►
From YouTube: Cloud Custodian Community Meeting 20230110
Description
Our community meeting is public and we encourage users and contributors of Cloud Custodian to attend! You can find the notes for this meeting on our github repo: https://github.com/orgs/cloud-custodian/discussions
To get an invite to the meeting join the google group and you'll receive one via email: https://groups.google.com/g/cloud-custodian
A
All
right
the
recording
is
on
before
we
get
started.
Please
remember
that
we're
under
the
cncf
code
of
conduct
so
be
excellent
to
each
other,
and
as
always,
we
do
record
these
meetings
and
publish
them
to
YouTube
on
the
regular
welcome.
The
date
is
January
10th.
A
We
are
week
two
of
2023
already
one
percent
through
the
year
great
job.
Everyone-
and
this
is
the
bi-weekly
Cloud
custodian
community
meeting.
It's
our
high
bandwidth,
bi-weekly
meeting.
Where
we
talk
about
all
the
stuff.
That's
happening
around
Cloud
custodian
development
and
kind
of
see
where
we're
where
we're
at.
As
always,
we
have
an
open,
Agenda
I
have
a
few
things
here
already
and
feel
free
to
add
your
own
I'll
go
ahead
and
toss
that
URL
in
the
chat
and
then
we'll
get
started.
A
Hey
all
right
yeah
as
I
was
saying
too
welcome
back
everyone,
the
calendar's
all
updated
for
2023.
You
can
always
find
that
at
Cloud,
custodian.io
calendar
and
that's
all
up
to
date
forever
and
then
I'll
keep
the
notes
there
for
the
slack
invite
page.
If
you
want
to
keep
inviting
people,
I
know
that
we've
seen
more
activity
here
over
the
break.
A
So
if
you
are
looking
for
a
way
to
help
out
the
community,
we
always
have
a
bunch
of
outstanding
questions
and
slack
and
people
looking
for
help
before
we
get
started.
We
have
a
release,
update
and
then
we'll
go
through
the
PRS
and
things
that
have
changed
since
the
last
one
opinions
on
releasing
in
January.
A
Okay,
if
you're
looking
at
the
notes,
the
the
amount
of
PRS
and
issues
is
smaller
than
usual,
but
all
right,
so
we're
just
going
to
crank
it
one
more
time.
Sunny.
You
have
any
outstanding
issues
or
any
color
from
last
time,
because
I
I
don't
think
we
had
a
I
think
we
just
released
a
roll-up
release,
and
that
was
it
for
the
end
of
last
year.
E
No
real
big
updates,
there's
I,
think
there's
some
regressions
in
the
tensa
provider,
but
those
since
my
fix
and
they'll
be
fixing
the
next
release
other
than
that
I.
Don't
know
off
the
top
of
my
head
anywhere
big
PR
instead
of
running
recently,
but
other
people
have
things
that
they
need
to
priorities.
Just
let
me
know
and
I'll
take
a
look
at
them.
A
And
then
I'll
put
on
here
normally
we
would
Target
to
release
around
this
week.
So
I
guess
we'll
just
say:
are
we
shooting
for
next
week?
What?
What
are
you
thinking.
E
Actually
I
think
they're
just
fixes.
Okay,
I,
don't
think
those
resources
were
ending.
Well,
that's
we're
in
the
two
releases
ago,
so.
A
All
right,
we
have
four
and
if
you
have
any
more,
please
feel
free
to
toss
them
in
chat
or
just
let
me
know
the
first
one
Kapil
disabling
trivia
I
saw
this
thread
in
chat
about
some
noise
here.
D
Yeah,
it's
mostly
so
we
use
trivia
to
scan
our
Docker
images
before
they
get
uploaded
and,
unfortunately,
I
have
to
download
a
vulnerability
database,
which
is
has
some
size
and
it
wasn't
being
cached
and
the
way
that
we
so
for
for
right.
Now,
it's
technically
just
disabled
I,
don't
know
that
we're
actually
doing
anything
useful
of
the
data
out
with
the
output
of
it
either.
So
there's
a
separate
question
there.
D
D
I
think
in
general,
like
as
I,
have
been
going
through
some
of
the
CI
stuff,
there's
definitely
a
need
for
a
revamp
but
I
think
on
just
updating
versions
like
GitHub
deprecated,
like
the
the
old,
the
old
way
of
passing
variables
around,
let's
say,
and
so
a
lot
of
our
CI
stuff.
Flags,
there's
also
an
opportunity
to
simplify
our
caching,
because
the
latest
python
major
versions
of
that
action
have
direct
support
for
poetry
and
incorporating
caches
directly
in
so
some
of
the
gymnastics
we're
doing
potentially
simplified.
D
So
I'd
like
to
take
a
look
at
that
just
to
see
if
we
can
simplify
some
stuff
yeah
see
all
those
print
statements
like
that's
no
longer
supported
I
think
is
the
issue.
Yeah.
The
set
output
is
no
longer
supported
in
the
former
correctly
is
was
the
underlying
issue,
the
thing
that
they
found
some
sort
of
security
issue
with
it,
and
so
that
is
verboten
now,
but
most
of
even
if
we
use
remember
our
own
uses
most
of
our
currently
pinned
actions
use
it.
So
just.
F
D
All
that
updated
I
think
is
really
what
I'm
talking
about
it.
Shouldn't
really
be
any
real
actual
changes
is,
is
probably
negligible
and
I
think
there
was
also
something
I
saw
that
I
might
have
already
added
this,
but
GitHub
now
has
support
for
like
Auto,
limiting
concurrency
and
canceling
previous
jobs
that
are
in
Flight
I
forgot
if
I
got
I
already,
I
already
got
added
in
so
just
looking
at
cleaning
some
of
that
stuff
up.
D
So
nothing
particularly
big
just
thought
that
might
be
just
some
just
some
yeah
yeah
there
we
go
March
just
some
housekeeping
on
CI,
because
before
I
was
looking
at.
E
Did
we
already
dropped
37
in
CI
that
we
did.
D
It
is
not
I,
don't
know.
If
it
is,
we
have
we,
we
should
add
311.
Now
that
looks
like
AJ
got
to
the
rate
limiter
thing
and
gcp
that
was
was
causing
us
issues.
There.
D
D
Oh
yeah,
if
you
go
to
ci.yaml
or
go
down
here
to
the
bottom
of
this
well
yeah
go
to
the
bottom
they're
about
not
the
bottom
up,
but
like
the
top
of
the
one
you're
on
there's
a
middle
Matrix
line.
Yeah
though
so
those
are
yep.
Those
are
four
four
versions
of
python
that
we
do
so.
The
mission
would
be
as
part
I
guess,
also
look
at
dropping
through
seven
and
adding
311.
G
D
And
311
has
some
significant
speed
Improvement,
so
it
would
be.
If
that's
all,
to
be
nice.
If
you
have
a
option
and
your
local
deployments
it
is,
it
should
be
a
nice
boost.
Yeah.
A
D
H
D
Only
thing
I
could
find
was
like
an
old
Rel
I
think
around
five
or
six
I,
think,
okay,
and
so
everything
for
all
seven
Adventures,
at
least
on
three
eight.
A
D
B
And
there's
the
there's
the
part
of
this:
if
we're
going
to
change
CI
the
minimum
version
of
CI,
we
would
probably
want
to
update
our
poetry
config,
also
like
the
Pi
Project,
tell
them
all
just
to
make
sure
we
don't.
We
don't
put
a
different
version
in
there
like
a
lower
version
in
there
that
we're
not
testing
and
then
runs
weird
like
on.
D
B
D
Yeah
sounds
good
and
if
people
are
on
older
web
Enterprise,
Linux
posters
I
think
the
general
suggestion
would
be
to
use
the
document
about.
D
Yeah
I
mean
if
they
upgrade
then
they'll
they'll
they'll
see
it
just
in
the
basically
the
breaks
total
non-run
policy
is
probably
the
one
where
I'm
okay
like
from
like.
We
can
only
assume
unattended
systems
on
upgrades,
but
I
don't
know
that
yeah.
It's
a
good
question,
like
one
policy
coming
versus
the
entire
CLI
run.
Failing.
D
A
At
a
minimum,
we'll
start
saying
it
over
and
over
and
over
again
I
think
yeah
all
right
anything
else.
Python
related
release
related
sorry
GitHub
sounds
like
we're
just
talking
about
all
the
GitHub
actions.
Now:
okay,
next
AWS
RDS
cluster
ad
DB
cluster
parameter
filter,
who
this.
G
Were
already
some
initial
review
and
discussion
and
I
think
it's
ready
for
some
final
review.
B
F
B
You
and
something
worked
through
stuff,
yeah
I,
don't
know.
A
G
A
You
know
we
like
the
meeting
is
to
kind
of
help
us
prioritize.
What's
actually
blocking
people
and
then
the
last
one
will
be
add.
Server-Side
filter,
query,
support
to
RDS
and
config
pull
rule.
B
Yeah
and
I'm,
taking
a
look
at
this
one
I
see,
there's
a
merge
conflict
that
looks
like
it
was
just
because
a
couple
different
RDS
tests
added,
that's
okay
and
there
was
a
there
was
a
change
in
December.
I
did
not
see
that
one
so
I'm,
so
yeah
thanks
for
bumping
it
in
the
meeting
the
boom
take
a
look
at
it.
Okay,.
A
So
two
for
two
for
sunny
one
for
AJ.
What
else
do
we
have
here.
A
I,
like
these
short
weeks,
any
issues
that
are
burning
anyone,
oh
I'd,
like
to
bring
this
one
to
everyone's
attention.
It's
just
a
tracking
bug
for
meeting
the
cncf
graduation
criteria
for
this
year.
Now
that
we're
incubated
we've
done
most
of
the
work
already
and
everything
is
in
draft,
so
I
think
for
the
governance
once
I've
linked
to
the
issue
there.
It's
mostly
pulling
that
out
and
proposing
it
and
then
fighting
to
get
that
merged.
A
There
is
a
third
party
security
audit
that
we're
going
to
have
to
go
through
that's
kind
of
run
by
the
cncf
and
I've,
went
ahead
and
filed
that
ticket.
You
might
not
have
access
to
the
cncf's
help
desk,
but
that's
the
number
in
case
any
in
case
anyone
ever
Wonders.
That
is
the
number
that
you'll
need
to
reference.
If,
if
I'm
not
here,
and
then
we've
got
most
everything
else,
the
adopters.md
we
have
the
CI
best
practices,
Badges
and
all
that
stuff.
A
So
that's
more
of
a
dotting
eyes
and
Crossing
tees,
and
that
is
the
I
follow
that
issue
this
morning
and.
B
Other
day,
yeah
that
that's
that
8118,
that's
a
few
down
from
the
issues
open
I
did
not
boom
that
one.
Okay,
all
right,
it
was
worth
it.
So
it
did
I
think,
and
you
had
mentioned
that
it
was
worth
a
ping
to
to
Harish,
which
I
I
did
oh
yeah,
that
that
this
one
Kristen
reported
the
issue
I.
Have
that
the
pr
the
pr
is
where
there's
the
actual
I
I
did
mention
Darren
in
a
review
only
because
I
think
you
into
it
folks
have
done
more.
B
Some
of
the
work
on
webv2,
at
least
using
some
of
the
yv2
stuff.
So
I
didn't
know
if
this
is
something
you
might
have
run
into
where
just
because
you
need
to
pass
that
Scope
when
you're
listing
web
ACLS
for
webv2,
it
was
doing
that
properly
on
poll
mode
policies,
it
wasn't
on
the
cloud
drill
mode,
the
Lambda
mode
deals.
So
this
change
is
supposed
to
to
fix
that,
but
I
I.
D
Had
a
look
at
this
I
was
a
little
bit
the
way
that
get
resources
there
is
implemented
it
like
catches,
everything
and
then
it
client-side
filters.
I
was
wondering
about
just
doing
describe
what
vcl,
instead
of
doing
the
list
would
get
us
just
the
one
thing
that
was
fetched
and
those
check
out
some
of
the
related
filters
and
or
some
of
the
graph
enabled
filters,
and
it
doesn't
look
like
they're,
actually
using
the
get
resources
directly.
D
So
it
would
never
be
an
array
I'm
just
wondering
about,
like
touching
a
bunch
of
data
and
Throwing
It
All
Away,
just
particular
one
versus
just
fetching
that
one
as
well
as
the
consideration
I
was
kind
of
thing,
clear.
B
E
D
I'd
say
probably
under
under
for
eight
to
like
99
of
the
time
for
an
event
base.
It's
just
a
single
one,
otherwise
I
mean
I.
Don't
know
that
org's
wrong
with
all
a
lot
of
web
ACLS
like
more
than
like
a
dozen
and
an
account,
but
I
I
could
be
overthinking.
It
too.
C
It's
interesting:
we
use
quite
a
bit
of
the
cloud
trail
mode
for
the
Waf
V2
policies.
Sorry
I'm
trying
to
understand
what
this
change
helps.
D
D
A
D
A
a
cloud
trail
event
based
policy
on
cloudtrail
for
the
for
the
web
itself,
wasn't
able
to
be
retrieved.
Okay,
yeah.
B
Yeah,
it
was
trying
to
do,
it
would
say
like
well
if
you
did
create
web
ACL,
and
it
would
have
that
on.
It
would
try
to
list
with
the
yarn
and
then
that
that
policy
run
would
fail
and
say,
like
all
right.
You
give
me
an
arm,
but
you
didn't
give
me
scope,
because
we
were
only
handling
that
for
the
the
pull
mode.
B
Sure
but
yeah
that's
good.
It's
a
reasonable
thing.
We
might
not
need
to
pull
everything.
Okay,
I'll
have
to
rethink
that
one.
D
Cool
actually
there's
a
separate,
interesting
question
here
on
something
that's
been
on
Warped
that
I've
been
wondering
about
trying
to
do
something
about
which
is
so
even
in
the
context
of
the
web
right
now.
I
think
it's
good
to
go
to
Regional
by
default.
D
This
query
because,
like
how
do
you
know
if
it's
regional
or
Global?
Well,
you
have
to
look
at
the
event
and
there's
no
real
way
to
specify
that,
instead
of
our
event,
IDs
expression
per
se,
which
is
a
similar
problem
that
happens
with
some
of
the
child
resources
where
we
need
to
explicitly
enable
support
for
it
and
I
was
wondering
about
just
adding
a
method
to
like
a
resource
type
is
like
a
static
method
to
be
able
to
try
to
extract
out
the
ID.
So
we
can
have
some
code.
D
Look
at
it,
the
like
her
child
resources.
You
have
to
get
the
parent
ID
and
the
Child
ID
out
of
the
event.
In
this
case,
you
need
to
potentially
grab
something
arbitrary
out,
it'll,
be
that
if
it's
a
great
way
of
scale
so
being
able
to
pass
that
through
to
some
pipeline
code,
that
I
could
actually
look
at
it
as
opposed
to
having
the
user
haven't
specified.
D
Both
of
them
might
be
easier,
but
it
the
the
challenge,
has
always
been
it's
a
little
bit
dice
to
the
pain
which
JPI
call
you're
doing,
but
if
there's
a
standard
format
per
type,
then
it
might
work.
Thank
you
Mr
or
another,
but
that's
not
directly
related.
It's
just
something.
That's
been
on
the
back
of
my
head
about
something
that
seems
to
come
up.
Come
have
come
up
written
in
the
last
few
months.
A
few
times.
C
George
I
added
something
to
the
chat.
There
was
something
that
just
got
approved
by
AJ.
Thank
you
so
much
for
doing
that
quickly,
AJ,
but
yeah.
We
did
see,
find
this
little
bug
in
redshift
backups.
It's
looking
at
the
end
time.
Instead
of
the
start
time,
it's
a
small
change
but
yeah.
Thank
you
for
getting
that
approved.
C
Are
we
trying
to
get
this
thing
like
merged
in
also
to
before
next
week
or
how
like?
How
does
it
work
if
something
is
approved
like
can
we
just
kind
of
assume
that
that's
gonna
get
merged
in
before
the
release,
or
should
we
usually
call
that
out
anyways?
Do
you
guys
run
like
a
check
before
it
releases
or
anything.
E
H
D
B
E
Crazy
here
is
good.
I
also
try
to
look
at
like
at
least
like
the
last
three
or
four
pages
of
pull
requests
check.
The
stuff
that's
approved
because
sometimes
there's
also
stuff.
That's
been
approved
with,
like
a
note
saying
like.
Oh,
if
you
yeah,
please
modify
this
one
little
thing
before
we
verse
it
in
and
then
the
contributor
you
know
either
forgets
or
doesn't
see
the
message,
and
so
sometimes
stuff
can
hang
like
that.
E
C
I,
don't
think
I've
really
ever
seen
that
happen.
Usually
stuff
does
get
version
stuff,
but
yeah
just
wanted
to
to
ask
basically
just
reach
out
to
to
make
sure
that
if
we
know
there's
like
an
upcoming
release
or
something
that
it's
on
the
agenda
for
merge.
A
Yeah
at
the
risk
of
becoming
really
unpopular
are
we
should
we
investigate
something
like
a
label
workflow
or
some
kind
of
automation
where
the
merge
just
happens
after
you
get
x,
amount
of
approves
or
some
other
I
know.
You
know
different
projects.
Do
it
a
different
way
but,
or
you
know,
is
like
do
we
have
a
way
to
say
all
of
these
have
been
approved
but
not
merged.
E
Yeah
I
mean
like
filtering
on
the
pr
Pages,
like
the
the
filtered
by
sadness,
zippers
yeah,
I
will
say
the
other
situation
like
positive
problems
is
because
we
require
full
requests
to
be
based
on
on
a
master.
Sometimes,
pull
requests
will
not
merge.
Cleanly
requires
the
interviews
together,
update
the
brains,
I
would
say
it's
relatively
rare,
like
most
of
the
time,
there's
a
there's
a
conflict,
it's
probably
dependency
related.
It's
like
a
Pokemon
file
or
byproduct
Tamil.
E
There
are
cases
where
people
are
just
working
on
the
save
file
at
the
same
time
and
one
gets
in
before
the
other
event
as
far
as
more
delicate
handiwork
but
yeah,
those
are
usually
I'd
say
that
that's
probably
the
most
usual
case
of
it's
approved,
but
we
can't
move
through
because
it
was
complex
and
I
mean
as
maintainers.
We
could
typically
fix
it,
but
sometimes
you
know
we
have
other
stuff
that
comes
out,
so
we
we
don't
get
to
it,
but
yeah
for
the
contributor
side.
E
If
there's
a
merge
conflict
like
definitely
appreciate
any.
A
Would
it
help
if
sorry
I'm,
just
kind
of
spitballing
here,
if,
if
people
are
wondering
whether
something's
going
to
make
a
release
or
not
I'm
wondering
if,
like
X
days
before
you
start
cutting
a
release,
you
know,
do
we
start
to
say
stuff
hey
if
you
haven't
approved
PR,
but
it's
not
merged
here's
your
window,
like
does
having
maybe
like
a
dedicated,
merge
window.
Perhaps.
E
Only
in
those
cases
that
have
brought
up
were
like
there's
some
additional
work
that
just
happened
on
the
contribute
side
but
yeah
for
the
most
part
like
if
it's
approved
and
there's
like
some
small
stuff
like
like
I've
gotten
and
fixed
up
for
contributors
like
you
know,
just
because
it's
it's
a
small
thing,
but
it
does
add
up
like
if
it's
like
there's
a
lot
of
birch
conflicts
and
like
trying
to
resolve
each
and
every
one
of
everything
it
just
takes
a
lot.
Sometimes
right
doesn't
get
done
so.
B
Out
of
date,
and
you
need
to
update
and
then
all
the
tests
we
run,
I
was
just
going
to
mention
a
a
comment
in
the
in
the
chat
too
about
you
know
with
the
new
year
new
thoughts
around
like
how
can
we
help
the
CI?
The
dev
experience
like
that
stuff
is.
E
Yeah
I
will
say
the
probably
the
biggest
thing
is
the
fact
that
the
CI
does
take
a
a
decent
amount
of
time.
I
mean
for
for
most
of
the
tests
and
for
most
of
the
operating
systems.
It's
not
it's,
not
bad.
It's
usually
like
Windows
will
be
the
one
straggler
and
then
from
there
you
know,
I
click
update,
Branch
I
come
back
like
five
minutes
later.
E
The
windows
test
is
still
running
and
then
I'll
leave
and
then
I
forget
about
it.
That's
the
typical
use
case.
It's
not
like
I'm
like
actively
ignoring
support
box.
It's
just
like
some
fantastic
finish
and
I
have
to
forget
so
yeah.
Maybe
if
there's
a
way
to
I
mean
I,
don't
think
we
should
make
the
windows
one
optional,
but
I
don't
know
that
that's
a
one
straggler
that
kind
of
makes
a
little
bit
annoying
and
then
I
think
code,
cuff
too,
sometimes
takes
a
bit
of
time
to
report
back.
E
It's
gotten
less
flaky
lately
like
because
I'll
check
in
for
anybody
that
wants
to
know
how
to
check
this.
If
you
check
the
Ubuntu
python,
310
Runner
and
then
check
the
upload
code
coverage
section
in
the
GitHub
actions
window,
you're,
basically
looking
for
a
success
message,
if
you
see
the
success,
the
code
code
stuff
will
come
back
eventually,
but
sometimes
it
fails.
So
you
have
to
like
kick
the
job
off
again
and
that's
a
bit.
E
D
Yeah
I
would
say
generally
CI
goal
like
goal
would
be
like
five
minutes,
because
after
that,
it's
really
too
easy
to
time
out
and
that
causes
delays.
But
I
don't
know.
If
there's
yeah
I
think
we
could
probably
move
code
cut
to
311,
because
311
will
probably
be
faster.
D
Yeah,
there's
also
options
on
getting
rid
of
code
I'm
just
doing
it
in
line
I
I've
used
things
like
that.
I'd
like
The
annotation
stuff,
can
be
nice
too
one
other
like
when
I
think
about
developer
experience,
though,
like
so
see
the
speed
of
CI
is
definitely
the
a
huge
thing
here.
D
I
think,
though,
is
being
able
to
just
install
and
work
in
a
single
provider,
because
most
people
don't
care
about
running
all
the
providers,
and
so
that's
just
that's
the
job
for
CI
so
potentially
like
adding
to
our
docs
on
how
to
just
get
a
Dev
environment
set
up
for
a
single
provider,
because
you
know
I,
don't
care
about
arrows
updating
my
my
Dev
environment
today
I
ran
into
some
dependency
and
Pi
crypto,
dome,
which
I've
never
heard
of
before
and
I
was
like.
D
It
wasn't
compiling
on
my
Mac
and
so
I
just
went
down
this
rabbit
hole
and
it
was
for
you
know
somehow
I,
don't
think
it
was
intense
provider,
but
that's
where
anyways
you
know.
If
you
just
want
to
work
on
a
single
thing,
you
can
get
it
done
fast
like
we
could
probably
update
our
developer
docs
on
like
how
to
how
to
just
work
on
a
single
provider,
because
otherwise.
G
D
Is
like
a
gigabyte
plus
and
then
the
other
thing
I
noticed
was
that
R,
which
has
been
warning
me
in
the
background,
is
our
repo
size
is
getting
really
big?
It's
like
I,
don't
know
450
Megs,.
G
D
I
I
don't
know
if
something's
happened
when
we
switched
out,
if
it's
literally
just
that's,
how
much
data
we
have
or
if
there's
something
wrong
with
the
GitHub
Pages
Branch
updates
that
are
causing
things
there.
I
was
just
a
little
bit
concerned
about
repository
sets.
I,
don't
have
any
any
fact-based
Awareness
on
what
that
is
it
just
it's
something
that's
been
in.
The
back
of
my
head
is
that's
getting
that
feels
a
little
too
large
I.
A
D
That's
what
I
have
yeah
if
you
run
okay.
D
A
A
D
B
D
And
MacBook,
it's
basically
just
one
of
the
latest
version
that
might
give
us
an
extra
minute
or
two
and
speed
improvements
and
then
switching
to
maybe
oh,
that
switching
action
is
going
to
help
at
all,
but
potentially
but
yeah
it
alternates.
Based
on
how
much
you
know
how
much
stuff
we
have,
how
much
how
many
tests
we
have.
But
you
know
we
need
to
figure
out
ways
to
take
the
tester
faster
to
get
the
SLA.
Then
that's
the
way
to
do.
E
B
B
Stuff
to
investigate
and
stuff
we've
talked
about
before
I
think
there
are
some
other
things
we
can
do.
That'll
make
things
more
Dev
friendly,
even
if
it
doesn't
directly
impact
the
CI
time
like
Rick
appeal
is
talking
about
with
only
working
on
certain
providers.
We
can
handle
some
of
that
with
with
defining
different
dependency
groups,
where
you
say
like
all
right.
If
I'm
working
on
gcp
install
like
C7
and
gcp
sure,
and
things
like
that,
and
that
might
be
nice.
D
F
A
Let
me
get
it
foreign.
B
F
D
Yeah,
probably
just
need
a
a
I,
see
it
I
think
I
understand
it.
D
Summary
yeah,
it's
basically
like
we,
we
try
to
the
related
filter,
tries
to
use,
get
resources
and
get
resources
tries
to
use
server-side
filters
when
they're
employed,
but
when
you
hand
it
a
list,
they
can't
do
that.
So
we
just
put
them
into
to
to
either
handle
that
in
in
the
get
resources
implementation
there
yeah
for
that
Source.
But
yeah,
that's
right,
but
but
no
thanks
for
doing
it.