do
I
get
some
filters
where
I
get
my
hands
on
those?
A
Do
I
have
to
go
write?
Those
like
you
don't
have
to
there
are
you?
Can
there
are
some
that
are
available
to
you?
Okay,
so
why
are
we
bringing
providers
into
this
like?
Why?
Don't
we
just
bake
those
public?
You
know
those
publicly
available
filters
into
a
given
release
of
measuring
it's
like
well,
that's
kind
of
what
we
are
we're
basically
saying
we're
basically
proposing
that
that
is
what
be
done,
but
that
baking
it
in
and
putting
it
right
behind
a
the
provider.
A
Interface
means
that
when
someone
uses
the
local
provider
like
they
have
to
make
that
choice
they're
in
one
of
two
modes,
they're
either
using
a
local
provider
or
a
remote
one,
and
having
this
content
behind
that
interface
means
that
the
provider
the
the
the
remote
provider
can
deliver
any
number
of
filters.
However
many
they
want.
A
Now
we
are
going
to
bake
them
in,
to
the
extent
that
they
would
be
packaged
and
available
out
of
the
box
for
anyone
using
the
local
provider,
and
they
would
be
resident
like,
like
the
those
filters,
would
be
act
literally
inside
of
the
same
same
docker
image
and
there
might
be
more
filters
available
on
a
subsequent
release
of
measuring
having
them
behind.
The
provider's
construct
is
what
we
need
to
do
for
all
content,
because
it
needs
to
be
clear
to
users
what
they're
getting
in
terms
of
persistence
of
that
content.
A
A
Cool
okay,
so
I
went
off
for
a
little
while,
but
subsequent
to
that
this,
this,
the
design
spec,
is
a
great
draft.
It's
it's
in
draft.
It's
needs
a
bit
of
refactoring
refinement.
It
is
going
through
that
and
that's
what
you
all
are
encouraged
to
go.
Look
at
one
of
those
open
challenges.
I
I
haven't
read
the
whole
spec
myself,
but
an
open
challenge
is
like
how
to
honor
those
customizations.
A
If
someone
takes
an
existing
like
a
filter
that
you've
given
them,
that's
out
of
the
box,
they
change
the
configuration
they're
running
one
version
of
mesheri.
They
go
to
update
to
the
next
version.
Well,
in
the
next
version
you
actually
have.
We
actually
have
a
whole
new,
better
filter,
better,
bigger,
better
nicer,
whatever
it
is
faster,
whatever
so
shouldn't
the
user,
get
that
newer,
nicer,
whiz-bang,
filter,
that's
the
same
filter,
it's
just
a
different
version:
oh
wait!
A
second!
I
just
said
version:
it's
a
different
version
of
that
filter.
A
A
Yeah,
I
don't
know
the
fair
enough,
no,
no
problem.
I
went
off
and
got
coffee
myself
too.
So
do
so
ashish
man,
I
just
like
there's
no
mercy
on
this
monday.
I'm
just
calling
people
out
like
any
ideas
that
you
might
register
content.
B
C
A
Cool,
so
all
right,
so
fair
enough,
I
am
my,
and
so
by
the
way,
I'm
not
I'm
only
I'm
I'm
harassing
everyone
in
the
best
of
ways,
hopefully
just
to
kind
of
help.
Everyone
help
reinforce
everyone's
understanding
of
what's
going
on
and
so
yeah
we
don't.
We
haven't
tossed
around
the
term
register
much
so
you
know
that
way.
A
The
question
is
a
little
bit
unfair
and
and
by
the
way,
those
that
didn't
answer
the
question
like
a
you
didn't
hear
it
so
don't
even
sweat
it
b,
you
didn't
even
get
a
chance
to
read
the
specs
or
don't
even
sweat
it
like.
Don't
don't
sweat
the
it's
not
I
don't
mean
to
make
anyone
feel
uncomfortable
in
that
way.
Only
to
like
get
your
juices
flowing,
get
your
getting
get
some
new
noodles
created
in
your
brain
like
to
wake
us
up
and
to
do
some
stuff.
A
So
the
way
that
like,
if
you
think
about
content,
being
managed
in
measuring
people
being
able
to
upload
their
own
content
or
import
a
pattern,
upload
a
filter,
they're
they're,
like
registering
content
with
messages,
so
messy
becomes
aware
of
the
fact
that
there
is
a
pattern
that
that
a
user
might
want
to
deploy
and
so
that
that
pattern,
that
content
becomes
registered.
At
that
point
with
metric
like
it's,
it's
in
mesher's
database
or
message
becomes
aware
of
it
so
that
yeah
yeah.
A
So,
okay,
so
like
there's
this
implicit
process
of
registration
happening,
it's
like
you
can
we
haven't
gotten
there
yet,
but,
but
we
will
like
measure,
has
a
long
roadmap
and-
and
it's
not
fully
articulated,
because
because
we
have
a
hard
enough
time
kind
of
getting
through
something
like
this,
where
we're
trying
to
understand
what
we
mean
by
registration
much
less.
What
do
we
mean
by
policy
driven
workflow
or
orchestration
like
that's
a
whole
like
we
desire
to
have
gotten
there
last
year.
A
Just
takes
a
long
darn
time
to
get
to
some
of
these
things,
and
so
so.
Registration
and
dealing
with
content
means
that
there
should
be
a
finite
state
machine,
an
fsm
everybody.
Everybody
recall
what
a
finite
state
machine
is
and
if
not,
maybe
some
of
us
haven't
gotten
to
it.
Yet
some
of
us
have
been
there
there.
It
sounds
like
a
big
old
word
and
it's
not
really
not
a
big
concept,
but
I'm
not
going
to
describe
what
a
finite
state
machine
is.
A
A
So
I
won't.
I
said
that
I
wouldn't
describe
it,
so
I
won't
okay,
all
right
good.
So
so
we
described
what
do
we
mean
by
registration?
What
do
we
mean
by
honoring
user
customizations
good?
What
do
we
mean
by
content?
Yeah?
Okay?
We
gave
some
examples
of
those.
We
gave
some
examples
of
the
fact
that
some
content
is
user
produced.
Some
is
provider
produced,
okay,
okay,
we
gave
we
said
well.
Why
are
why
did
providers
come
into
this
like
shouldn't?
We
just
package
it
out
of
the
box.
It's
like
okay!
A
A
Well,
the
provider
offers
that,
like
the
mesherie's
goals,
one
of
us
in
like
in
in
the
charter
of
measuring
the
vision
of
the
thing
it's
not
dissimilar
from
well
one
aspect
of
it.
One
aspect
is
not
dissimilar
from
prometheus.
A
What
does
measury
have
to
do
with
metrics?
Well
well,
in
this
way,
the
prometheus
project
at
mesri
are
the
same.
You
know
what's
really
interesting
about
one
interesting
factoid
about
prometheus.
Is
that
so
so
prometheus?
If
you're
not
familiar
like
you
know,
it
collects
a
bunch
of
metrics
tons
of
metrics
so
quickly
so
fast
and
it
collects
a
lot.
They
change
so
frequently
the
environments
are
ephemeral,
things
are
coming
and
going
all
the
time
it's
gotta
like
so.
A
The
way
that
it
does
storage
of
metrics
is
like
at
the
core
of
one
of
is
one
of
the
core
capabilities
of
prometheus
one
of
the
ways
that
it
does
so
it
needs
to
do
it
efficiently.
It'll
burn
through
a
physical,
hard
drive
fairly
quickly
in
a
large
kubernetes
environment.
Oh
yeah,
you
can
and
there's
there's
tests
out
there
that
show
so
the
way
that
so
they've
rewritten
the
storage
model
a
couple
of
times.
A
It's
interesting
they've,
you
know
sort
of
gotten
about
as
small
as
you
can
get
it
as
compact,
as
you
can
get
it,
and
anyway,
so
for
as
front
and
center
as
storage
is
and
storing,
metrics
is
to
prometheus.
A
What's
one
of
what's
one
of
not
the
project's
goals,
this
project
does
not
deliver
persistent
storage
like
that
is
an
anti-goal.
This
project
doesn't
do
it.
They
say.
Storage
is
hard,
we're
not
doing
it.
We're
facilitating
the
collection
of
metrics,
the
writing
them
to
disk.
Whatever
disk
you
bring
whatever
storage,
that
is
they
facilitate
that.
But
beyond
that,
it's
like
hey,
that's
in
your
head:
that's
not!
They
don't
want
to
deal
with
it.
Consequently,
other
projects
have
popped
up,
so
is
amman
on
yeah.
A
So
oman,
can
you
know,
like
he's
sitting
here,
he's
following
he's
following
this
he's
like
yeah,
yeah
amman,
you
want
to
tell
us
about
some
other
projects
that
have
popped
up
to.
C
Yes,
so
I
guess
you
are
talking
about
thanos,
so
there
are
some
other
projects
like
thanos
and
cortex,
which
are
which
arise
around
the
prometheus
ecosystem,
because
there
were
some
drawbacks
or
some
limitations
which
prometheus
has,
which
likely
mentioned
right
now
that
it
does
not
provide
long-term
storage
capability.
C
So
for
that
we
have
other
tools
of
from
other
tools
like
thanos,
which
does
that
of
providing
the
long-term
storage.
It
has
a
side,
car
and
a
complete
thing,
so
the
complete
detail
might
not
be
good
to
share
it
right
now,
but
yeah.
A
A
Well,
it's
not
a
dissimilar
story.
It's
like
mastery
is
trying
to
help.
You
manage
cloud-native
infrastructure,
particularly
this
newfangled
service,
mesh
layer
and
all
that
it
can
do
good.
My
goodness,
that's
enough
onto
its
own
trying
to
persist,
persist
your
filters
and
applications,
your
patterns
and
stuff
and
then
provide
it
to
you
and
guarantee
all
that.
Based
on
what
storage
you
want
to
use.
A
Let's
draw
the
line,
so
we
drew
the
line,
but
we
also
acknowledged,
like
oh,
it's
gonna,
be
a
pain
in
the
rump.
If
people
have
been
running
performance
tests
for
a
while
they've
been
benchmarking,
their
result,
baselining
their
environment,
looking
at
the
results
and
then
they're
trying
to
baseline
like
hey
last
month.
How
am
I
doing
last
month
to
now
wait
a
second.
What
happened
to
last
month's
results?
Oh
they're
gone,
oh,
like
hey,
like
hey
they're
gonna,
want
long-term
storage.
A
Let's
provide
an
interface
behind
which
a
storage
provider
a
provider,
a
remote
provider
from
sri.
That's
what
we
call
them,
because
these
remote
provide
these
providers
don't
just
deliver
in
measuring
they
don't
just
deliver
storage.
They
can
deliver
other
things
like
filters
like
like
what
we're
talking
about
now.
So
if
you
want
to
kind
of
grok
more
about
the
concept
of
a
provider,
this
is
an
excellent
place
to
you
know.
A
So
anyway,
that's
in
part
like
prometheus
and
mashri
drew
a
similar
line.
Prometheus
has,
oh,
I
think
it's
actually
called
remote
storage
providers.
I
think
it's
the
local,
a
remote
storage
provider.
You
can
write
your
own.
I've
had
my
teams
in
the
past
write,
read:
write
storage,
provide
remote
storage
providers
for
some
of
the
things
we
were
doing.
It
was
prometheus,
but
but
point
is
like
there's
that
line
the
point.
A
So
so,
my
goodness
I
can
talk.
I
apologize.
I
didn't
realize
it's
been
like
been
so
long
for
me
to
so.
Okay,
you
know
I
I
get
that
most
of
you
are
kind
of
just
seeing
this
for
the
first
time.
You
have
any
time
to
go
through
it.
It's
very
fresh,
it's
still
being
written,
please
jump
in!
Oh
look,
profs
all
over
every
brock.
Nice,
you
guys
are
oh,
very
good.
Get
in
ask
some
questions.
Learn
some
learn,
learn
some
things
suggest
some
things.
Some
of
this.
A
A
It's
pretty
neat
neven,
dude
and
rude
rocks.
I
know
you
were
thinking
on
this
as
well
that
yeah
all
of
a
sudden.
This
isn't
just
about
filters.
This
starts
to
become
this
is
a
big
old
consideration
across
measuring,
so
pretty
cool.
F
Yeah,
thankfully,
so
just
an
update
on
editing
filter
conflicts.
So
we
can.
F
We
are
already
have
support
for
deploying
or
designing
filters
on
mesh
map,
but
we
didn't
have
support
for
editing
the
configuration
for
filters
yet
because
in
last
or
maybe
previous
to
that
week's
development
meeting,
I
showcased
that
we
could
drop
filters
onto
notes,
but
the
configuration
tooltip
that
basically
popped
up
wasn't
editable
for
filters
yet
and
also
I
had
shown
that
we
could
edit
configurations
for
other
workloads
in
general
button.
F
But
last
week
we
like
I
found
out
that
there
was
a
small
block
with
with
regards
to
the
whole
tool,
tip
and
editing
it.
So
just
fixing
that
and
also
implementing
it
for
filters
also.
F
Oh
yeah,
so,
as
he's
typing
that
there
are
three
ways
I
guess
yeah,
there
are
three
ways
in
which
users
can
configure
filters.
There
is
a
configuration
tab
in
the
machine
ui
which
lets
you
upload
and
configure
filters.
There
is
mesh
map
itself
where
you
can
drag
and
drop
your
filters
and
mess
receipt.
L,
I
guess
novena,
would
be
a
better
person
who
can
answer
that.
A
A
A
How
else
could
we,
it
might
be
the
the
built,
the
binary,
the
filter
itself,
the
binary,
maybe
it's
the
same,
but
maybe
the
configuration
is
different.
Yeah
yeah
now
that
I
open
this
can
of
worms
like.
I
think
it's
a
little
bit
of
a
can
of
worms,
because.
A
Because
the
configuration
that
you
have
for
a
filter-
well,
it
may
not
be
just
as
simple
as
like.
Here's
a
binary
file
and
here's
a
small
configuration
string
that
it
may
be
deployment
specific
or
at
the
time
that
you
go
to
apply
the
filter.
You
go
to
deploy
it,
that
small
configuration
string.
You
know
the
binary
stays
the
same,
the
configuration
string,
it
might
be
deployment
or
or
application
specific,
and
so.
A
Okay,
well
I'll
I'll,
write
it
down
so
that
it
and
then
I'll
harangue
everyone
next
week
about
about
the
complexities
there
and
how
we
should
potentially
go
to
solve
them.
So,
okay,
well
moving
on.
A
Well,
yeah,
I'm
sorry,
sorry
shankar!
I
apologize.
I
still
have
a
question
for
you,
so
you
don't
have
to
have
mesh
map
up.
Do
you
there's?
I
want
to
tease
out
that
next
set
of
interactions.
If
if
we
can-
and
it
may
be
very
straightforward
that
like
when
someone
goes
to
apply
a
filter
to
a
deployment,
I
think
you
know
there's
at
least
in
the
istio
environment.
There's,
like
I
don't
know,
10
or
15
different
configuration
items
like
you
know,
and
your
questions
for
users
to
answer.
F
A
Okay,
no
problem
I'll
try
to
bring
it
up,
but
in
the
meantime
yeah
surya
shankar
yeah.
Please
continue.
C
Yeah
so
currently
what's
happening
is
that
when
there's
responses
with
multiple
levels
of
nesting
like
I'm
unable
to
display
it
like,
I
can
use
a
for
loop
to
like
check
if
it's
an
array
and
then
just
like
show
it
again.
But
if
there
is
multiple
level
of
levels,
then
it's
getting
kind
of
difficult
like
I
can
show
you
what
I
mean
if
I
can
like
share
my
screen.
C
Yeah
so
like,
for
example,
this
is
returning
a
machine
pattern
array
and
if
we
go
there
like
technically,
this
is
also
returning
an
array
and
like
I'm,
having
some
troubleshoot
like
getting
that
display
properly,
like
I'm
just
going
like
checking
if
it's
an
array
and
then
just
using
a
simple
photo,
but
we
can't
just
add
infinite
for
loops.
C
For
like
multiple
stuff,
I
guess
and
another
problem
is
that,
like
some
stuff
like
these
areas
like
these
are
not
getting
generated
properly
so
like
the
id
is
coming
kind
of
blank,
the
rest
of
the
stuff
is
kind
of
working.
A
A
Who's
got
a
who
can
help
here.
A
G
Sorry
I
was
distracted.
I
didn't
listen.
C
So,
what's
going
on
is
like
when
we
have
multiple
levels
of
nesting
like
a
look
at
this,
an
area
of
memory
pattern,
and
if
it
also
has
some
other
array
like,
is
there
any
simpler
way
to
like
display
it
like
right
now?
What's
going
on
is
like
I'm
just
checking
if,
if
it's
an
array
and
then
just
looping
through
it,
but
what
if
there
are
like
multiple
levels
like
that
isn't
working
properly?
For
me.
C
C
Like,
for
example,
this
is
an
array,
but
it's
not
getting
like
a
displayed
properly
like
ideally
like
for
swagger
like
you
just
need
to
go
to
its
definition,
the
struct
and
then
add
a
comment
over
there,
but
I
believe,
like
these
are
like
like
libraries,
and
we
can't
just
edit
them.
B
A
Yep
so
like,
when
a
call
is
made
to
get
patterns,
if
there's
multiple
patterns
there'll
be
an
array
of
patterns,
is
that
it.
B
C
Yeah,
like
that'll,
probably
like
fix,
fix
this
case,
but
I'm
sure
like
there
must
be
something
that's
returning
an
array
like
some
other
api
or
something
like
that
will
definitely
make
this
patterns
api
work.
But
what
I'm
sure
like
there'll,
be
something
like
which
returns
an
array
at
one
point.
A
Yep
so
part
of
so
serious
you're
shanker
just
to
give
context.
Part
of
your
goal
is
to
you're
writing
up
api
docs
yeah,
and
so,
as
someone
goes
to
get
it,
this
is
for
the
rest,
api
right
yeah.
This
is
for
the
get
get
request.
C
Like
this
is
just
a
example
like
these
are,
these
are
just
static
stuff.
These
are
not
dynamically
or
generated
like
these
are
just
generated
from
the
swagger
annotations.
C
Like,
for
example,
I
have
like
two
problems
right
now,
like
two
issues
like
first,
is
that
if
we
are
returning
anything
like
an
array
or
a
object
like
this,
then
it's
not
getting
generated
properly
in
the
swagger
documentation
and
it's
returning
an
empty
string.
You
could
see
like
it's
basically
empty.
So
that's
one
problem
and
the
other
is
that
like
like,
if
it's
like
and
like
this
is
kind
of
salt
but
like
if
this
was
an
area
like
then
like,
we
would
get
stuck
like.
C
B
Oh
so,
but
they
are
not
getting
generated
on
the
basically.
Is
it
not
getting
rendered
or.
A
B
It
could
be
that
because
swagger
is
not
generating,
and
it
could
be
that
that
the
uid
n
sql.map
does
not
implement
a
certain
interface
which
swagger
that
courseware
tool
is
looking
for.
That
can
totally
happen
for
sql
dot
map,
because
that's
something
that
we
just
created,
I'm
not
sure
why.
B
That
was
why
that
would
be
happening
for
uid
or
uid,
because
that's
the
properly
maintained
package,
so
most
of
the
other
thing
that
is
happening
is
that
ghostwebber
is
looking
for
something
a
certain
interface
to
marshall
stuff,
and
it's
not
finding
that
for
location
I
mean
secret.map
and
for
ui
dot,
so,
probably
something
we
will
have
to
look
into.
We
we
can't,
even
if
that's
the
issue,
we
can't
fix
it
for
sql
dot
map,
but
not
for
g
y
direction.
B
If
not
right,
and
they
are
not
yet
sure
why
that
would
be
happening,
because
if
that's
the
case,
then
time.
time
shouldn't
work
either,
because
that's
just
another
structure
and
we
have
to
find
example
right
on
top
of
it,
I'm
not
sure
if
it
like.
I
don't
think
that
would
be
present
in
co-compiling,
so
I
I
still
there's
still.
The
hunch
is
that
there
is
certain
interface,
certain
marketing
industries.
That
is
missing.
So
probably
you
can't
create
an
issue
and
look
into
it.
B
Yeah,
actually,
I
think
timer
is
also
okay
yeah.
We
can't
create
an
issue
in
traffic,
probably
not.
A
D
Okay,
so
it's
a
very
small
update,
the
pr
is
still
open.
You
mentioned
that
you
faced
some
issue
while
you
built
it
earlier,
so
I
think
the
reviewers
are
waiting
for
your
update
on
that.
I
didn't
see
any
other
unresolved
stuff
there,
so
that's
pretty
much
it
one
set
is
merged.
I
can
I'll
open
up
here
for
the
patents
and
applications
one
so
that
should
close
the
other
issues.
Then
I'll
work
on
the
special
case.
A
Nice
yeah,
okay,
that's
good!
I
wonder
yeah
I
didn't
realize
there
was
something
you
just
wasn't
gonna
find
me
yeah,
I
okay!
Maybe
then
maybe
it's
my
environments
is
I
get
a
missing
configuration
menu
now,
that's
weird
which
okay
yeah,
which
is
what
you
just
said.
Okay,
all
right,
good,
okay,
fair
enough
good
yeah!
Let's
get
this
one
merged,
then
I
think
we're
good
to
go.
So
that's
yeah!
That's
great!
Then
your
plan
of
action
after
that
sounds
great
as
well.
A
A
Yeah,
you
know
what
petition.
Actually
this
is
a
great.
Let's,
that's
a
great
one
for
the
websites
meeting
we're
over
time
here
and,
and
so
that
has
to
deal
with
the
layer,
5.
io
the
website
itself.
We
should
yeah
good
good
suggestion.
You
just
missed,
or
were
you
on
the
websites
meeting
and
a
couple
hours
ago.
A
Yeah,
let's
get
I'll
point
you
to
those
meeting
minutes,
so
you
can
get
it
tee'd
up
there.
A
Cool
okay,
good,
so
we've
got
at
least
one
item
for
next
week:
megana,
maybe
we'll
have
you'll
be
on
a
roll.
So
you'll
get
like
the
other
two
upload
by
url
for
next
week.
A
A
Am
I
missing
anything
so,
hopefully
the
bulk
delete
gara,
I
would
say
if
we,
if
you
don't
get
it
by
next
week,
let's,
let's
swap
that,
let's
get
you
onto
something
else,
because
because
otherwise,
we'll
I
think
you'll
you'll
be
swimming
in
it.
For
probably
too
long.