►
From YouTube: Gateway API GAMMA Meeting for 20230620
Description
Gateway API GAMMA Meeting for 20230620
A
Wow,
let
me
do
that.
Evidently,
because
Rob
made
everybody
a
co-host,
yeah
well
Rob,
do
you
want
to
take
it
away
or
shall
I.
B
Oh
I've
got
no
preference,
I
guess
I
have
the
first
few
couple
things
on
the
agenda.
So
let
me
let
me
start
with.
A
B
Cool
all
right,
so,
first
up
here,
a
couple
maintainer
updates,
first
off
as
I
think
we've
discussed
previously.
Mike
is
stepping
down
I.
Think
huge
thanks
to
everything.
You've
done
to
get
this
Community
up
and
running
and
then
I
think
effective.
Today
is
what
we
said:
Flynn
is
now
a
gamma
lead,
so
welcome,
and
also
huge
thanks,
because
I
know
you've
been
very
involved
for
a
long
time
so
yeah.
This
is
a
great
transition
into
two
very,
very
instrumental
people,
so
yeah.
C
So
yeah,
thank
you
so
much
appreciate
it.
Congratulations.
A
Sorry
to
see
you
go
you're,
leaving
a
big
pair
of
shoes
to
fill
as
I
said,
an
email,
so
I'm,
not
sure
if
it's
congratulations
or
but.
C
B
Awesome
cool
and
then
I
also
wanted
to
highlight
that
Leora
is
becoming
a
conformance,
reviewer
and
Ingress.
The
Gateway
reviewer,
so
he's
been
taking
on
increasingly
more
and
more
in
those
projects.
I'll
just
thanks
thanks
to
him
for
for
leading.
So
much
of
this
work
and
I'm
excited
to
see
Ingress
get
the
Gateway
moving
again.
So
yeah
really
really
excited
about
this,
and
just
shout
out
to
anyone
else
on
this
call
or
watching
later
we
do
have
a
contributor
ladder.
B
B
Next
up
but
oh
go
ahead.
No.
A
B
All
right,
cool
yeah,
the
next
thing
I
had
was
the
080
Milestone
I.
Think
we've
already
discussed
this
last
time.
I
just
wanted
to
make
sure
we
we
could
Circle
back
around
because
realistically
we're
trying
to
get
080
and
1.0
out
in
between
now
and
October.
Basically,
so
it's
going
to
be
a
pretty
accelerated
stretch
here
and
080
is
primarily
just
gamma's
release.
B
So
this
is
you
know
this
is
a
broader
meeting,
but
the
idea
is
to
cover
both
topics,
and
so,
as
far
as
I
know,
based
on
our
last
meeting
we're
really
just
meeting
missing
Docs,
which
Flynn
is
going
to
work
on.
We
said
we're
gonna
ping
you
this
week,
and
so
this
is
the
ping
yeah.
A
C
A
Think
I
can
I
think
I
can
tackle
Doc's
stuff
I
will
probably.
A
Yeah,
let
me
let
me
look
over
things
a
bit
more
and
then
I'll,
probably
ping
a
couple
of
people
with
with
some
suggestions
there
and
then
we'll.
C
A
Great
skipping
ahead,
a
little
bit
I
believe
I
will
also
be
able
to
handle
the
blog
post
David
from
hash
record
reached
out
already.
It
would
be
lovely
indeed
to
have
other
co-authors.
So
if.
C
Yeah
someone
wants
to
reach
out
to
Michael
over
at
Kuma.
That
would
probably
be
good.
C
One
wants
to
reach
out
to
that
would
probably
be
good
too
yeah.
A
Yeah
so
yeah
just
ping
me
on
slack
grabbing
email,
whatever
we'll
we'll
go
from
there
and
hopefully
be
able
to
handle
that
asynchronously
for
a
bit.
B
Cool
and
then
the
the
other
bit
was
API
spec
updates
themselves.
So
I
think
the
only
thing
that
we
need
here
is
something
like
on
parent
ref.
That
says,
oh
by
the
way,
this
can
also
be
attached
to
a
service
and
that's
what
it
means
in
this
context,
maybe
there's
some
other
gamma
specific
guidance
that
I
haven't
thought
of.
Yet
so
definitely
ideas
welcome
there.
B
But
you
know
I've
been
thinking
again
about
this
kind
of
aggressive
timeline
that
we
have
and
I
think.
That
means
that
you
know
I,
guess
we've
kind
of
discussed
this
before,
but
we
really
should
try
and
have
everything
wrapped
up
by
the
end
of
this
month,
so
we
can
at
least
hand
it
off
to
reviewers.
B
You
know
it
does
take
some
time
for
Tim
and
Cal,
depending
on
their
availability,
to
actually
get
through
this,
but
I'm
hoping
this
is
a
small
enough
release
that
we
can
just
actually
get
it
out
sometime
in
July,
so
yeah
that
that's
what
I've
been
thinking
as
far
as
timeline,
but
I'm
open
to
comments
discussion.
Whatever
on
this.
C
High
availability
is
opened
up
a
little
bit
recently,
so
yeah
I
I
think
I'll
be
able
to
stuff
wrapped
up.
Hopefully
this
week
see
if
there's
anything
else
that
feels
relevant,
but
also,
if
anybody
else
notices,
please
feel
free
to
pay
me.
B
Yeah
thanks
so
much
for
for
taking
this
one
on
I,
really
appreciate
it:
okay,
well,
I'll
hand
it
back
to
Flynn
or
leor,
or
whoever
to
take
us
to
our
next
next.
One.
D
Thanks
yeah
I
wanted
to
bring
up
this
product
called
PRS,
where
I
think
we
we
discussed
on
this
before
where
this
is
the
community.
Although
it's
kubernetes
like
this,
is
the
community
to
review
them.
D
I
think
this
one
is
laid
out
for
a
while,
and
it
has
some
some
accomplishments
and
improvements
here
where
we
both
think
that
the
web,
sockets
and
websocket
secure
should
be
added,
I.
Think
Dave's
here
as
well,
we'll
be
chatting
as
well
about
it,
but
yeah
I
just
wanted
to
know
what
are
the
blockers
for
adding
it
and
if
someone
else
thinks
differently.
A
D
C
I
think
it's
widely
beneficial
I
think
there's.
Maybe
it
was
some
questions
about
websites
to
help
previously
around,
like
the
upgrade
handling
and
how
that
works.
I
think
that's
more
implementation
details
rather
than
someone
that
should
block
this.
This
seems
like
a
good
addition
to
me.
E
Oh
I
added
to
the
issue
here:
if
you
look
at
like
Gip
119,
I
kind
of
added
a
table
based
on
next
feedback
about
app
protocol,
Service
Port
protocol
and
also
the
route
type
and
weather
like
if
those,
if
those
combinations,
how
do
they
match
up
hp3,
makes
things
very
interesting
because
it
changes
the
underlying
protocol
to
be
over
UDP.
So
then,
what
raises
the
question?
Sort
of
for
these
protocols
is
like
Hey
how
technically
ws
and
WSS
are
strictly
like
hp1
rfcs.
E
E
Does
that
expand
then
into
like
Leo's
looking
for
use
cases
for
like
multiple
app
protocols
and
a
list,
or
do
we
change
sort
of
the
websocket
constants
in
this
PR
to
include
like
WS
2?
That's
three:
do
what
I
mean
like
that
it'd
be
nice
to
kind
of
like
have
a
plan
and
then
make
the
necessary
changes.
D
Is
this
also
following
the
LPN.
F
Issue
about
there's
a
separate
issue
about
IOP
and
negotiate
discussion
actually
about
aop
and
negotiation
and
using
if
we
can
use
that
for
as
another
component
on
top
of
Sni
for
TLS
route,
so
I
think
yeah,
I,
I
sort
of
said
on
that
one.
That
seems
reasonable,
that
you
might
use
the
the
fact
that
you
want
alpn
negotiation
or
the
results
of
the
aopa
negotiation
like
you
can't
really
have
the
results
of
the
aopia
negotiation
because
it
has
to
be
negotiated,
see
it's
not
available
for
routing.
F
So,
like
yeah
like
how
much
can
we
do
that,
I
don't
know
like,
but
but
we
should
be
able
to
do
something
with
Alp
and
negotiation
right
like
in
the
same
the
same
yeah.
Sorry,
no,
no
go
ahead
and
I
was
gonna.
Tell
you
like
to
be
honest.
I
I
do
think
that
it's
probably
worth
putting
I
think
we
can
put
a
sort
of
star
on
the
on
the
it's
okay
to
put
unanswered
questions
in
again,
it's
okay,
to
put
you
know,
future
work
in
a
gap.
F
It's
okay
to
say:
HTTP
3
is
unknown
at
the
moment
this.
What
we're
doing
here
explicitly
does
not
say
anything
about
HTTP
3.
If
you
want
to
say
something
about
HTTP
3
come
back
and
do
another
PR,
you
know
or
something
like
that,
like
I,
think
I
think
it's
worth
if
we
need
to
put
that
in
the
out
of
scope,
bits
at
the
top,
that's
fine,
but
like
out
of
scope
for
now
or
proposed
future
working.
You
know
like
we
have
a
couple
of
examples
and
saying
it's
not
that
we'll
never
do
this.
F
It's
that
we're
not
doing
it
now,
because
it's
hard
and
we'll
come
back
to
it.
You
know
I
think
that
a
lot
of
stuff
about
HTTP
3
is
is
like
that
I
mean
I,
almost
think
that
we
might
need
another
protocol
value
on
The
Listener.
To
say
this
listener
is
a
HTTP
31,
because
that
then
has
implications.
F
You
know,
but
I
haven't
thought
of
it
enough
about
it
to
be
sure
that
that's
what
I
really
think
we
should
do,
but
so
like,
like
I,
think
that
that
all
of
that
sort
of
really
puts
a
lot
of
this
into
the
you
know.
You
know
into
the
200
right
now
basket
for
the
you
know:
let's
get,
let's
get
the
one,
the
bits
that
we
want
for
the
things
now
in
and
markdown
you
know
has
to
be
three
things
as
we
know
it's
important.
F
We
will
get
to
it,
but
not
right
now,
because
otherwise
we're
going
to
end
up
stuck
having
to
boil
the
ocean
of
HTTP
3
support
to
be
able
to
do
to
be
able
to
add
functions
that
are
very
useful
for
HTTP
one
and
two
right
now.
D
By
the
way
is
HTTP
3,
Iana
protocol
or
no.
F
A
A
Yeah,
it's
like
you
said
that
that
is
usually
a
point
at
which
you
can
consider
it.
Hard
currency,
as
you
put
it
H3
in
a
lot
of
ways,
feels
to
me
like
something
more
immediately
relevant
to
the
Ingress
side
of
the
Gateway
API,
as
opposed
to
the
gamma
side
of
the
Gateway
API
I
could
imagine
that
changing
in
the
future,
but
I
mean
probably
not
so
much
right.
Now,
it's
most
of
the
benefit.
G
A
Yeah
yeah
and
yeah.
D
D
The
time
box,
yet
because
there
are
many
more
things
on
the
agenda,
so
I
think
Rob
like
or
anyone
else
I
can
get
your
help
helping
this
one
merge,
because
if
there
is
a
consensus
here,
but
also
I,
think
question
C
remains
like
also
with
HTTP
2,
like
how
do
we
express
if
we
want
to
express,
like
you
know,
WSS
on
HTTP,
2
or
WS
or
like,
for
example,
any
protocol
I
was
thinking
to
like
just
leave
it
as
like,
like
what's
the
LPN
negotiation
is
and
like
the
community
is
gonna
encourage
using
alpn,
but
they
know
there
are
some
implementations
to
them.
B
Yeah
I
I
haven't
said
much
here,
but
this
this
feels
websocket
feels
maybe
slightly
different
in
that
I
I
feel
like
it's
almost
always
done
via
negotiation
or
via
upgrade
right
so
like
what
does
it
mean
like
the
way?
I
would
interpret
this
as
an
app
protocol.
Is
that
we're
allowing
the
upgrade
to
occur
like
this?
This
endpoint
supports
that
not
necessarily
that
all
connections
should
use
websocket
like
and
and
so
that
feels
just
a
little
bit
different
than
what
we
have
so
far
in
app
protocol.
A
Do
you
think,
there's
a
better
place
to
to
talk
about
websockets
I
mean
sorry
I.
Do
you
think
there
is
a
better
way
to
indicate
that
websockets
should
be
expected
on
a
given?
You
know
a
given
place.
B
E
B
Okay,
okay
to
me,
I
feel
like
this.
This
specific
thing
seems
to
indicate
if
we
move
forward
with
this
as
an
app
protocol,
we
almost
certainly
need
to
support
a
list
of
app
protocols
and
if
we
don't
want
to
do
that,
we
should
evaluate
others,
but
it
does
seem
like
all
the
momentum
is
towards
this
change,
plus
the
list
of
app
protocols,
but
if
anyone
has
concerns
raise
them
now,
because
I
think
this
just
you
know,
pushes
us
further
further
in
that
direction.
A
B
There's
also
some
of
the
the
pain
that
we've
we've
found
along
this
way.
Is
there
sometimes
implementation
specific
variations
of
protocols
like
only
supporting
HTTP
2
over
TLS
and
not
clear
text,
and
not
you
know,
there
are
some
weird
variations
out
there
that
can
be
expressed.
This
way
is
this
is
one
of
the
things
I
support.
A
H
A
This
is
not
something
that
I
think
is
going
to
be
a
problem
or
inside
the
cluster,
so
there
are
ways
that
I
kind
of
feel
like
it
makes
more
sense
to
attach
this
sort
of
thing
to
an
HTTP
route
than
anything
else
or
some
other
kind
of
a
route
where
you
get
to
say,
here's
a
particular
path,
and
this
is
what
we
expect
on
this
particular
path,
but
I
also
suspect
that
people
like
Laura
and
Dave
have
done
a
lot
more
thinking
about
this
than
I
have
so
I
am
I'm,
definitely
willing
to
be
convinced
on
this
one.
E
I
just
want
to
answer
Candace's
question
about
like
why
not
make
a
route
for
protocol
and
I
kind
of
have
a
distinction
where
I
view
the
existing
Gateway
routes
to
be
external
facing
and
not
sort
of
like
the
Upstream
API,
because,
as
an
example
for
key
native
people
come
out
with,
like
newer
visions
of
their
software.
Let's
say
they
upgrade
them
from
like
an
HP
endpoint
to
a
grpc
endpoint
I.
E
Don't
want
to
have
to
tear
down
the
Ingress
resources
in
order
to
then
be
able
to
send
traffic
to
the
back
end
on
upgrade.
E
E
So
how
you
would
do
your
like
upgrading
to
grpcs?
You
would
just
specify
the
HTC
string
kind
of
like
how
we
see
who
does
it
on
the
on
the
port
name
actually
or
the
container
Port
I
think
is
How
We
Do
It,
but
we
don't
touch
the
Ingress,
because
we
know
that
these
Protocols
are
sort
of
compatible
right.
F
Yeah
so
yeah
the
the
way
that
I
would
answer
that
same
question
is
we're
not
types
of
Base.
We
have
route
types
based
on
how
the
request
should
be
routed
when
it
reaches
the
Gateway.
So
that's
the
that's.
The
sort
of
determinant
that
we
need
to
use
about
whether
or
not
we
create
a
new
route
type
is:
is
this
new
route
type
relevant
for
other?
F
How
the
things
should
be
routed
when
it
reaches
the
Gateway,
not
like
the
capabilities
in
the
back
end
right
like
or
what
what
the
protocol
should
be
when
you
talk
from
the
gateway
to
the
back
end
right
like
that
was
one
of
the
reasons
we
I
wanted
to
do.
The
back
end,
property
big
chunky
version
is
earlier
was
that
you
could
put
a
lot
of
that
this.
That
would
be
the
bugger
for
this,
but
in
terms
of
like
what
would
what
would
he
cons?
F
What
would
I
personally
consider
a
new
routers
for
I
would
actually
consider
talking
about
a
HTTP
throughout
you
know
in
that
it
might
be
a
useful
sort
of
compare
and
contrast
between
like
if
we
did
that,
would
there
be
any
useful
gain,
as
opposed
to
just
having
a
HTTP
route
and
a
pro
and
a
protocol
type
on
the
listener.
That
says
you
should
expect
traffic
to
this
listener
to
be
HTTP
3,
because
that
implies
UDP
and
blah
blah,
but
but
the
you
go,
the
and
so
yeah
I
think
I.
F
F
It's
changing
the
transport
details,
and
so,
like
you,
we
really
should
be
trying
to
con
to
support
HTTP
3
at
a
transport
layer
rather
than
you
know,
rather
than
at
sort
of
the
the
higher
level
application
layer,
stuff,
yeah
layer,
four
to
five,
not
that
seven
to
usually
slightly
forward.
Oh,
it's
like
a
model
yeah,
so
I
think
I
just
wanted
to
make
clear
like
how
I
think
of
that
and
I.
Think
that's
that's
one
of
the
things
we
explained
really
poorly
about
the
API.
At
the
moment.
F
It
is
online
tutor
list
of
things
to
fix
before
we
go.
Ga
is
to
have
better
explanations
about
why
why
we
have
multiple
route
types
and
and
what
they're
for?
But
the
important
thing
to
remember
is
that
right
now,
the
way
so
the
way
I
think
of
it
and
the
way
I
tell
it
to
people.
Is
that
we're
out
types
of
bay
based
off
the
information
you
use
for
routing
discrimination.
F
Tls
routes
are
Sni
and
that's
pretty
much
it
so
when
we
started
talking
about
like
do,
we
have
traffic
where
we,
you
know,
terminate
TLS
with
TLS
routes,
it's
like
you
can
as
long
as
you're,
only
using
the
Sni
as
the
routing
as
the
routing
discriminator,
because
then
it
doesn't
break
violate
the
TLs
contract,
so
yeah
I
think
those
that
sort
of
I
wanted
to
sort
of
have
recorded
a
response
to
you
know
why
would
we
consider
a
new
wrap
up
thanks
for
the
question
Candice.
A
B
Yeah,
sorry
yeah,
that
that
is
a
perfect
description.
I
I
just
raised
my
hand,
because
I
realized
that
we
were
just
finishing
this
conversation
up
and
then
I
I
opened
it
right
back
up.
So
sorry
but
I
know
we
do
need
to
time
box
this,
and
we
have
many
very
interesting
items
on
the
agenda
so
yeah.
Maybe
we
can
continue
this
discussion
offline
on
this
one.
A
D
Yeah
yeah
I'll
try
to
follow
up
with
maybe
another
discussion
on
this
things
just
finished,
but
yeah
I'll
expect
to
see
some
traction
because
people
have
opinions
so
yeah.
The
next
one
is
about
like
enhancing
the
Gateway
class
status
API.
So
both
in
terms
of
reporting
conformance
features,
we
seems
to
be
not
controversial.
D
Actually
we
didn't
get
a
lot
of
responses,
but
the
second
one
which
is
seems
to
be
more
complicated,
is
about
like
policies
or
extensions
support,
say
I
wanted
to
get
like
a
little
bit
like
more
thoughts
about
it
like
what
people
think
for
conforming
feature
is
pretty
simple.
We
want
to
be
able
to
visualize
for
people
like
for
users
what
conformance
features
their
implementation
support,
but
for
policies
it's
a
bit
more
complicated.
Actually
then,
I
I
tagged
you
in
this
one
yeah.
D
A
F
Yeah
sorry
yeah
yeah,
so
sorry,
the
idea
here
is
for
me
is
to
make
is
to
make
this
thing,
be
the
here's
a
list
of
relevant
cids
that
you
might
want
to
know
about
that
are
relevant
for
this
implementation
right
like
that's.
Why
it's
in
status?
It's
not
about
you
know
declaring
something
about
the
cities.
It
is
about.
The
implementation
telling
you
these
are.
These
are
crds
that
you
might
want
to
use
right.
These
are
the
extra
cids
that
are
not
included
in
the
base.
Gateway
API
set
that
you
might
want
to
support
So.
F
Thus,
any
any
crd
that
you
used
for
a
parameters
ref
would
should
be
listed
in
this
extension.
You
know
list
of
extension,
resources,
the
you
know.
If
you
used
a
config
map,
then
that's
not
I,
think
that's
not
a
non-standard
resource,
so
you
wouldn't
need
to
use
that
right
like
so,
but
the
intent
here
is
to
give
you
a
way
that
you
can
be
like
something
weird
is
going
on
with
this
implementation.
Is
there
config
relevant
that
I
had
that
I?
F
Don't
know
about
cubecat
or
get
go
to
this
thing:
API,
Group,
dot,
resource
name?
Okay,
Dash
a
now
I
can
see
all
of
them
in
the
cluster
that
I
have
the
rights
to
see
whatever,
but
like
the
the
intent
here
is
to
sort
of
take
some
of
the
things
that
you
know
clean
with
good
reason
was
complaining
about
in
the
parable
that
you
know
about
discoverability
and
help
people
discover
things
about
the
implementation
that
they're
using.
F
If
they
don't
know
them
already
right,
it
says
you're
having
to
go
to
the
implementation
stocks
website
and
be
like
what
resources
can
I
use.
You
know
do
I
support.
Does
it
support?
Does
it
have
like
a
bunch
of
extension,
custom
extension
filters
right
now,
for
example,
Gateway
has
a
bunch
of
like
pretty
nifty.
F
You
know
oil
filters
and
stuff
like
that,
but
there's
no
way
for
you
to
know
that
from
the
cluster
itself
you
have
to
go
to
the
onboard
Gateway
website
and
look
up
exactly
what
they
are
and
how
to
use
them
and
all
that
sort
of
stuff.
At
least
this
way
you
can
be
like.
F
Is
there
something
relevant
I?
Don't
know
right
like
in
that's
that's
my
intent
here
is
just
it
should
that's.
Why
it
should
literally
be
you
know
a
you
know,
a
list
of
you
know,
group
kind
or
and
or
resource,
and
maybe
an
optional,
like
purpose
that
like
tells
you
something
about
what
it's
for
yeah
anyway
see
I'd,
probably
talked
enough
Rob.
Your
next.
B
Yeah,
no
big,
plus
one
on
everything
you're
suggesting
there
I
I,
think
it.
It
makes
sense
I,
like
the
idea
that
you
have
of
some
kind
of
purpose
category
whatever
to
to
describe
the
what
that
specific
kind
is
being
used
for
yeah
I.
Think
this.
This
is
great
yeah
would
love
to
move
forward
with
it.
B
Yeah,
I,
I,
guess
really
just
to
to
Echo
what
you're
saying
here,
and
maybe
since
this
is
how
I
shouldn't
open
this
can
of
forms,
but
just
I'll
throw
this
out
here,
because
this
is
very
much
because
it's
a
Gateway
class
centered.
This
is
something
that
that
solves
Ingress
implementation
issues
and
leaves
gamma
completely
out
in
the
current
form
of
the
world
so
longer
term.
We
probably
need
some
kind
of
solution
for
gamma
yeah.
B
Yeah
and
I
think
this,
combined
with
the
other
thing
Leora
had
mentioned
on
on
Gateway
class
status.
The
what
features
do
you
support
is
also
very
relevant
for
any
gamma
implementation
because
I
imagine
it's
not
going
to
be
100
percent
yeah
cool
okay,
I'll
hand
it
over
to
leor.
D
All
right,
I
have
this,
and
this
is
its
purpose
like
in
terms
of
flow.
What
you
have
in
mind.
F
So,
in
my
mind,
in
my
mind,
this
is
just
a
it's:
a
must
requirement
that
we
can't
conform
this
test
for
right,
like
it's
just
a
this
is
going
to
be
an
honor
System
thing.
This
is
going
to
be
a
you
know.
You
must
do
this
and
we
can
all
sort
of
be
like
raise
eyebrows
with
people
if
they
don't
and
that's
pretty
much
it
right.
Like
you
know,
it's
like
it's
more
intended,
as
you
know,
sort
of
very
strongly
worded
guidance
for
implementations.
F
You
know
to
say
you
have
to
do
this
so
that
people
will
be
able
to
find
all
the
bits
that
you're
making
available
in
the
same
way
that
the
for
conformance
features
it's
a
little
better,
because
I
would
hope
that
in
the
future,
if
we
do
the
conformance
profile
stuff
correctly,
we
should
actually
be
able
to
change
the
performance
test
such
that
you
point
them
at
a
Gateway
class
and
it
picks
which
conformance
test
to
run
because
you've
told
it
which
features
you
support
and
does
those
and
then
tells
you
how
you
did
so
that
so
that
you
don't
need
to
sort
of
specify
as
part
of
your
test,
harness
like
right.
F
What
what
features
doing
you
make
your
your
thing,
publish
that
to
its
Gateway
class,
and
then
the
tests
can
understand
that
and
they
can
understand
what
they
should
be
doing
and
do
it.
And
that
way,
if
you
add
a
new
feature,
all
you've
got
to
do
is
publish
it
as
a
Gateway
class
and
boom
your
testing
infrastructure.
Now,
we'll
we'll
run
the
components
test
for
you,
and
so
that
you
know.
F
You
know,
ideally,
people
should
be
able
to
write
tooling,
to
look
at
Gateway
classes
in
the
cluster
and
be
like
hey
I
know
that
controller
name
screen
that
is
still
in
I
know
that
control
on
answering
that's
Contour,
you
know,
and
they
can
then
I
can
make
and-
and
they
can
say,
okay
and
this
Contour
installation
supports
these
additional.
You
know
crds
and
those
are
the
ones
that
we
should
be
looking
for
in
the
cluster
or
something
like
that
yeah.
F
It's
both
useful
for
users
in
the
sense
that
you
know
that
they
need
to
type
Q
Kettle,
but
it
also
gives
the
possibility
for
people
to
write
tooling
to
help
you
do
that
automatically
and
again
Flynn
I'm
sort
of
Leaning
heavily
on.
You
know
you.
What
would
Flynn
do
here
just
to
be?
Like
you
know,
users,
you
know
like
we
want
users
to
be
able
to
use
this
thing
without
hate
without
hating
us,
as
the
people
who
wrote
the
spec
right,
like
you
know,
we
want
people
to
I
think
should
be
discoverable.
F
A
F
Yes,
so
I
think
the
yeah
like
I,
think
the,
but
the
key
part
for
the
for
the
extension
ballistic
extensions
is
that's
not
going
to
be
tool
impossible.
That's
going
to
be
an
online
system
and
and
effectively
extremely
strongly
worded
guidance
rather
than
you
know
something
that
we
can
actually
check
in
any
way
other
than
having
a
person
go
and
actually
install
it
and
be
like
hey.
You
forgot
a
yeah,
you
forgot
an
extension
or
something
like
that.
F
A
Maybe
it
was
a
shame
on
that
particular
particularly
colorful
not
occur,
but
anything
else
on
this
one
just
didn't
in
interest
of
time
it.
It
definitely
seems,
like
everybody,
likes
the
idea
of
extending
Gateway
class
to
do
this
sort
of
thing
and
then
finding
a
way
for
gamma
to
take
advantage
of
it,
I
I
kind
of
feel,
like
Nick's
suggestion
of
you
know
just
listening
groups
and
kinds
or
groups
and
resources,
and
that
makes
a
lot
of
sense
to
me.
C
F
Yep
so
I
I
finally
opened
APR
to
rolling
the
changes
that
we
talked
about.
Oh
people
have
commented
that
I
hadn't
noticed
sorry,
I
haven't
read
those
yet
but
yeah.
F
The
intent
here
is
to
sort
of
yeah
fold
in
the
fold
in
parallel
that
that
Flynn
wrote
as
part
of
the
get
2014
and
to
to
sort
of
talk
about
what
that
leads
us
to
conclude
how
you
know
how
possible
ways
we
could
things
we
need
to
think
about
when
trying
to
solve
this
discoverability
problem
and
then
include
sort
of
proposed
Solutions
now,
a
couple
things
that
that
are
important
here
is,
if
you
keep
going
down
if
you're
driving,
yeah
cool.
These
are
sort
of.
F
This
is
listing
the
problems
this
part
and
then,
if
you
keep
going
down,
then
I
do
actually
include
like
here
are
the
things
that
we
should
have
the
solutions.
However,
there's
a
solution:
cookbook
there
are
some
things
where
I'm
like
I,
don't
think
there'll
be
any
objection
here.
You
know
so
some
of
it
is
already
is
already
added.
You
could
just
go
back
up
a
little
bit
wrong.
I
just
want
to
hit
the
first
one.
The
one
about
the
label,
so
the
standard
label
on
crd
objects,
is
actually
supposed
to
be.
F
You
know
already
done,
and
that
is
actually
part
of
the
spec
already
it
was
just.
You
know
a
little
bit
hard
to
find
to
sort
of
know.
You
look
yeah
yeah,
so
this
one
calls
that
out
and
then
the
other
thing
that
I
wanted
to
call
out
here
is
that
I've
added
this
like
design
consideration
section
to
each
of
these
solution,
things
to
sort
of
say
a
little
bit
about
hey.
F
If
you're
going
to
do
this,
you
need
to
be
careful
about
apis
of
a
fan
out,
or
you
know
this
seems
like
a
no-brainer
or
some.
You
know
some
other
sort
of
concept
here.
So
you
know
I've
got
like
this.
The
idea
of
having
a
standard
status,
condition
I,
don't
think
that
that
one
seems
very
would
be
very
controversial
to
me
like
if
something
is
affected
by
a
policy
at
all.
F
Instead,
you
know
that
is
just
an
escape
hatch
for
for
core
things
that
we
can't
update
and
won't
be
updated
to
have
status
and
then
so
yeah
and
then,
if
you
keep
going
down,
there's
a
few
more
ones
further
down
where
they
are
sort
of
you
know,
I've
put
this
solution
is
in
progress
and
so
is
not
binding.
F
Yet
you
know
this
is
that's
the
you
know,
extension
types
listing
that
we
were
literally
just
talking
about,
and
then
you
know
having
a
standard
status
stanza
that
lists
that
lists
all
of
the
all
of
the
policies
that
are,
in
effect
on
a
particular
object.
F
If
we
control
the
status
and
can
do
that,
you
know
that
has
big
fan
out
problems,
yeah
and
so
I've
got
an
example
there
of
what
it
might
look
like,
but
there's
a
bunch
of
stuff
here
saying
you
know
this
is
experimental,
don't
use
this,
yet
you
just
think
about
it.
F
When
you
talk
about
it
in
this
in
this,
in
this
Gap,
and
so
thank
you,
I
tried
to
sort
of
be
very
clear
here
about
which
bits
are
in
progress
and
which
bits
aren't,
and
you
know
some
of
these
I
expect
that
we
will
talk
about
as
part
of
the
Gap
and
then
and
then
sort
of
either
remove
them
from
the
gap
or
mark
them
as
like.
We
didn't
put
them
in
Alternatives
considered
or
something
like
that.
F
So
the
intent
here
is
to
sort
of
make
it
very
clear
when
you're
reviewing
this,
which
bits
are
sort
of
up
for
discussion
like
I,
think,
are
up
for
a
discussion
and
which
bits
are
sort
of
seem
to
me
to
be
complete
no-brainers
that
we
shouldn't
just
do
also
a
couple
other
things
that
I
would
call
out
here:
yeah
yeah,
so
that
one
that
one
was
based
on
a
suggestion
from
Tim
as
part
of
one
of
the
API
reviews.
Rob
and
I
have
talked
about
it
before
you
know
again.
F
Yet
because
we
haven't
agreed
right,
so
that's
I'm
trying
to
make
it
so
that
we
have
a
way
to
avoid
the
problem
that
we
had
with
overall
policy
attachment
where
people
started
using
it
before
we
had
all
sort
of
finished
working
on
it
and
so
I'm,
trying
to
I'm
trying
to
manage
that
problem
here
from
the
outset
in
terms
of
what
else
to
call
out
yeah
I
think
those
are
the
main
things.
I
tried
to.
F
The
things
are
yet
there's
a
couple
of
ideas
that
are
a
bit
wild
like
having
a
validating
and
Mission
controller.
That
tells
you
if
your
thing
is
affected
by
policy,
but
so
I.
Just
wanted
to
throw
a
whole
bunch
of
ideas
in
there
so
that
we
could
talk
about
them
within
this
PR
I
have
sort
of
the
other
thing
that
I've
done
as
part
of
this.
That
I
is
going
to
come
up
in
another
discussion
that
I've
got
listed.
F
There
is
that
I
had
I
talked
to
Flynn
about
this,
but
I
changed
the
names
of
the
people
from
Julian
and
Jane
to
Charlie
and
Anna,
with
the
intent
being
that
we
should
give
standard
names
to
our
personas.
We'll
talk
about
that.
F
The
last
thing
is
that
that
other
sitesource
references,
doc,
yeah
policy
attachment
currently
duplicates
like
most
of
the
policy
attachment,
get
and
also
gives
policy
attachment,
maybe
an
appearance
of
being
more
like
more
stable
and
less
experimental
than
it
actually
is,
and
so
this
is
one
of
the
primary
contributors
to
the
state
that
policy
attachment
has
ended
up
in
and
so
I
have
removed.
This
and
pointed
pointed
us
to
off
to
the
off
to
the
main
Gap,
so
yeah
tried
to
sort
of
make
it.
F
You
know
like
clear
that
this
thing
is
not
ready
yet,
and
you
can
see
in
the
Gap
that
it
is
provisional
or
whatever
so
anyway.
That's
that's
all!
That's
what
I
have
to
say
there
please
go
and
review
it
and
give
me
comments
on
it.
I
am
going
to
be
after
this
week,
I'm
going
to
be
on
leave
for
three
weeks,
so
I.
Don't
think
that
this
is
in
a
huge
rush,
so
you
know
I
wanted
to
give
it
time
you
know.
Have
it
have
people
time
to
think
about
it?
A
I,
definitely
like
that
better
than
waiting
another
four
weeks
to
open
it.
Yes,
I
think
it
might
be
nice
to
get
the
the
change
to
the
policy
attachment
document
in
sooner
than
a
month
from
now
the
one
that's
you
know,
removing
most
of
it
and
pointing
back
to
a
gap,
but.
F
A
I
can
split
that
up
yeah
yeah.
That
would
I
think
that
would
be
a
nice
thing.
Okay,
I
just
modified
the
the
meeting
notes
to
note
that
I'm
going
to
close
Gap
2014
as
superseded
by
Nick's
work
here.
Thank
you.
Nick
much
appreciated
I
still
have
concerns,
but
this
is
this
at
least
covers
a
lot
of
things
that
I'm
most
concerned
about
so
yeah.
H
H
It
relates
to
a
situation
that
we
often
find
ourselves
in.
Maybe
because
we
are,
we
are
not
implementers
API
in
the
same
time
solve
all
the
implementations
that
actually
provide
gateways
and
the
actual
workload
to
know
about
for
for
the
apis,
but
we
are
nevertheless
providing
custom
controller
for
policies
right.
Some
implementation
for
policies
effectively
leveraging
existing
Gateway
implementations
right.
So
that
is
the
main
idea,
and
we
often
whenever
we
see
things
such
as
you
know,
updating
these
status
on
a
stencil
of
a
resource
right.
H
We
already
always
wonder
you
know
what
we
can
do.
You
know
in
this
kind
of
situation,
because
we
clearly
we
we
would
have
multiple
different
controllers,
trying
to
update
the
same
resource
right
so
yeah.
F
Yeah
yeah,
so
yeah
I
think
this
there's
definitely
some
work
in
Spec
worker.
We
probably
need
to
do
here
to
to
make
that
clear,
but
I
I
in
general,
it's
supposed
to
be
sort
of
a
Boolean
like
something's,
affecting
this
I,
think
that
we
might
need
to
have
some
guidelines
in
here
to
say:
hey
if
this
is
all,
if
you're
trying
to
set
it
to
True,
something
else
has
already
set
it
to
true
just
leave
it
alone.
Right,
like
you
know,
the
important
part
here
is
to
say
some
policies
affecting
this
go.
F
Look
for
the
policy
right,
like
you
know,
and
you
know
this
is
a
slightly
unusual
thing
for
a
condition,
because
it's
generally
expected
that
you
won't
coordinate
on
one
condition
in
kubernetes,
like
conditions
are
supposed
to
be
namespace
to
avoid
that
sort
of
thing,
however,
I
kind
of
also
want
to
avoid
having
ending
up
with
like
implementation,
specific
policy
affected
conditions,
so
you
might
end
up
with
like
six
different
affected
by
this
person.
This
implementations
policy.
F
It
is
a
nice
little
bit
of
extra
information,
but
it
also
dilutes
the
sort
of
tool
utility
of
being
able
to
say
some
policies
affecting
this
goal
before
right,
like
so,
I
think
that
it's
probably
I'll
have
a
think
about
that
and
see.
If
I
can
you
know,
change
the
wording
a
little
bit
to
make
that
clear
off.
F
But
in
my
in
my
mind,
that's
what
it
is
like
you
know,
it's
just
that
we
need
to
have
a
thing
there
saying
you
know
if
you're
going
to
try
and
update
this-
and
it's
already
present
with
this
data
set
to
true,
then
just
leave
it
alone.
Right
like
don't,
you
know,
do
you
need?
You
will
need
to
have
a
special
check
in
your
contusion
checking
luggage
logic,
unfortunately,
that
you
know
that.
F
Doesn't
that
just
checks
you,
you
really
need
to
check
the
observed
generation,
the
type
and
the
sort
of
the
reason,
a
message
and
stuff
like
that,
and
we
will
need
to
be
very
specific
about
reasons
and
reasons
and
then
say
ignore
the
message
or
something
like
that
like
we'll
need
to
sort
of
be
very
specific
about
condition,
checking
Logic
for
this
specific
condition,
I
think
unless,
unless
there's
some
other
way
around,
this
problem.
B
Yeah
I
I
agree
with
what
you're
saying
there
I.
This
is
more
of
a
meta
level
comment
here:
I
what
what
key
described
as
kind
of
a
policy
controller.
It
sounds
very
interesting,
so
if
you
have
any
docs
or
anything,
you
can
link
us
to
I'd
love
to
see
what
this,
what
this
looks
like
how
it
works
Etc,
because
it's
it's
kind
of
an
area
that
I'm
kind
of
blind.
You
know
I,
don't
know
much
about,
but
I'd
love
to
see
a
different
yeah
cool
awesome.
Thanks
for
the
link,
yeah.
H
The
quadrant
Yeah,
so
basically
you're
implementing
functionality
suggests
such
as
rate
limiting
Global
rate
limiting
and
other
authentication
authorization,
also
multi-cluster
Gateway
consistence
through
K2
API
policies
and
and
obviously
attachment,
obviously,
and
while
relying
on
existing
proper
gate,
API
implementations,
for
example,
we
spot
that
we
have
istio,
for
example,
delivering
a
Gateway
API
functionalities.
So
we
from
we
we
reconcile
our
policies,
configuring,
the
underlying
istio
deployment,
if
not
istio,
all
the
gate,
API
implementations.
Next
one
on
the
list
is
invoice.
For
example,.
A
H
Yeah,
that's
awesome.
Yeah
thanks,
Nick
yeah
I.
Think
overall
I
just
like
to
add
that
we
see
that
our
first
impression
of
Europe
is
very,
very
positive.
Just
like
pretty
much
everything
that
you
guys
have
been
doing,
I
think
a
Craig's
point.
There
is
only
about
raising
the
awareness
about
this
older
Persona
that
might
be
sometimes
also
part
of
the
picture.
There.
F
Importantly,
I
actually
added
a
Persona
I
want
to
add
a
Persona
as
part
of
this
called
policy
admin.
Mental
note
for
myself
I
actually
need
to
give
that
that
person
the
name
if
we
all
agree
that
they
should
have
names
but
yeah.
So
you
know
the
yeah.
The
idea
there
is
to
sort
of
yeah
give
the
you
know.
I've
actually
got
an
update
to
the
diagram
as
part
of
that.
F
The
sort
of
canonical
here
is
the
hierarchy
of
Gateway
API
Resources,
with
the
policy
admin
sort
of
poking
off
to
the
side
and
affecting
everything,
because
that's
really
what
happens
so
yeah,
so
that
sort
of
where
that
one
is
at
I
think
we
should
definitely
time
box
this
one.
We
could
talk
a
lot
longer
about
this,
but
I
would
rather
have
a
lot
of
these
more
of
these
conversations
on
the
pr
so
that
there's
more
history,
yeah
and,
as
I
said,
Yeah,
the
more
the
more
that
we
can
have.
F
Okay,
so
I'm
going
to
move
on
to
my
next
one,
which
is
like
I,
said
giving
personas
names.
F
You
know
I'm
proposing
you
that,
yes,
that
we
give
our
Persona
standard
names
and
that
we
give
them
names
that
are
sort
of
reflective
of
the
initial
letter
of
the
name
of
the
current
Persona,
so
Ian
with
the
infrastructure,
admin
Charlie
would
be
cluster
admin
and
the
application
developer
would
be
Anna
with
the
intent
being
that
you
know
just
it
gives
us
a
shorter
hand
to
do
this.
It
makes
it
easier
to
understand
what
these
people
might
be
yeah.
F
So
please
have
a
look
at
that
comments.
Votes,
Etc
welcome.
The
idea
here
is
to
start
up
this
discussion
so
that
when
I
get
back,
I
can
open
an
actual
Gap
and
then
make
it
happen.
A
Okay
notes
that
we're
down
to
10
minutes,
which
is
a
good
point,
I,
absolutely
100,
think
that
the
personas
need
to
have
names
in
order
to
be
personas
as
opposed
to
abstract
descriptions
and
yeah.
You
know
it.
It
makes
me
sad
to
get
away
from
all
the
J
names
but
like
Jane,
but
that's
okay,
I
think
I
can
I
think
I
can
support
this
proposal.
F
Yep,
the
other
one
is
I
suspect.
Maybe
more
controversial.
I
was
thinking.
The
other
day
about
people
have
been
complaining
since
I've
been
talking
about
Gateway
API
about
how
hard
it
will
be
to
write
a
Helm
chart
that
automatically
drops
our
HTTP
route
in
your
cluster.
That
just
works,
and
so
here
is
like
a
bit
of
a
strong
person
argument
about
about.
You
know
a
so.
F
F
So
you
know,
implementations
could
provide
this
as
part
of
their
install
or
they
could
you
or
an
administrator
could
create
this
as
part
of
their
install
and
then
people
writing,
Health
charts
can
sort
of
expect
they
can
Target
the
internet
gateway
in
the
default
namespace
and
have
HTTP
routes,
but
it'll
probably
work
right,
like
the
intent
here
is
just
to
sort
of
cut
a
couple
of
those
layers
of
like
oh
now,
I've
got
to
go
and
find
out
like
what
Gateway
I
need
or
and
which
namespace
it's
in,
and
you
know
like
a
bunch
of
stuff
like
that
and
if
the
route
is
allowed
a
few
other
things
like
that
to
sort
of
give
a
slightly
easier
on-ramp
to
sort
of
trying
this
thing
out,
where
maybe
you
don't
control
the
gateways
where
you
are
Anna
and
not
Charlie,
so
yeah
Leo,
you
were
first.
D
Yeah
just
curious,
like
is
this:
like:
did
you
get
his
feedback
like?
Where
did
you
get
this
feedback?
It's
interesting
because,
like
I
mean
again.
F
Yeah
basically
I
mean,
but
for
people
who
are
sort
of
like
I
I
like
that
I
like
the
idea
of
the
API,
but
one
of
the
things
that
I
liked
about
Ingress
was
that
you
could
you,
if
you're
installing
an
app
you
can
just
include
an
Ingress
and
be
pretty
confident
that
it'll
probably
work
right,
like
you
know,
and
right
now,
you
can't
do
that
with
Gateway
API,
because
there's
no
conventions,
there's
no
sort
of
places
to
have
any
of
this
standard
stuff.
All
right.
So
that's!
F
The
idea
here
is
not
I
put
a
suggestion
like
a
proposal
in
there,
because
I
wanted
to
have
something
concrete
to
discuss,
rather
than
just
let's
talk
about
this
much
easier
to
have
people
be
like
that's
a
rubbish
idea
and
here's
a
better
one
than
than.
G
Yeah
I
can
probably
only
provide
the
first
part
of
that
statement
that
I
I,
really
like
the
idea
and
I
think
Austin
who's
on
vacation
would
be
jumping
to
hear
this.
Because
he's
talked
about
this
many
many
times,
I
kind
of
hate.
The
the
idea,
though,
of
the
kind
of
standard
Gateway
like
we
shouldn't,
be
the.
E
G
Namespace
usage
do.
G
This
particular
one
so
I
think
there's
potentially
better
ways
we
could
do
it.
I
don't
have
any
good
ideas
offhand.
This
one
just
makes
me
really
uneasy
to
like
say
that
we're
encouraging
kind
of
this
default,
namespace
gateway,
I,
don't
think
that's
really
a
good
practice
and
it's
like
allowed
routes.
All
namespaces,
maybe
is
not
necessarily
best
practice
either.
A
This
feels
like
a
thing
where
a
given
implementation
should
be
allowed
to
not
provide
it
and
clearly
document
that
but
yeah
if
you're
you're,
if
you're
an
app
developer
as
opposed
to,
if
you're
on
an
instead
of
Charlie,
then
this
really
could
make
your
life
a
whole
lot
easier
in
some
ways.
So
yeah
I.
G
G
G
It
goes
to
the
global
default
parent
ref
that
may
be
also
concerning,
but
that
feels
a
bit
more
intuitive
to
me
like
I,
like
the
idea
of
unset
means
default
rather
than
we
have
this
canonical
name
that
maybe
or
maybe
not
someone
follows
because
if
it's
just
this
kind
of
fuzzy
canonical
name,
then
every
Helm
chart
is
going
to
need
to
provide
the
ability
to
override
it
anyways.
And
then,
if
someone
doesn't
do
that,
then
they
just
provide
the
hard-coded
internet
default.
A
G
Know
this
is
all
off
the
cuff,
so
I
think
it's
a
good
idea.
It
probably
is
worth
discussing
more.
A
I
I
actually
agree
with
John's
statement
that
I
don't
think
that
users
should
be
just
having
gateways
by
default
and
stuff
like
that,
and
yet
I
also
have
the
use
case.
That
tells
me
that
it
can't
be
so
so
we
have
both
a
Gateway
operator
which
dynamically
creates
gateways,
which
we
would
consider
the
the
best
way
to
basically
use
Gateway
API,
but
we
have
our
classic
Ingress
controller
with
Gateway
API
support,
which
does
it
in
what
we
called
a.
What
did
we
call
it
a
non-provisioning
mode
or
something
like
that?
I
F
Again,
a
large
part
of
what
I'm
trying
to
do
here
is
just
to
draw
out
the
discussion
and
to
actually
have
a
discussion
about
this
and
not
sort
of
like
that.
Fester
is
a
thing
that
no
one
is
thinking
about
or
doing
anything
about,
I
actually
don't
care.
If
this
specific
proposal
I
have
there,
it
doesn't
happen.
F
Yeah,
like
the
you
know,
like
all
I
wanted
to
do,
was
start
off
the
conversation
and
sort
of
say,
hey
you,
I
have
literally
got
feedback
when
I
have
presented
at
this
at
sort
of
non-cubecon
places,
and
sometimes
even
a
coupon
I
have
got
feedback
nice
API,
but
how
the
hell
am
I
going
to
use
it
in
Hamilton
how's
it
going
to
fit
in
a
health
chart
right
like
your
daddy's,
and
that
is
the
extent
of
the
use
case
right
like
nice
APA,
but
how's
it
going
to
fit
in
the
health
chart.
F
Right
like
you
know,
and
that's
the
sort
of
the
you
know
and
that's
the
sort
of
things
that
I
just
think.
We
need
to
think
about
right,
like
you
know,
if
it
doesn't
turn
out
to
be
this
like,
even
if
we
have
a
convention
or
something
like
that
that
says,
if
you
know
you
know,
the
sort
of
default
expectation
is
that
there's
going
to
be
a
Gateway
somewhere,
that's
called
this
right
like
or
the
Gateway
classes
called
this,
and
we
do
have
a
sort
of
default
parent
rafting
like
like
John
was
talking
about.
F
That
would
be
fine,
but
I
like,
but
the
intent
here
is
to
make
it
so
that
you
know
right
now.
You
can
include
an
Ingress
in
a
Helm
shot
and
have
a
reasonable
expectation
that
something
will
happen
right,
like
you
know,
as
long
as
the
university
contract
is
fulfilled
somewhere
in
the
cluster
right,
like
you
know,
you
can't
do
that
with
go
API
like
for
some
good
reasons,
but
also
we
need
to
sort
of
have
a
way
to.
Let
will
say,
I
want
this
tall
thing
to
be
magic
right,
like.
I
F
Yeah
so
yeah.
This
is
what
this
is.
What
I
mean
like
this
is
the
exact
discussion
that
I
wanted
to
have
I
would
encourage
everybody
to
please
go
and
write
your
comments
on
the
thing.
I
will
not
be
offended
if
you
say
you
know
having
a
Gateway
in
the
default.
Namespace
is
a
terrible
idea
and
you
should
feel
bad.
It's
fine,
but
I'm.
F
You
know
my
skin
is
thicker
than
that
now,
so
you
know,
like
you
know,
do
do
that,
like
that's
fine
again,
the
intent
here
is
to
sort
of
have
a
discussion
so
that
we
can
try-
and
you
know,
get
this
just
state
where
we
write
a
gap
at
the
end
of
the
day.
That
says,
if
you
want
things
to
be
magic,
here's
what
you
do,
and
even
in
the
absence
of
that,
we
should
provide
feed,
like
you
know,
guidance
for
people
who
want
to
ride
a
Helm
shot.
F
That
includes
Gateway
API
things
as
to
how
you
know
as
to
how
that
how
that
Helm
chart
should
handle.
If
you
want
to
have
a
HTTP
route
or
something
like
that
right.
If
you
want
to
have
a
HTTP
route,
what
should
your
HTTP
route
say
right,
like
you
know,
what
value
should
you
provide?
Should
you
error
out?
G
D
Just
one
last
night
about
it,
like
a
b
like
it'll,
be
useful
to
understand,
like
all
those
things
and
people
for
for
the
use
cases
because,
like
it,
depends
how
much
we
want
to
invest
in
on
it.
So,
like
you
could
you
know,
write
helpers
that
sell
it
validations
and
help
like
there's
a
lot
of
magic
you
can
do
and
how
to
help
the
people
do
on
board
and
take
it
forward,
but
it
depends
how
much
you
want
to
invest
on
it.
F
You
know
and
yeah
exactly
the
opposite
to
what
Finn
is
joking
about
in
the
chat
you
know
like,
we
should
be
very
light,
touch
to
start
with
and
just
sort
of
be
like.
We
suggest
that
you
do
this
right,
because
you
it
doesn't
take
much
of
a
suggestion
for
a
suggestion
to
become
you
know
a
convention
right
like
and
that's
all
we
need.
F
We
do
right
like
the
most
stuff
that
we
probably
we
could
get
by
here,
with
having
a
set
of
conventions
about
like
the
ways
that
you
would
handle
this
particular
thing
right
like
and
that's
all,
that's
all
I
really
kind
of
want,
like
you
know
it's
just
to
have
some
conventions
to
say.
If
you
want
to
do
that,
you
should
do
it
like
this.
A
One
minute
at
negative,
one,
okay,
I
was
gonna,
say
I
would
strongly
encourage
everybody
to
go
and
take
a
look
at
the
things
that
Rob
has
listened
in
a
triage,
but
I
think
we're
going
to
have
to
pick
those
up
next
time.
A
A
So
so
I'll
be
taking
a
look
at
that
shortly
soon
and
I
think
that
is
it.
Unless
anybody
has
any
horribly
urgent
things,
they
want
to
jump
in
on
going
once
going
twice.
No,
let's
really
not
create
a
home
health
replacement.
All
right!
Thanks
much!
Everybody
we'll
see
you
all
next
time
take.