►
From YouTube: Kubernetes Sig Docs 20180918
Description
Meeting notes: https://docs.google.com/document/d/1Ds87eRiNZeXwRBEbFr6Z7ukjbTow5RQcNZLaSvWWQsE/
The Kubernetes special interest group for documentation (SIG Docs) meets weekly to discuss improving Kubernetes documentation. This video is the meeting for 18 September 2018.
https://github.com/kubernetes/website
A
A
C
C
A
A
Folks,
yep
all
right.
Let's
move
on
to
updates
and
reminders
he's
not
in
the
meeting,
but
I
think
Brad
Topal
may
get
something
like
a
wartime
service
medal
for
being
the
PR
Wrangler
during
the
hurricane
we
did.
It
survived
like
he
offered
where
I
offered
on
Tuesday
to
take
over
his
PR
shift
last
week.
Well,
he
was
well
Florence
is
barreling
down
on
this
house,
and
it's
like
as
long
as
I
have
power.
I
have
nothing
else
to
do.
This
is
actually
good.
A
A
D
They
112
is
happening.
It's
in
full
steam.
I
think
this
is
the
last
2
weeks,
2
3
3
weeks
of
the
release,
ideally
by
this
Friday.
The
milestone
is
actually
today
but
I'm,
giving
people
who
would
like
to
have
conversations
until
Friday
and
then
I'm
gonna,
merge
everything
in
I
had
to
back
out
core
DNS,
which
is
a
bummer.
But
now
it
looks
like
based
on
the
discussion
for
the
release
meeting
today.
D
I
may
be
backing
it
back
in,
depending
on
how
so
right
now,
scalability
tests
for
5,000
and
greater
node
clusters
are
causing
out
of
memory
errors
for
accordion
s.
So
it
looks
like
it
can't
go
in
112
unless
they
fix
it
or
stick.
Auto-Scaling
says
this
is
an
acceptable
bug
and
as
long
as
we
have
something
like
an
a
dot
release
very
quickly
now
that
would
be
fine
and
I
ended
up
accidentally
hitting
the
trip
wire
that
restarted.
D
Yes,
I'm
sure
Zack
will
be
able
to
discuss
this
as
he
continues
to
look
at
his.
You
know:
I
do
okay,
it's
the
the
long
and
short
of
it
is
by
far
as
far
as
heterogeneous
is
concerned.
As
our
contributors
and
contributions
are
concerned,
they
are
very
wide
and
very
diverse
and
very
much
so
it's
like
trail
mix.
If
you
took
enough
time,
you
could
split
everything
back
out
into
its
individual
components,
exactly
I'm,
sorry
and
laugh,
but
the
heterogeneity,
I'll
mix
I
can't
wait
to
see
where
this
is
going.
Okay,
sorry!
D
Basically,
what
I'm
saying
like
is
because
the
difference
between
a
heterogeneous
mixture
and
homogeneous
mixture
is
I'm
not
just
make
sure
once
you
mix
it,
it's
mixed.
Nothing
else
can
happen.
Heterogeneous
is
like
trail
mix
as
Lisa
sort
of
started
in
chemistry.
You
can
split
everything
back
out
into
its
original
ingredients
at
the
end.
D
D
You
know,
barring
that
they've
offered
suggestions,
but
unless
we
move
to
a
release,
cadence
of
you
know
once
every
time
kubernetes
is
released,
then
I
don't
know
that
we
could
necessarily
adopt
any
of
the
same
principles
that
they
would
have,
but
I'm
open
for
discussion
and
that's
the
end
of
my
soliloquy.
If
112
is
going
great
so
who
are
some
of
the
people?
You've
talked
to
the
temp
regice
this
time
because,
like
we've
had
it's
been
various
different
contributors.
This
time
it's
been
mostly
live,
Amir
Ivanov,
who
is
tagged
as
neo
lit
one
two
three.
E
E
D
A
Yeah
yeah,
the
TLDR,
is
basically
Luba.
Mir
had
the
same
reaction
that
some
developers
do
at
realizing
that
our
strategy
is
kind
of
a
mirror
inverse
of
K
K
and
that
there
are
reasons
for
that
and
that
there
is
not
a
an
easy
or
easily
preferable
solution
to
changing
a
release
strategy
that
what
we
do
works
very
well
for
who
we
serve
I,
think
I.
A
Think
Luther
Mir
was
frustrated
because
he
had
an
open
PR
that
he
had
to
rebase
because
of
emerge
from
an
upstream
PR
and
as
anything
like
anything
happens,
that's
sort
of
annoying
II
wondered
whether
we're
possible
to
not
be
annoyed
by
it
in
the
future,
and
the
answer
is
well.
A
What
we
have
right
now
is
minimally
annoying
for
them
for
the
most
amount
of
people
and,
if
you
know
of
a
way
to
make
it
even
less
annoying,
we're
certainly
open
to
it,
but
you're,
not
the
first
and
we're
still
open
to
hearing
it.
So
Zack
handled
it
very
well,
as
always,
we're
open
yeah
to
a
better
technical
improvement
in
our
in
our
branching
and
release
strategy,
but
I
think
what
we
have
works.
A
A
A
B
With
modeling
yeah
so
just
to
give
a
little
context,
if
any
of
you
guys
are
fans
of
Big
Bang
Theory
like
Sheldon
Cooper,
has
this
like
fun
with
flags
segment?
He
does
periodically
so
I
thought
it
would
be
fun
to
have
one
for
modeling,
because
we've
kicked
off
the
the
fundamental
modeling
concepts
project
with
Dominic
who's
here,
so
we're
gonna
just
start
doing
like
a
regular
cadence
and
just
show
little
bits
of
like
things
that
we're
trying
to
clarify
and
make
sure
you
know.
Concepts
and
definitions
are
are
crisp.
B
G
G
G
Wonderful,
so
before
we
get
into
fun
with
modeling,
we
do
three
minutes
made
a
presentation
that
answers
the
question:
what
is
a
model
right
and
in
this
context,
a
model
is
an
abstraction
of
a
system
so
that
each
fact
about
the
model
is
also
affect
about
the
system,
so
the
reader
may
expect
that
every
fact
that
he
or
she
may
derive
from
the
model
is
also
effect.
That
is
the
arrival
from
the
system.
As
a
little
example,
we
have
a
system
on
the
right
side,
an
electrical
circuit
and
we
have
an
abstraction,
a
circuit
diagram.
G
So
in
this
case
the
facts
about
the
model
and
the
system
are.
There
is
an
electronic
circuit
with
two
components:
a
power
source
and
the
light
bulb.
However,
there
are
additional
effects
about
the
system
that
are
not
captured
in
the
model,
so,
for
example,
in
this
case
it
would
be
physical
dimension,
weight
operating
temperature.
You
only
can
derive
these
facts
if
you
study
the
system,
but
not
the
model.
G
G
However,
if
one
person
studies
two
system
and
has
to
communicate
his
or
her
mental
model,
vir
system
model
information
transfer
suffers
from
encoding
loss.
I'm
a
sloppily
create
a
system
model,
may
forget
some
details
and
may
not
explain
it
well,
and
it
also
suffers
from
encoding
loss,
which
means
that
as
a
receipt
decoding,
loss
decoding
loss
correct.
G
Thank
you
that,
as
a
as
a
receiver,
I
may
not
understand
what
was
Bell
explained
to
me,
which
results
in
the
fact
that
for
the
first
person
based
on
our
operating
assumption
capable
person,
the
mental
model
matches
to
system,
but
for
the
second
person
the
mental
model
does
not
match
the
system
at
all,
and
therefore
second
person
cannot
reason
well
about
the
system
in
its
behavior.
So
what
we
are
trying
to
do
is
to
minimize
encoding
loss
and
decoding
loss
by
complete
and
accurate
conceptual
models
and
also
very
applicable,
complete
and
accurate
for
memorials.
G
So
that
sets
the
stage
and
then
we
can
jump
right
in
into
fun
with
modeling
and
today
I
want
to
talk
about
imperative
versus
declarative.
When
we
read
about
kubernetes
or
when
we
talk
about
kubernetes,
we
usually
state
kubernetes
is
a
declarative
system
and
all
of
us
have
somewhat
like
a
picture
in
mind.
What
does
that
actually
mean
right?
But
if
you
ask
somebody,
what
does
it
mean?
G
It
is
actually
a
concept
that
is
hard
to
explain
so
for
this
presentation,
I
want
to
look
at
a
simple
model
of
computation,
and
that
is
we
transition
through
states
over
time.
That
is,
we
start
in
a
status
and
applying
a
command
C
to
state
s,
transitions,
the
system
to
a
stick,
similarly,
applying
see
tick
to
s,
tick,
transitions
assistant
or
a
stick,
tick
and
so
forth.
G
Now,
when
we
talk
about
declarative,
let's
start
with
imperative
in
an
imperative
system,
the
user
has
a
desired
state
in
his
or
her
head.
I
know
where
I
want
to
be
at,
and
it
is
now
the
responsibility
of
the
user
to
determine
the
set
of
commands
that
transitions,
the
state
of
the
system
from
the
current
state
to
the
desired
state.
G
Now
to
contrast
that,
with
the
declarative
system,
it
is
a
systems
responsibility
to
determine
the
set
of
commands
that
transition,
the
state
of
the
system
from
the
system's
current
state
to
the
desired
state
and
in
the
context
of
a
declarative
system,
the
desired
state
does
not
only
exist
in
the
user's
head.
It
also
exists
in
an
encoding
form
as
an
input
to
the
system,
and
now
the
system
is
split
in
two
components:
a
controller
and
the
already
previously
present
process.
G
But
that
still
leaves
us
one
question.
We
hear
a
lot
about
control
loops
as
well,
and
currently
there
is
no
loop
in
this
right,
the
controller
does
it
once
the
processor?
Does
it
once
and
we
are
in
a
blissful
state
eternally?
However,
reality
usually
is
not
as
fun
so
a
control
loop,
then,
is
a
continuous
process
that
accepts
a
representation
of
the
desired
state,
determines
a
set
of
commands
and
performs
a
set
and
performs
a
set
of
commands
to
transition
the
current
state
of
the
system
to
the
desired
state
of
the
system.
G
In
other
words,
the
control
loop,
is
a
controller
that
performs
its
function
continuously.
Therefore,
compensating
adverse
state
changes
now,
what's
an
adverse
state,
changed
adverse
state
changes
or
all
changes
to
the
state
that
are
not
performed
by
the
processor
context
of
kubernetes
container
goes
down.
A
node
goes
down
now.
To
summarize
this
right,
the
difference
of
a
declarative
versus
I'm,
sorry
of
an
imperative
versus
declarative
system
is
who
is
the
driver
right
in
an
imperative
situation?
I
had
the
desired
state
in
my
head
and
I
am
the
driver
of
the
system
in
a
declarative
system?
G
I
have
the
desired
state
of
my
head,
and
the
system
has
it
as
an
input
in
an
encoded
form
and
the
system
itself
or
more
specifically,
a
component
of
the
system,
is
the
driver
of
the
system,
and
if
we
take
that
one
step
further,
we
actually
see
that
this
fits
neatly
into
the
world
of
kubernetes,
where
I
do
not
have
only
one
controller.
I
have
multiple
controllers.
G
The
processor
in
this
case
is
the
API
server
that
accepts,
commands
and
alters
the
kubernetes
object
store,
so
the
kubernetes
objects
accordingly,
and
the
controller's
act
in
a
control
loop
run
continuously
and
are
therefore
capable
of
reacting
to
adverse
state.
Changes
like
a
node
goes
down
or
a
container
goes
down,
and
that
concludes
today's
episode.
I
give
you
a
little
bit
of
an
outlook
for
next
week's
episode,
which
is
titled
the
great
lie.
G
We
were
talking
about
declarative
versus
imperative
systems
today
and
saw
how
kubernetes
is
indeed
a
declarative
system,
but
next
week
we
actually
step
down
into
the
nitty
gritty
details
of
kubernetes
and
explore
where
there
are
actually
mismatches
in
the
mental
model
and
also
in
the
communication,
when
we
say
kubernetes
is
a
declarative
system.
Where
does
his
statement
hold
up
and
where
does
his
statement
not
hold
up?
Thank
you
very
much.
Oh
actually,
I
want
to
take
the
chance.
G
If
you
do
not
mind,
I
have
I
create
a
document,
an
RFC,
a
request
for
comments
about
a
kubernetes
conceptual
overview.
So
this
is
a
high-level
model
of
the
structure
of
kubernetes,
its
individual
components
and
connections.
If
you
don't
mind,
I
would
like
to
send
this
out
later
to
everybody,
and
if
you
have
any
comments,
concerns
please
do
comment
to
this
request
for
comments.
So.
B
I'm
going
to
add
some
instructions
to
the
RFC
to
give
a
little
context,
because
what
we're
looking
for
is
technical
accuracy.
This
is
to
help
validate
the
models
that
Dominic
is
coming
up
with.
So
one
if
you're,
you
are
familiar
with
the
the
details
of
communities
like
make
sure
it
does
reflect
what
you
believe
of
the
the
system
right,
but
then
to
also
like.
Does
this
help?
You
understand
communities
better?
B
Does
this
you
know
help
clarify
your
mental
model,
that
sort
of
thing,
so
those
are
the
two
main
types
of
feedback
we're
looking
for
and
Andrew.
How
do
you
plan
to
distribute?
How
do
you
plan
to
distribute
this?
Where
will
it
be
available?
So
I
was
planning
on
sending
a
link
to
the
sig
Docs
email
list,
and
then
we
can
also
send
it
more
broadly
to
the
engineers
and
the
other
kubernetes
email
lists,
and
that
can
also
put
it
on
flak.
Can.
B
H
H
B
So
I
can
answer
some
of
those.
So
as
we're
going
through
the
work
I
mean
you
know,
we
we
realized.
We
need
to
clarify
certain
things
like,
for
example,
the
imperative
versus
declarative,
because
that's
a
really
foundational
thing
so
I
will
be.
You
know,
working
on
a
doc
plan
that
will
once
you
know,
as
we're
encountering.
These
things
will
organize
like
what
it
is
that
we
need
to
modify
and
add
to
the
to
the
Korean
a
zyo
doc
set,
and
then
some
of
them
could
actually
be
blog
post.
They
don't
necessary
in
the
documentation.
B
So
we
have
to
kind
of
determine
you
know
with
the
appropriate
channel.
Is
for
that
and
then
I
think
what
I
was
going
to
propose
after
Dominic's
presentation
was
to
turn
the
working
group
that
regularly
meets
on
Friday
to
be
the
modeling
group,
so
people
can
join
in
on
that
and
we
can
figure
out
like
you
know
how
you
know
that
how
we
want
to
operate
and
like
dissipate.
This
and,
like
you
know,
collaborate
on
this
stuff.
H
That's
great,
it
means
straight.
The
range
of
diagrams
could
grow
to
such
a
point
that
it
becomes
difficult
to
know
which
diagrams
should
go
where
or
how
to
gracefully
incorporate
them.
So
having
some
kind
of
plan
full
universe
in
the
grain
of
sand,
for
how
this
will
turn
into
kubernetes,
diode
docks
I
think
will
be
focusing
function
for
everyone
and
for
everyone
in
the
group.
A
B
I
I
A
G
G
Well,
if
you
do
not
mind
for
me
to
ask
a
question,
so
please
do
think
about
as
I
think
Andrew
or
Tom
is
from
stated.
We
can
make
this
somewhat
regular
occurance
about,
let's
say
random
topic
fun
with
modeling
all
about
kubernetes.
Please
do!
Let
me
know
if
you
think
this
is
viable
for
your
time
or
if
you
think
that
this
meeting
is
not
the
correct
stage.
Word.
A
I
think
it's
great
to
have
a
presentation
in
review
I
like
Andrews
idea
of
having
a
working
group
specifically
to
dive
in
and
have
time
that's
more
dedicated
to
the
material
itself,
but
in
terms
of
reviewing
progress
and
getting
sort
of
a
high-level
overview.
I
think
periodic
updates
in
weekly
meetings
are
great
good.
A
I
Yes
Luke.
He
wanted
me
to
check
if
there
were
any
tools,
free
tools
that
we
could
use,
and
these
were
the
two.
The
two
tools
that
I
put
up
on
the
sake,
Docs
tooling
channel,
are
the
two
tools
that
seem
to
have
some
good
reviews.
All
of
you
I
invite
all
of
you
to.
Please
take
a
look
at
it.
Please
post
your
suggestions,
comments
or,
if
you
know
of
any
other
tools,
that
would
be
better.
A
I
B
A
Right
so
I
would
say:
let's
review,
let's
review
these
tools
and
revisit
this
in
one
week.
I
know
you
mentioned
earlier:
Luke,
that's
a
Luke
Perkins
and
he's
not
able
to
be
with
us
this
week.
So
I
would
say:
let's
defer
for
one
week
and
revisit,
and
hopefully
this
gives
us
time
to
review
these
tools
and
have
a
more
informed
discussion.
A
So
we
had
originally
talked
about
doing
just
passing,
regular
commits
back
and
forth
between
branches
and
the
the
content
that
we're
passing
from
a
website
out
to
internationalization
repos
is
sufficiently
different
from
the
content
that
those
repos
are
passing
back
in.
In
return
that,
just
you
know,
a
straight-up
PR
and
a
strategy
of
just
making
opening
PRS
and
each
other's
repos
and
merging
that
way
seem
pretty
straightforward.
A
Zack
suggested
that
we
take
a
look
at
the
set
subtree
strategy
to
reduce
the
just,
reduce
the
complexity
even
more
and
the
more
that
I
look
at
especially
like
right
now
we
have
three
internationalization
projects:
open,
Korean,
Japanese
and
Chinese,
and
the
like.
The
Korean
team
they're
super
far
ahead.
A
They're,
like
their
comfort
with
the
toolchain,
is
very
high,
and
the
the
Chinese
team,
for
example,
is
still
still
struggling
with
prowl
still
trying
to
find
like
the
right
fit
and
structure
for
how
to
do
their
do
their
internationalization
work
inside
of
the
kubernetes
org,
so
anything
that
we
can
do
to
reduce
barriers
for
internationalization
teams
as
part
of
our
structure.
I
think
is
a
good
thing,
so.
A
The
the
question
of
our
branching
and
merging
strategy
for
internationalization
is
still
an
open
one.
I
guess
it's
the
TLDR
for
that
I'm
working
right
now
on
a
prototype
of
what
that
what
a
subtree
set
up
would
actually
look
like,
and
the
answer
so
far
that
looks
really
cool
in
theory,
but
getting
to
looking
really
cool
in
practice
may
be
difficult
or
maybe
straightforward.
That's
what's
this
dedicated
to
is
getting
some
sense
of
how
much
signal
loss
occurs
between
cool
in
theory
and
cool
in
practice.
A
Say
I
think
like
another
working
group
meeting
is
worthwhile
just
because
there
are
some
things
that
would
be
really
good.
I
think
it
would
be
effective.
We
could
pass
through
really
quickly,
so
I
will
set
up
a
working
group
meeting
for
branching
strategy
again
right
now.
That's
just
Andrew
me
and
Zack,
but
if
you
are
interested
in
branching
strategy
for
internationalization
Rico's,
even
just
as
a
listener,
you'd
be
welcome
to
come
join
us.
So
if
you
are,
if
you
are
interested
now
well
actually
here,
let
me
just
pull
it.
F
A
Right
now
we're
focusing
there
are
no
Spanish
localization
projects
in
the
works,
but
if
you
wanted
to
drive
one
congratulations,
you
have
found
the
right
do,
but
the
the
questions
that
we're
answering
are
going
to
be
broadly
applicable
to
all
internationalization
projects,
because
we're
talking
about
the
process
for
internationalization
projects
as
a
whole,
how
they
move
back
and
forth
between
their
repos
and
and
key
websites.
So
a
GB,
if
you
want
to
do
I,
can
only
send
you
an
invitation.
A
A
A
A
So
I'd
like
to
revisit
this
again,
have
a
have
a
sig
meeting
where
we
talk
about
and
set
our
goals
and
priorities
for
the
coming
quarter
in
December
at
coupe
con
in
Seattle,
so
by
a
quick
show
of
hands
or
like
a
ping
in
in
the
chat
channel.
Who
is
planning
on
being
in
Seattle,
okay,
Tom
Brad
Zak
me
Steve,
Jim.
Are
you
gonna?
Are
you
planning
on
being
there
Svetlana
Raji?
A
A
Awesome
Jim,
Oh,
excellent,
gee,
I
hope
you
get
it.
That's
they're,
usually
pretty
good
about
that
cool.
A
Yes,
so
I
would
like
to
I'd
like
to
focus
on
having
more
concrete
goals
going
forward.
We
I
think
we
did
a
lot
of
process
stuff
for
about
who
we
are
what
we
value
in
May
and
I'd
really
like
to
focus
on
on
specific
goals
and
specific
projects
in
December
and
go
forward
after
that.
So
does
that
see
folks
nodding,
but
does
that
sound
reasons
that
sound
like
a
good
idea
to
folks?
Does
that
sound
like
worthwhile
to
invest
that
kind
of
work.
A
H
I
think
it's
a
great
idea
and
that
I'm
kind
of
in
the
witness
protection
program
here
hold
on
there.
We
go
I
think
it's
a
great
idea
to
have
some
sort
of
True
North
ideas
for
where
we
want
to
go
and
what
we
want
to
work
on,
because
this
project,
you
know,
there's
there's
so
much
going
for
it,
but
there's
so
much
going
on
going
on
with
it
that
when
everybody
just
does
a
little
bit
of
everything,
it's
like
adding
drops
to
the
ocean.
You
know
and
I
think
sorry.
H
H
If
we
focus
on
a
few
big
things,
then
we
might
be
able
to
see
some
significant
changes
if
we
scope
those
things
based
on
and
apparent
need
that
there's
evidence
for
you
know,
and
so,
but
then
part
of
that
is
like
how
do
we
actually
do
that
and
we
agreed
that
there
are
a
lot
of
nice
things
we
should
do
and
we
need
some
kind
of
structure
that
results
in
those
things
getting
done.
I.
D
Think
we're
fortunate
enough
to
have
a
large
enough
and
diverse
enough
group
of
people
inside
of
this
sake
that
it
is
like
if
I
were
to
offer
a
word
I
super-sick
in
the
sense
of
like
you
know,
it's
not
just
people
who
maintain
this
one
aspect
of
kubernetes,
like
literally
everybody,
reads
the
dot.
Well,
that's
not
true.
A
lot
of
people
read
Docs
and
a
lot
of
people
contribute
to
the
docs.
D
A
H
A
J
A
So
we
spoke
with
flood
schlossberg
and
I
have
not
turned
on
any
BOTS
personally,
but
I
noticed
that
other
folks
have
turned
on
BOTS
in
like
the
sigmax
channel
sigmax
tools.
So
at
this
point
to
be
honest,
I'm
not
sure
how
to
turn
them
off,
even
if
I
wanted
to.
So
it
appears
that
we
are
going
forward.
A
D
See
how
it
goes
if
I
may,
as
the
essence,
we
sort
of
made
the
unilateral
decision
I
apologize
about
talking
anyone
first
to
invite
it
to
our
channel.
My
part
of
my
thought
process
from
the
community
meeting
in
last
week
when
he
presented
to
the
wider
broader
communities
community
was
as
a
sig
as
as
sort
of
briefly
outlined
earlier.
We
have
interesting
pull
request,
guidelines
and
things
and
common
gotchas
that
we've
all
answered
for
many
people
before
in
the
sig
Docs
channel.
D
I
Yeah
thanks
Zack,
yes,
I
did
catch
a
fat
round
about
it.
I
tried
some
of
the
questions
in
our
sig
Docs,
for
example,
how
do
I
contribute
and
the
bots
didn't
seem
to
give
the
relevant
answers
so
when
I
did
catch
a
fat
weld
offline,
he
mentioned
that
we
may
have
to
provide
the
answers
and
add
it
there.
So,
yes,
we
we
could
contribute.
A
You
know,
I
am
all
in
favor
of
participating
in
in
a
bot
experiment
that
relieves
the
burden
of
some
of
the
most
common
questions,
but
I
still
have
not
received
a
satisfactory
answer
about
what
so
kubernetes,
and
indeed
all
of
the
work
of
the
Linux
Foundation
is
on
open
source
projects.
What
is
focus
profit
model?
Are
they
taking
data
data?
That's
open
right
now
and
funneling
it
into
a
model
that
will
consider
that
a
proprietary
that
hasn't
been
answered
to
my
satisfaction,
I,
am
disinclined
by
nature.
I
follow
the
lazy,
programmers
school
of
ethics.
A
I
D
D
I
F
We
need
to
be
a
little
bit
careful
here,
I
mean
you
know
things,
things
get
started,
perhaps
good
intentions,
but
but
not
without
you
know,
being
fully
thought-out
and
then
then
then
people
think
oh,
it's
it's
official
and
that's
part
of
the
thing
and
then
you
know
people
forget
how
it
really
came
to
be,
and
you
need
to
be
a
little
bit.
Careful
I
think
with
these
things,
because
otherwise
we
end
up
with
legacy
stuff,
and
nobody
knows
why
it's
there
does
it
make
any
sense.
A
Yes,
so
let
me
pose
this
a
different
way.
Is
there
anyone
who
was
wildly
in
favor
of
continuing
with
focal
in
in
the
channels
that
cig
docks
is
in
custody?
If
you
are
in
favor
of
let's
do
you
simple,
thumbs-up
thumbs-down?
If
you
are
in
favor
of
focal
thumbs-up,
if
you
are
in
favor
of
folklore
thumbs
down.
C
That's
for
my
customers
like
bi-weekly
basis.
Now
it
does
I
loved
all
your
point,
Zach
about
the
lazy,
the
lazy
protocols,
I
will
say:
I
just
want
to
plant
the
see
that
if
you
guys
do
find
that
there
is
a
process
that
you
can
put
stuff
in
a
spreadsheet
and
then
I'm,
obviously
speaking
to
cutter.
So
maybe
this
doesn't
work,
but
if
you
found
that
there's
a
very
important
thing
that
needs
to
be
out
there,
that's
not
in
documentation
and
you
want
to
lazily
roll
it
out.
C
A
E
E
You
know
like
in
a
console
or
Anna
and
slack
have
they've
just
they
fragmented
our
our
effort
to
provide
complete
information,
and
they
tend
to
you
know
not
get
maintained,
and,
and
so
not
only
do
we
have
two
places
where
there's
information,
but
one
or
both
of
those
places
get
out
at
a
date.
So
I
I,
just
I,
adhere
to
this
idea
that
we
want
to
concentrate
our
efforts
on
one
and
only
one
place.
I
A
It
is
at
least
partially.
There
are
I
think
Vlad
explained
that
there
are
three
sources:
there's
the
documentation
website,
natural
language,
learning
from
the
channels
themselves
and
Stack
Overflow.
So
we
have-
and
we
don't
really
know
what
percentage
that
is.
I
mean
if
I
can
I
think
Vlad
showed
us
a
diagram
of
what
that
looks
like,
but
we
don't
really
have
any
control
over
that
and
we
would
also
with
focal
we're
also
hosting
and
promulgating
content
from
sites
and
sources
that
aren't
our
docks.
So
it
gets
down
to
what
do
we
want
to
do?
A
Do
we
want
to
do
BOTS
that
pull
information
from
natural
learning
and
the
internet
well,
or
do
we
want
to
focus
our
effort
on
maintaining
our
docks,
well,
I
and
given,
given
that
we
have
given
that
we
struggle
sometimes
with
the
resources
to
take
on
some
of
the
most
pressing
projects
for
our
own
documentation?
This
doesn't
seem
like
a
wise
idea
for
distribution
of
energy
and
resources.