►
From YouTube: GMT 2018-03-20 API WG
Description
No description was provided for this meeting.
If this is YOUR meeting, an easy way to fix this is to add a description to your video, wherever mtngs.io found it (probably YouTube).
A
A
A
B
B
C
D
D
D
I've
already
noticed
emails
go
out
and
I
asked
Benjamin
to
send
out
an
email
about
lunch
chain
of
his
recently
and
I
think
actually
I'm
realizing
now,
since
I
sent
that
email,
how
many
small
API
changes
get
made
without
any
real
review,
so
I'm
hopeful
that
we
can
get
a
little
more
strict
about
that.
Maybe
and
even
when
we're
making
smaller
API
changes
just
reach
out
to
the
community
and
Steve,
it's
gonna
impact
people.
F
E
Think
I,
don't
think,
have
enough
scrub
enough
date
has
really
altered
a
structure
in
a
structured
way
like
a
one-hand.
We
all
I
had
any
people
pray
for
a
more
free,
slightly
more
frequent
releases
because
it
makes
getting
a
high
priority
feature
easier
faster.
E
E
Well,
all
the
way
to
apply
I
have
almost
every
minor
version,
some
kind
of
format,
every
matter
where
he's
running
in
different
clusters,
I
think
that
each
class
that
we
keep
will
keep
them
consistent,
but
across
different
transference
we
have
old
machines,
that's
really
hard
to
make
to
upgrade
so
I
think
I,
think
being
able
I'll
agree
with
EGA
that
being
able
to
check
this
track.
This
correctly,
like
really
having
a
correct
expectation
about
when
to
get
back
to
the
lease.
Is
that
important?
E
If
we
say
we
want
to
do
frequent,
and
we
can
do
it
that's
better,
but
if
we
don't
feel
like,
we
can
actually
do
that.
That's
a
community
like
we
need
more
time
to
really
making
sure
every
released
is
there
in
high
quality
I'm
okay
to
do
is
less
frequent
as
long
as
what
we
say
on
the
document
actually
reflects
what
we
do.
G
E
We
apply
the
work
we
either
to
ultimate
when
there
is
high
priority
features.
Oh,
we
release
the
first
point,
fixed
version
like
one
point,
for
example,
one
point
four
point:
one
lump
like
3.1:
we
either
release
that
100
to
one
point
three
point:
zero
one
point
X
point
zero,
but
only
if
there
is
some
kind
of
high-priority
feature
there
that
we
really
need
to
use.
Yeah.
H
I
did
it
for
CSI
stuff,
because
extra
education
we
did
for
other
releases
like
do.
We
have
some
notion
of
why
it
took
so
long
for
us
like.
Are
we
trying
to
get
justice
a
powerhead
release
itself
that
we
think
it's
not
as
well
over
two
months
or
are
we
like
always
trying
to
block
it
for
some
feature
to
get
into
not
really
time
based
and
not
cutting?
Or
is
it
like?
What
do
you
have
a
good
sense
of?
What's
the
fundamental
reason
like
when
stay
in
four
months
is
it
example.
H
Are
you
cutting
the
RCEP
months,
but
by
the
time
it
actually
gets
taken,
one
more
one.
That's
had
another
month
for
the
voting
attesting
stabilizing
of
candidate
is
happening
like
what
I
mean
to
me.
Ideally,
we
should
be
even
faster
releases
as
long
as
overhead
of
doing
really
small.
If
we
can
minimize
that
because
we
are
to
be
it's
alien,
people
can
jump
questions,
so
they
don't
really
have
to
upgrade
to
a
version
that
they
are
not
comfortable
with
or
whatever.
If
their
cadence
is
six
months,
they
can
skip
three
versions
on
chopper
2.
H
It's
like
we
don't
mandate
that
they
have
to
jump
to
it.
Each
version
for
people
who
want
to
move
faster
they
can
move
faster
at
KC.
2
month
is
like,
if
you're
going
to
make
it
even
longer.
It
can
tell
you
people
that
they
have
to
wait
at
least
that
long
to
get
a
new
feature,
but
if
you
do
it
faster
and
then
I
think
it's
more
flexible
for
people
who
want
to
go
slower,
you
go
slow
because
we
have
a
compatible,
both
Academy
for
people
who
couldn't
go
fast.
H
They
can
go
fast
if
they
want
to
in
the
neck
startups
and
they
don't
really
care
about
making
some
kind
of
change
just
for
whatever
the
cadence
is
like
facts
like
about
Twitter.
It
should
be
here
two
weeks,
but
that's
because
a
lot
of
my
skills,
expertise
in-house
and
we
can-
you
know
water
engine.
So
that's
another
extreme
like
in
my
mind,
as
long
as.
H
I
H
D
J
Depends
on,
as
mentioned
like
the
so
for
now,
if
we
pepper,
1
5
1
5
eggs
is
straightforward
but
like
if
we
dip
up
to
like
1
3,
it
might
be
conflict.
You
have
to
adjust
xn
person,
their
son,
like
style
change,
some
like
finger
improvement,
so
those
can
be
searched
and
all
the
project
in
most
of
the
component.
So
we
have
adjusted
small
country
but,
for
example,
if
you
in
1
3
&
1,
we
have
suddenly
process
changes.
We
have
some
like
group,
look
information.
We
have
some
life
if
sunlight
repair.
H
It
would
still
be
six
months
going
back,
so
the
amount
of
code
Charlotte's
there
in
the
budget
will
still
be
the
same.
Everybody
it's
the
same.
Difficulty
like
even
if
it
just
to
is
it's
from
me,
then
I'd
still
have
been
that
it
was
a
shame
that
you
need
to
report
it
to
that
six
month
old
release.
You
might
have
to
do
it
twice,
maybe,
but
they
won't
have
Cochin,
that's
going
to
happen
in
the
sixth
months,
which
is
hard
enough.
Flash
period
for
a
release.
H
B
I
J
J
And
then
we
have
some
new
pub
and
then
people
would
divide
about
is
a
WLAN
regression
with
active
fix+
is
so
that
part
might
be
the
biggest
part
of
which
we
have
the
release,
delay
and
I
don't
feel
like
it
has
been
happening
for
not
just
4
1
5.
Then
it's
been
happening
for
a
couple
days
and,
like
so
I,
think
I
think
we
have
some
fun
cushion
before
it
is
like
one
way
to
Rufus.
This
might
be
like
B.
J
H
H
H
J
J
I
almost
gonna
do
the
cut
for
the
one
five
most
of
the
talk
about
some
of
its
some
of
the
early
some
of
the
local
park
there
from
the
new
feature
which
we
introduced
for
ice
is
more
likely.
We
have
been
fighting
how
one
more
new
feature
park.
If
we
wait
for
that
feature,
which
is
another
factory,
would
make
the
release
delayed.
D
D
H
B
E
So
this
item
is
mostly
about
reviewing
the
new
API
that
I'm
proposing
to
add
to
support,
resizing
or
volumes
so
so
out.
I'll
skip
them
so
the
background
that
we
added
prison
bottoms
about
I
think
at
least
two
years
ago
a
user
were
never
able
to
resize
up
since
volume
once
it's
created.
Hoover
has
some
really
high
priority
tasks
that
we
realized
this.
If
this
caused
a
lot
of
Pisgah
space
waste
in
our
cluster,
because
we
cannot
really
sizing
your
purchase
volume
correctly
in
the
very
beginning
of
a
database
long.
E
H
E
H
E
H
H
E
People
can
figure
things
out
from
the
operator
API
like
there's
no
risk
condition,
assuming
there's
no
risk
envisioned
and
I'm
here
the
same
time,
and
we
will,
I
think
the
operation
will
see
an
error
at
the
validations.
I
think
I'll
friction
will
see
an
error,
so
if
the
task
is
being
used
by
other
thing,
but
if
the
volume
is
you
spend
different
tasks,
yeah.
H
E
H
E
E
E
E
E
E
So
there
was,
there
was
I
can't
think
of
three
different
options.
One
needs
Express
the
size
difference
in
as
a
resource
object,
which
should
also
include
correct
allocation
information,
reservation
info
correct
the
raw
that
the
resources
been
with
your
vendor.
A
second
option
would
be
simply
put
the
difference
as
a
scalar
value
and
omit
all
the
additional
informations
for
the
person's
modem
case.
We
can
still
be
in
front
of
out
because
it
has
to
be
consistent
with
the
volume
and
the
third
would
be
expressed,
a
target
volume
in
the
resource
format.
E
L
G
Actually
so
I
actually
have
different
opinions
or
grow
volume
in
finger
and
submitting
these
options
for
shrink
volume.
I
feel
number
two
is
easier,
because
so
we
just
specify
that
either
the
net
space
we're
on
the
free
or
the
file
sizes,
okay,
but
we
we
can
that
either
agent
or
in
the
future,
the
mrs.
provider.
Besides,
what
would
be
the
result
in
previous
books,
but
it
can
be
a
nun
of
hath
disk.
It
can
be
raw
if
you
are
shrinking
from
out.
This
depends.
G
That's
that's
the
reason
that
I
I
would
suggest
not
we
don't
specify
the
target
result,
but
for
grow
volume,
I
have
a
different
opinion
and
that's
why
I
feel
it's
better.
If
we
can
split
this
resizing,
the
two
different
calls,
so
the
different
between
shrink
and
grow
is
that
train
takes
one
volume.
We
create
two
and
therefore
grow
on,
and
we
take
two
input
volumes
and
merge.
K
E
G
M
A
A
G
Let
me
Connie
so
for
shrink
volume.
My
thought
is
that
we
take
a
persistent
volume
and
pass
that
into
either
agent
or
the
research
provider
and
ask
whoever
to
shrink
it
optionally.
They
might
be
a
by-product
that
represent
the
resource
of
the
freed
space
and
it's
up
to
the
either
agent
or
the
resource
provider.
To
present
that
resource.
It
can
be
another
path,
volume
or
it
can
be
converted
to
raw
volume
in
C,
psych
plugin.
G
So
that's
why
I
suggest
that
for
shrink
volume
we
should
take
the
input
resource,
but
with
an
additional
excise
either
the
tummy
size
or
the
free
space.
You
want
to
remove
a
shrink
and
left
the
agent
or
the
supervisor,
decide
what
are
the
result
looks
like,
but
for
grow
volume.
My
opinion
is
that
groom
takes
the
original
persist
imodium,
as
well
as
an
additional
resource.
G
So
we
took
these
as
two
inputs
and
try
to
merge
them
and
four
persist
volumes
as
to
how
sad
for
the
agency
for
default
resources,
we
can
infer
that
what
that
the
additional
resource
must
be
of
the
same
type
of
the
element
Tony
grow,
except,
except
that
it
doesn't
have
any
persistence.
But
in
future,
if
we
want
to
use
the
same
call
to
extend,
say
a
CC
volume,
we
may
take
a
mount
volume
as
well
as
a
row
of
storage,
poor
resource
and
ask
the
CSI
plug-in
to
convert
to
two
to
grow
to
grow
them.
G
So
in
this
case,
this
not
is
not
true
that
we
can
completely
infer
the
type
or
the
other
attributes
of
the
additional
resource
ardougne
use
unless
we
specified
the
whole
the
whole
resource
product.
That's
why
I'm
I'm,
leaning
towards
like
we
have
the
two
separates,
calls
and
have
slightly
different
signatures
for
these
two
calls.
Well.
G
M
E
Volume,
that
is
the
case
for
the
CSI
volumes.
My
argument
is
this:
cars
not
going
to
be
symmetric,
say
if
you
can
make
a
volume
class
difference
and
you
make
a
new
if
you
cannot
take
useful
I
mean
what
I'm
trying
to
what
I'm
trying
to
achieve
here
is
try
to
make
these
to
call
more
or
less
a
Simic
to
each
other,
don't
I
think
otherwise.
The
framework
author
will
have
quite
some
confusion
about
how
to
construct
these
calls.
Yeah.
G
But
we're
gonna
use
symmetric
calls.
Then
then
we
can
actually
go
use.
A
Jain,
peaches,
J
species
idea
about
just
one
single
resize
call,
but
if
you
just
want
to
reduce
the
complexity
at
all
like
the
coming
off
the
Pembroke,
then
one
costs
of
like
one
recess
coal
would
be
sufficient.
But
this
recommendation
is
that
well
it
cannot
specify
what
we
like
specified
this:
the
type
of
the
additional
resources
we
only
use
in
the
future.
G
D
Yeah
I
agree.
It
would
be
nice
to
avoid
new
calls
just
for
CSI,
and
so,
if
I,
if
I
understand
number
one
correctly,
we
would
just
Express
the
size
difference
as
a
resource,
so
we
still
don't
have
to
like.
We
don't
have
to
specify
the
resulting
resource
that
comes
out
of
the
operation.
Is
that
accurate?
Yes,
that's
accurate.
G
D
M
E
For
the
persons,
but
in
case
I,
think
if
I
would
be
the
framework
author,
it's
going
to
be
like
I,
make
a
copy
of
the
original
volume
clear.
The
positions
part
change
the
size
and
appended
on
to
the
object,
either
that
all
the
scan
through
may
offer
find
a
resource
object,
which
has
some
kind
of
matching
method.
Matching
information
append
it
into
there.
E
E
G
G
G
If
we
don't
only
specify
size
for
growing
a
volume,
we
are
losing
the
ability
to
say,
I
want
to
close
on
and
from
this
particular
resource,
or
that
particular
resource
and
another
points
that
g20
allying.
The
review
of
that
half
is
that
we
also
if,
by
making
the
call
like
the
resource,
making,
is
clear
that
these
resource
our
input
resource
and
we
can
do
validation
that
we
want
to
yeah,
do
all
like
just
that.
G
The
frameworks
like
which
resource
on
the
merge
and
that
the
master
do
all
the
validation
to
check
the
types
the
in
other
things.
That's
that's
why
I
think
for
Crowe
volume,
it's
to
more
feature,
prove
if
we
can
have
for
resource
or
like
specify
the
full
result.
Promo
for
the
additional
not
just
assigns
a
scalar
yep.
G
D
C
D
E
D
G
E
Also
looking
looking
at
other
three
combinations
here
wrote
out
here
to
see
for
a
CSI
volume
if
we
shrink
them
who
the
results
come
back
in
this
format,
the
case
again
was
a
last
last
question.
Last
question
is
for
the
three
things
you
list
out
here.
Well,
if,
following
the
roots,
this
practice
for
multi-disc
would
the
difference
also
coming
back.
It's
the
same
resource
object.
G
E
That's
the
case
right.
It's
not
like
I
said
if
volume
is
among
disc
edition
should
be
a
rod.
A
scrum
same
result,
providers
in
profile.
That
means,
if
it's
about
this
and
we
shrink
it,
will
shoot
expecting
new
raw
disk,
a
same
raw
disk
from
same
result,
provider
and
same
profile.
Coming
back
as
an
ad
in
a
different
coming
back
in
a
different
resource
object,
we
offer
yeah.
G
E
D
K
D
E
How
much
did
you
use
so
it's
I
think
I
think
they
may
even
have
different
workflows
about
the
different
API,
so
I
think
in
terms
of
even
helium
they
may
get
the
I
think
they
need
some
clarification
right,
good
comments
to
explain
not
only
whether
field
what
each
field
should
be.
Also,
while
it's
a
ritual
in
this
way
and
with
that
I
think
they
can,
it
should
be
approachable.
G
So
if
you
actually
want
to
make
this
call
symmetric,
I
think
we
it's
okay,
I,
think
I'm,
okay
with
like
making
think,
also
sapped
a
resource
for
love
as
a
difference
just
that
we
were
going
to
just
extracted
size
and
nothing
else.
So
it
seems
a
little
redundant
to
me
for
freeing
shrinking
about
him.
But
if
you
feel
like
it's
way
easier
for
a
framework
developer
to
have
a
symmetric,
maybe
is
okay,
I,
don't.
E
Think
it'll
be
sick,
David.
Second,
it
complains
easier.
I
think
this
is
actually
effect
operator.
Ap
is
more
because
the
operator
cares
I've
seen
people
hand
crapped
out
trees
on
object
to
send
to
us.
That's
where
the
really
mistakes
happening
for
the
framework
has
everyone
writing
metal
framework?
Most
of
us
using
strong,
typed
language
with
the
generate
findings
there.
It
here
will
be
the
wrong
thing:
it's
it's
not
like
they're
called
Freeman
compiled
and.
D
What
about
if
we,
if
we
make
them
symmetric
and
allow
a
shrink
to
take
a
resource,
we
could
document
the
API
such
that
we
explicitly
tell
users.
You
just
need
to
give
us
a
disk
resource
with
a
scalar
size
and
that's
all
we're
gonna
pay
attention
to
just
send
us
any
disk
resource
with
a
size.
We're
going
to
use
that
size
to
shrink
your
volume,
and
that
might
be
one
way
to
kind
of
eliminate
difficulties
that
operators
would
have
in
constructing
the
call.
D
D
E
E
G
Also
with
a
single
call
like,
we
still
need
to
explain
that,
okay
for
growing
volume,
where
you
already
expect
everything
in
the
prover
you
provide
for
thinking
volume,
we
only
look
at
the
size
and
we
don't
care
about
anything.
You
can
not
rely
on
the
love.
You
provide
to
say
that
you're
expecting
a
free
resource
with
the
same
resource
for
above.
So
we
still
need
to
all
these
explanation
and
to
make
the
family
they're
aware
of
that.
These
two
have
different
behaviors
like
have
different
meanings.
G
B
M
D
E
D
E
D
I
E
C
E
Think
we
have
seen
previous
presidents
of
say,
create
a
field,
that's
required,
but
later
extended,
optional,
I
think
that
does
not
break
anything
from
from
from
a
memory.
So
we
can
consider
that
path.
I'll
just
make
a
deny
the
optional
former
beginning,
let's
say
I,
think
without
the
external
resource
provider
I
be
in
that
message,
making
agent
ID
optional
taxes
seem
reasonable
to
me,
because
then
people
will
be
confused
said
do
I
need
this.
They
were
likely
ignore
that
well
constructing
object,
but
then
got
an
arrow.
E
Okay,
so
we
can
make
it
optional.
With
a
comment
say,
this
is
required
as
long
as
you
are
working
on
the
present
volume,
all
local
resource
provider,
but
this
will,
but
once
we
have,
but
once
we
add
support
for
external
resource
provider
and
explore
its
a
provider,
ID
will
be
the
one
of
them
should
be
set.