►
From YouTube: 2020-04-21 Apache Cassandra Contributor Meeting
Description
This meeting we talk about testing and the addition of the new fallout project. An update on the Kubernetes Operator. Discussion about how to do an official docker image for the project. DataStax donation of Java and Python drivers
A
So
we
can't
do
Calendar
widgets
inside
of
see
wiki,
so
I
just
created
a
link.
There's
two
things
in
there
there's
this
meeting,
which
has
multiple
months
run
out
and
also
the
reoccurring,
kubernetes
meeting.
You
know
that's
in
there.
So
if,
if
you
add
that
one
calendar
you
get
all
those
for
free
can.
A
A
Good
to
start
my
notes.
The
first
thing
that
was
on
the
agenda
for
today
was
just
an
update
on
what's
happening
with
kubernetes
the
kubernetes
meeting
and
we
have
actually
I'm
sorry.
I
did
I,
have
an
update
on
testing
and
kubernetes
operator,
but
make
wanted
to
do
the
update
on
testing
and
he's
not
here.
So
maybe,
if
somebody
could
ping
him
real,
quick.
Oh.
C
E
E
Crickets
and
yeah
Jordan
I'll
also
mention
that
I
saw
all
your
pings
and
slack
on
some
of
the
tickets
and
I
think
that
we
are
in
JIRA
and
some
of
the
tickets,
and
that
would
be
really
helpful
if
we
could
track
down
some
of
that
feedback.
Just
so
we
don't
duplicate
any
effort
from
anybody
else.
There
yeah.
F
I
think
one
of
the
first
steps
is
just
sort
of
like
enumerated.
I
mean
we
can't
be.
You
know
where
I
find
other
stuff,
but
at
least
starting
to
enumerate
what
we
want
to
test
and
in
some
sort
of
priority,
like
you
know,
storage,
engine,
stuff,
more
important
than
I,
don't
know
some
feature,
nobody
uses
kind
of
stuff
so
that
we
can
make
decisions
about
where
to
put
that
effort.
E
F
We
absolutely
want
performance
covered
for
a
while
that
was
kind
of
being
handled
by
Netflix,
like
with
the
messaging
service.
They
were
offering,
like
their
suite
of
tools
and
their
sort
of
resources,
to
run
a
lot
of
performance
benchmarks,
so
we
can
catch
up
with
them,
but
I
mean
I.
Think
absolutely.
We
need
to
figure
out
the
performance
coverage
also
to
directly
answer
your
question.
No,
there
isn't
a
plan
right
now.
We
should
make
one.
G
So
I
had
developed
the
tools
that
now
are
in
the
data
stacks
portfolio
for
tailpiece,
stress
and
cluster,
and
in
there
there
were
things
like
Griffon
dashboards.
Tail
key
stress
had
the
workloads
out
of
the
box
and
tippy
cluster.
Could
fire
up
arbitrary
size
clusters
in
Amazon?
It's
not
perfect,
there's
definitely
a
whole
bunch
of
bugs,
but
there
might
be
some
bits
and
pieces
out
of
there
that
can
be
used
at
some
point
for
folks.
D
E
People
want
to
use
it,
ok,
sweet
and
yeah.
What
we
were
thinking
is
that
we'd
have
part
of
the
report
is
like
okay.
What
did
this
test
run?
What
were
the
metrics
maybe
shown
by
a
graph
on
a
dashboard
or
something
like
that
then
kind
of
a
summary
of
each
thing
that
was
run
through,
that
we
could
post
back
to
zero
or
something
like
that
just
to
keep
track
of
what
was
run,
but
we'll
definitely
check
out
those
those
other
options
too.
For
the
tooling.
A
A
Fair
enough
yeah
boundaries.
What
are
that?
Okay,
so
I'm
gonna,
I'm
gonna,
just
put
what
he
put
in
the
dev
list
in
this
doc
there
he
had
a
series
of
zeros.
That
I
think
are
these
two
jurors
that
you
were
talking
about
Oh
Chris.
This
is
you
am
I
doing
here.
Sorry,
yeah
I
was
just
gonna
copy
and
paste
those
him
yeah,
yeah
yeah.
Sorry
Mick
had
to
know
differently
all
right,
all
right,
cool.
E
A
Right,
the
other
thing
that
I
had
on
here
was
the
cubic
eighties
stuff,
the
kubernetes
meetings
that
we
had
went
reasonably
well.
We
had
good
good
attendance
on
both.
There
was
a
some
feedback
on
the
death
list
from
from
about
having
these
one
having
one
meeting
instead
of
two
now.
The
initial
reason
for
having
two
is
just
to
be
a
little
more
friendly
about
24
time
zones.
A
G
I
can't
speak
for
everybody,
but
I
think
that
those
of
us
that
don't
have
any
real
kubernetes
like
experience
I,
have
a
feeling
we're
gonna
end
up
sort
of
just
not
really
engaged.
It
seems
like
it
seems
like
there
really
are
two
meetings
like
you're,
either
running
Cassandra
or
you're,
helping
build
it
and.
A
Obviously,
there's
some
folks
that
you
go
well
there's
two
kubernetes
meetings:
John,
that's
talking
about
there's
one
and
all
late
at
night
on
West,
Coast
time
and
early
morning,
West
Coast
time,
so
that
we
can
get
this
a
pack
and
European
coverage
going
so
kubernetes
meetings
and
on
top
of
this
one.
So
that
was
what
he
was
talking
about.
It's
like
how
about
one
kubernetes
meeting
and
yeah.
A
We
we
talked
about
this
time
like
for
this
meeting.
This
time
is,
is
kind
of
in
the
weird
middle.
You
know
it
overlaps
a
lot
of
time
zones
some
and
it
rotates
and
I
rotate
these
every
every
meeting
time
so
that
we
have
the
pain
shifts
from
one
place
to
the
other,
and
it's
like
a
round
robin
pain,
so
I
I
mean
I.
A
A
The
other
thing
worth
mentioning
is
what
some
of
the
outcome,
when
that
was,
we
are
working
on
CEP,
Ben
and
I
worked
on
it
some
yesterday,
so
Ben
Bromhead
and
I
are
kind
of
heading
up
this
initial
documentation
effort.
This
was
part
of
that.
Those
meetings
was
all
right.
Let's
get
it
all
written
out,
we
will
we'll
have
it
all
wrapped
up
for
the
kubernetes
meeting
this
week
and
you
know
start
hopefully
working
on
a
group
doc
that
outlines
a
lot
of
the
initial
expectations.
I.
A
Think
it
really
it's
going
to
kind
of
shape
around
the
levels
that
are
being
specified
by
the
kubernetes
operator
project.
There's
like
11,
1,
2,
3,
4,
5
and
level.
3
goes
up
to
a
full
lifecycle
management
and
then
once
you
get
beyond
that
you're
in
cutting
in
a
different
place.
But
if
you
know
that's
where
most
people
are
like,
we
need
to
get
to
level
3
on
operator.
Compliance
at
this
point
that
makes
full
lifecycle
management
of
a
Cassandra
cluster
possible.
H
So
to
John's
point,
but
this
was
a
pretty
large
divide
between
Cassandra
database
developers
and
then
companies
operator
developers,
and
we
have
some
that
have
a
foot
in
both
well.
They
have
a
lot
of
people
have
a
foot
in
both
camps,
but
I
mean
that
does
run
the
risk
of
becoming
a
de-facto
deployment
approach
to
things,
because
a
boatload
of
people
are
migrating
into
kubernetes,
probably
worth
it's
just
at
least
having
in
our
radar
of
some
kind
of
whether
it's
documentation
or
on-ramp,
or
something
where
it's
like.
H
Now,
like
so
so
normally
you'd
say:
okay,
I'm
gonna
write
some
documentation
for
a
user
who's
that
who's
that
user
gonna
be
person.
I'm
writing
this,
for
you
write
it
for
a
person.
That's
gonna
be
running
Cassandra
using
urban
at
ease,
but
there's
another
user
here
right,
just
like.
If
you
go
to
the
documentation
for
getting
started
as
a
developer
on
Cassandra,
that's
very
specific
documentation,
saying
like
here's,
how
you
clone
it
here,
so
you
run
CI.
H
Here's
how
you
actually
work
as
developer
on
Cassandra,
targeting
that
same
persona
and
saying
and
here's
how
you
as
a
Cassandra
developer,
also
work
on
the
operator.
Here's
that
you
understand
the
basics
of
kubernetes.
Here's
how
you
understand
how
it
sockets
into
the
operator
for
Cassandra
would
be
useful.
That's
the
specific.
A
That
was
that
was
definitely
surfaced
in
those
meetings.
Is
we
have
a
lot
of
kubernetes
expertise
showing
up?
There's.
Here's
like
where
some
of
the
weighting
is.
We
have
a
lot
of
people
who
run
kubernetes
or
company
X.
You
know,
so
that's
a
scale
problem
that
they
and
their
experts
in
running
kubernetes,
they're
approaching
Cassandra's,
yet
another
thing
that
they
have
to
support.
What
can
we
do
to
help
that
effort?
You
know
this
is
the
scratch
they're
trying
to
inch,
which
is
cool
because
I
mean
that's
a
cool
scratch
right
there
like
I.
A
F
Think
I
think
the
concrete
example,
though,
regardless
of
expertise,
is
like
it
should
be
documented
somewhere
how
you
building
not
an
official
release
of
the
kubernetes
operator
so
that
you
work
on
it
just
like
there's
documentation
of
how
you
build
like
a
non-official
local
Cassandra.
So
you
can
work
on
it
back.
F
H
Think
the
last
thing
we
want
is
for
the
kubernetes
operator
to
become
like
this
thing
that
exists
somewhere
else
where
Cassandra
developers
to
get
on
the
project
looking
at
askance
are
like
that's
a
pain
in
the
ass
to
learn,
I'm
really
busy
I,
don't
understand
that
partly
view
much
better
for
us,
I
think
holistically.
If
that
was
like
a
part
of
the
ecosystems
that
was
consumed
like
table
stakes
before
working
on
Cassandra,
it's
also
understanding
and
working
on
that.
If
you
need
to
or
choose
to.
C
Yeah,
we
honest
I,
don't
think
it's
really
going
to
the
pipoint
to
do
it,
because
I
can
tell
you
at
the
beginning
when
I
started
at
the
end
of
last
year,
we
had
a
lot
of
to
do.
Sections
in
the
cassandra'
documentation
and
I
was
asking
for
some
stuff
around
and
before,
like
some
people
were
from
community
and
other
developers,
they
were
like,
oh
I,
don't
know,
I
use
this
or
that
sometimes,
but
it
was
like
bori
and
I
never
managed
to
make
it
work
or
whatever,
and
we
had.
C
We
didn't
have
documentation
from
all
the
stuff
or
checking
different
forms
and
stuff,
so
it
shouldn't
be.
We
definitely
should
take
care
of
it
like
a
high
priority.
No,
this
is
my
my
vision
from
what
I
saw
before
and
now
when
we
don't
have
all
those
Buddhist
people
way
more
happy
and
way
more
going,
there
checking
and
doing
a
lot
of
stuff.
That's
just
my
observation.
A
That's
good
feedback,
yeah
I'm,
a
I
want
to
also
combine
a
couple
of
top
is,
it
seems,
like
there's
a
directionality
with
the
testing
it's
happening
in
Cassandra
and
running
it
on
kubernetes
and
and
docker.
You
know
creating
images
tons
of
like
whatever
base
image
we
had
this.
This
is
another
hot
topic.
Inside
of
this
is
there's
no
official
docker
image
for
Cassandra
other
than
what's
been
created
by
the
darker
people
and
it's
kind
of
out
of
bounds
of
the
ASF
rules.
They,
like
you,
can't
do
that.
G
A
H
G
Doing
this
would
actually
be
remarkably
easy
if
we
I
mean
we
have
to
change
a
few
things,
but
creating
and
publishing
a
docker
image
is
actually
not
that
hard.
If
the
build
system
is
set
up
a
little
bit
differently
and
we
haven't
right
now,
we
did
it
with
the
TLP
stress
and
it
was
only
like
adding
six
lines
of
configuration
to
the
build.gradle
I.
Just
did
it
for
us,
I
mean.
H
D
A
H
G
A
G
Can
be
like
it's
easy
to
build
it
as
a
just
tar
ball,
that's
fairly
straightforward,
so
honestly,
I
think
that
I
think
it
should
be
while
there
might
be
some
benefit
to
the
snapshots.
It's
also
I
think
the.
If
the
primary
goal
is
make
it
so
easy
that
it's
just
like
typing
in
exactly
one
command
into
the
build
system
that
generates
that
docker
image,
then
it's
like,
then
you
don't.
G
We
don't
really
need
you
to
worry
about
infrastructure,
and
we
kind
of
need
to
do
that
anyway,
like
right
now,
you
know
it's
like
building
your
own
Deb
package,
right
of
course,
I.
Nobody
knows
how
to
do
that,
but
in
theory
it's
something
that
you
should
be
able
to
do
very
trivially,
and
it
is,
if
you
know
the
mystical
magical
commands
to
make
it
work
guillotine.
I
You
could
we
put
images
of
tinker
pop
out
on
on
docker
hub
on
releases,
so
we
have
a
standard
vote
release
we
just
published
directive
docker
and
I
was
just
amazed
and
command
sent
it
up
there
I
don't
know
what
Apaches
issued
I
know
they
have
there's
a
lot
of
things
they
don't
like.
They
don't
like
they
might
not
like
docker
hub.
They
might
not
like
type
I
and
all
those
other
places
that
you
can
publish
stuff
to.
But
typically
it's
done
as
part
of
a
release
like
at
least
my
interaction
with
them
has
been.
I
They
don't
seem
to
have
a
huge
problem
with
at
least
that's
been
kind
of
my
dealings
with
them
on
that.
So
that
kind
of
gets
around
issues
with
you
know,
quote-unquote
releasing
something
that
isn't
source
code.
So
that's
that's
kind
of
been
the
what
we've
done
with
them
for
a
long
time,
I
think
a
lot
of
other
projects.
Do
it
too
I
didn't
last
that
the
Cassandra
doctor,
because
they
were
out
there
already,
that
I've
been
playing
with
or
actually
they
were
totally
they
were
told.
A
Well,
apparently,
you
take
about
nothing
yeah,
I,
there's
I,
think
there's
just
like
some
consternation,
but
a
lot
of
folks
in
the
kubernetes
community.
They're
like
why
isn't
there
an
official,
Cassandra,
docker
image-
and
you
know
it's
like
from
the
project
it's
like
I-
have
to
go
to
a
docker
hub,
and
so
there's
already
a
little
side.
I
like
is
that
is
that
just
some
enthusiasts
putting
it
together
and
is
that
you
know,
is
that
the
right
freeIn
or
what
have
you
yeah?
This
might
be
something
a
little
more
PMC
oriented
like
hey.
A
D
A
I
A
I
H
H
I
We
always
we
did
on
all
the
everything
whether
it's
Popeye
or
new,
get
or
docker
hub.
We
put
some
language
in
there.
It
makes
it
really
clear
what
is
an
official
release
right,
so
you
might
have
released
candidates,
you
might
have
we
like
nightbeat's
or
anything
like
that
and
stay
with
your
mother.
So
it's
always
a
something
based
on
an
official
bill
that
is
repeatable
you.
Could
you
can
check
out
that
specific
3.46
tag
and
reproduce
that
exact
artifact
that,
in
the
same
exact
way
right?
I
But
we
haven't,
we
definitely
put
language
in
there.
That
says:
hey
this,
you
might
use
this
pie.
Pie
thing
for
convenience,
recognize
that
if
something
is
marked
with
it
is
I
think
we
use
Farsi
for
a
release
candidate
or
something
that's
like
that.
So
I
developed
a
test
was
for
that's
for
developers
only.
This
is
not
official
in
any
way.
So
we
had
like
all
those
kinds
of
like
caveat
built
into
all
those
things.
So
hopefully
that's
kind
of
enough.
It's
been
enough
for
a
lot
of
the
other
binaries.
I
We
release
like
different,
actually
no
Brenna,
console
and
sort
of
like
these.
These
you
know
non
source,
release
things
and
we've
always
been
told
to
put
those
kinds
that
kind
of
verbiage
in
there,
and
that's
made
a
bit
of
a
difference
in
how
those
things
were
accepted
by
folks
and
Apache,
as
though
so
I
don't
know
that.
That
might
be
something
else
to
consider
if
you,
if
you
discuss
this
further
and
decided
move
forward
with
it,.
A
A
E
Yeah
Patrick
I
got
one
more
topic
yeah,
so
we
at
data
stacks
have
been
exploring
the
donation
of
our
drivers
and
I
wanted
to
get.
The
sense
of
the
group
here
on
is
there
interest
from
the
project
for
taking
ownership
of
those
drivers,
if
so,
which
ones
and
then
I'm
also
interested
in
your
opinion
of
whether
this
makes
sense
to
start
pursuing
now
or
whether
we
should
wait
until
after
for
us
that
we
don't
distract
from
the
release
really
anything
but
open
to
thoughts
and
opinions.
J
J
H
H
H
E
E
L
J
People
can
contribute
to
without
complicated
what
the
name
of
the
agreement
contributors
agreement
without
delays.
Yes,
because
I
know
if
at
least
one
organization
that
was
very
willing
to
contribute
for
a
very
long
time,
but
was
stuck
in
a
legal
hell
with
with
regards
to
CL
a
certain
anonymous
unnamed
organization.
But
it's
much
easier
to
deal
with
if
it's
okay.
H
Dog
I
think
also
did
a
great
point
on
the
last
call
we're
talking
about
this,
where
there's
like
a
an
organizational
bus
factor
where,
if
it's
on
data
sex
is
github
in
its
open
source,
that's
great,
but
if
data
stacks
you
know
explodes
wipes
off,
face
the
earth.
Whatever
happens,
you
can't
fork
things
and
donate
them
to
the
Apache
foundation,
so
those
drivers
are
all
will
forever
live
outside
the
Apache
ecosystem
and
never
be
part
of
the
project
itself.
J
Yeah,
like
the
really
the
need
for
project
control
driver
is
given
tight.
Integration
with
Cassandra
itself
has
been
raised
multiple
times
and
basically
the
only
reason
we
haven't
done
it,
because
we
don't
have
resources
like
X
resources
to
actually
do
that
ourselves
as
a
project.
Right
now,
maybe
like
an
absence
of
data
stacks
donating
in
the
future,
we
would
develop
an
independent
one,
but
if
data
stacks
can
donate,
I
mean
yeah.
Why
wasted
effort
in
reinventing
stuff
yeah?
It
would
be.
It
definitely
would
be.
J
F
F
J
Developer
of
some
alternate
library
complaining
that
this
makes
something
that
they
aren't
working
on.
Official
I
think
it
was
like
Hector
related
or
some
some
mapping
library
related
a
while
ago,
doesn't
matter
anyway,
I,
don't
think
right
now
we
have
MIDI
I,
don't
think
we
have
a
lot
of
like
alternative
drivers
that
have
major
adoption
today.
So,
like
the
reasons
for
the
any
competition,
so
I.
C
J
A
A
There's
different
things
like
we
have
a
ton
of
kubernetes
people
that
do
go
programming
and
they
want
to
be
committers
on
a
kubernetes
operator
and
if
it's
a
part
of
the
kampachi
cassandra
project,
then
they
need
to
be
committers
on
the
apache
cassandra
project
same
with
a
driver,
there's
gonna
be
driver.
People
yeah.
J
F
J
It
is,
it
is
a
little
bit
unfortunate
that
in
the
whole
SF
hierarchy,
there
is
just
commuter
BMC
member
I'd
like
this
is
basically
it
at
while
at
the
same
time,
we
do
recognize
explicitly
in
an
SF
guideline
that
a
contribution
is
not
just
code
and
it's
not
just
documentation.
I
there
like
it's
stuff,
like
helping
people
on
the
user
mailing
list,
for
example,
but
I
think
the
same
same
basic
rule
applies
like.
J
H
H
A
J
I'm
just
saying,
if
that,
if
that
happens-
and
we
will
list
this
in
our
guidelines,
we
will
take
away
there,
but
I
I
highly
doubt
that
someone
thoughtful
enough
to
invest
a
lot
in
helping
people
on
the
lists
and
stuff
like
that
is
gonna,
go
rogue
and
start
committing
to
drivers
kubernetes
and
pour
out
of
nowhere.
So
I
think
we
should
absolutely
write
this
down
in
our
guidelines,
but
I
I
don't
see
us
needing
it
in
practice.
You.
A
A
I
think
there
needs
to
be.
There
needs
to
be
people
in
the
PMC
that
are
focused
on,
like,
for
instance,
I've
heard
plenty
of
folks
in
our
current
PMC,
say:
I,
don't
know
anything
about
kubernetes.
Well,
we
should
fix
that.
You
know
there's
like
hey.
We
need
to
make
this
change
based
on
the
operator.
Sdk,
there's
gonna
be
a
whole
lot.
I,
don't
know
you
know
it
would
be
crickets
on
the
dev
list.
A
H
I
was
gonna,
say
you
know
more
holistically.
Extreme
I,
like
I,
need
to
look
into
this
more,
but
I
heard
on
the
podcast.
My
not
must
be
true,
but
I
think
spark
actually
has
like
people
in
marketing
that
are
committers,
if
not
PMC
members,
because
they
market
spark
like
their
role,
is
to
evangelize
that,
like
you,
could
take
it
even
further
and
say
you
need
people
that
their
entire
role
is
to
be
user.
H
It's
kind
of
a
vestigial
legacy
of
the
Apache
foundation
that
being
a
committer
on
a
project
means
you
have
good
access
to
the
stomp
all
over
code,
but
at
the
end
of
the
day,
like
I'm
I,
very
much
err
on
the
side
of
trust
until
that
trust
gets
violated
and
then
someone's
dead
to
you
and
lock
it
down.
So
where
we
could
just
like
leave
the
whole
thing
open
and
let
people
in
and
say
just
don't
do
anything
stupid.
You
got
to
commit
flag.
You
can
figure
out
what
stupid.
J
F
So
question
actually
about,
like
I,
think
for
the
drivers
and
stuff
it's
easy
with
kubernetes
I
feel
like
there's
this
chicken
egg
thing
one
and
maybe
Josh
what
you're
just
saying
kind
of
fixes
this
like
if
we
leave
the
doors
more
open,
but
historically
the
doors
have
been
relatively
closed
to
commit
or
ship
that
you
need
to
show
some
level
of
experience,
and
you
know,
let's
say
dedication
to
the
project.
How
does
that
work
with
like
a
totally
new
piece
of
code
like
kubernetes?
F
We're
like
none
of
us
can
even
review
it
that
our
or
not
even
me
but
like
the
existing
committers,
can't
review
it
because
the
patrick's
point
we
have
no
kubernetes
expertise
currently
like.
So
how
does
that
chicken
egg
problem
get
solved?
We
just
sort
of
say
like
the
two
or
three
people
who
are
contributing
it
like.
We
trust
you
and
here's
the
bit
and
you
screw
up
that
trust
like
like
Josh
said,
and
we
take
it
back.
I
mean
again
I'm
not
totally.
H
I
would
trust
them
to
work
on
the
operator
way
before
I,
trust,
myself
and
I
would
trust
myself
on
the
database
way
before
I
trust
them.
So
it's
just
kind
of
a
like
if
we're
opening
up
for
widening
the
aperture,
but
we
consider
part
of
the
Cassandra
ecosystem
in
the
Apache
foundation
inside
the
repo
with
sub
products
and
other
things,
we're
gonna
start
getting
more
diverse
voices
into
the
governance
of
the
project,
which
you
know
probably
a
good
thing.
B
G
G
B
G
H
Miss
yeah,
the
commit
thing
can
be
fixed
pretty
easily
it's
where
the
governance
comes
in.
That
gets
a
little
confusing
to
me
because
it's
like
I,
don't
think
that
be
on
the
PMC.
I
would
really
have
a
good
idea
of
what
good
governance
of
the
kubernetes
operated
for.
Cassandra
looks
like
without
educating
myself
right
governance
over
the
drivers,
like
I,
have
written
some
shitty
driver
code
in
my
time
and
that's
as
far
as
my
relationship
with
it
is
gone.
So
there
is
a
piece
of
this
that
will
probably
need
to
hash
out
and
figure
out.
H
F
If
I
mean,
is
it
worth,
you
know
reaching
out
to
some
of
them
like
we
like,
we
spoke
to
the
Hadoop
community
before
I'm
sure
the
SPARC
community
is
accessible,
I,
try
to
figure
out
what
they're
doing
and
I
guess
not.
The
last
thing
I'd
ask
is
like
this
problem
has
to
affect,
like
all
modern
Apache
projects
like
people
with
Apache
experience
here
like
is
there
any
Avenue
Nate.
A
F
I
I've
been
fascinated
by
how,
when
there's
a
kind
of
a
smaller
project,
that's
been
consumed
in
to
tinker
pop.
You
just
kind
of
a
driver,
type
thing
and
I'm
always
thinking
this.
This
person
will
clearly
make
that
you
don't
just
start
digging
into
core
and
start
learning
about
that
and
start
contributing
there,
and
they
just
never
do
like
the
thing
they
brought
in
was
the
thing
they
stuck
with
and
that's
just
never
tripped
it
out
of
that.
I
So
are
you
talking
about
the
GL
beads
there
in
that
example,
yeah
yeah
for
the
most
part
for
the
most
part
like
you
just
no
one
ever
really
makes
the
jump
out
of
that,
for
whatever
reason,
I
haven't
really
seen
that
yet.
But
it's
just
kind
of
interesting,
because
there's
like
this
concern
that
oh
there's,
a
guy
there's
somebody
coming
in
who
doesn't
necessarily
know
a
lot
about
the
system
and
they'll
start
kicking
things,
and
they
never
quite
do.
I
I
This
is
just
kind
of
just
kind
of
interesting,
even
I,
guess
even
like
some
docker
development
stuff,
like
that,
like
he
just
thought.
Folks
were
like
you
know,
I
didn't
know
anything
about
docker
until
it
kind
of
came
in,
and
you
know
once
it
did.
It
was
just
like,
oh
okay,
well
now,
I
have
to
vote
on
whether
or
not
this
docker
thing
goes
out
and
I
have
no
idea.
I
How
can
you
feel
worse
and
I
just
have
to
kind
of
sit
back
and
kind
of
trust,
the
people
that
you
know
trust
the
people
who
were
managing
and
dealing
with
it?
And
you
know
so
far.
That's
that's
work
pretty
well,
you
know
now
I
kind
of
at
least
have
a
handle
on
it,
but
yeah.
It's
just
kind
of
interesting
how
that's
developed
this
quick
point.
I
in
case
they
folks
didn't
see
it
in
the
chat.
I
I
A
A
We
we
had
a
pretty
lively
discussion.
I'm
gonna
go
back
and
re-watch
this
recording,
so
I
can
finish
up
the
notes
and
get
those
posted
right.
It's
so
hard
doing,
I
I
can't
walk
into
you've,
got
him
either.
You
know
that's
okay,
fun
to
hang
out
with
sometimes,
but
we
we
had
pretty
good
conversation.
I
I
think
some
of
the
things
that
came
out
of
this
were
trying
to
sort
out
the
sub-project
stuff,
which
is
really
I.
Think
this
is
a
good
time
to
talk
about
it
with
drivers
and
kubernetes
and
even
sidecar.
These
are.