►
From YouTube: 2020-11-18 meeting
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).
B
Oh,
no,
all
right
a
little
sleepy
tuesday,
tuesday
night.
We
have
a
10
p.m,
meeting
with
asia,
pacific
with
honorable
and
other
folks
who
are
in
that
region.
So
it's
a
little
bleary
on
wednesday
morning
after
that,
but
it's
an
excellent
meeting.
It's
really
good
to
get
a
chance
to
connect
with,
especially
with
honorag.
B
B
A
That's
great
yeah
so
that
I
don't
know
if
you're
waiting
for
other
people
to
join
before
we
talk
on
topics
yeah.
B
A
Cool
all
right,
I'm
just
going
through
a
couple
of
the
core
language
sigs
and
the
compliance
matrix
and
seeing
how
things
are
tracking
and
it
looks
like
for
java
there's
a
lot
of
it's
taken
care
of
according
towards
the
trace
table
and
the
resource
contacts
propagation
table.
Sorry.
B
B
I
actually
wanted
to
maybe
you're
a
good
person
to
chat
with
about
metrics
at
some
point,
so
the
I
mean
historically,
both
bogdan
and
josh,
put
together,
metrics
sdks
one
for
java
and
one
for
go
and
they
don't
look
very
much
alike
like
they're
implemented
quite
differently.
A
Heard
through
the
grapevine
that
tyler
is
in
the
process
of
redoing,
quite
a
bit
of
the
go
code.
B
A
I'm
just
trying
to
identify
like
you're
saying
java
is
based
on
go
and
if
joe's
go's
going
to
be
redoing
it
and
stuff
like
that,
if
there's
any
type
of
like
restructure,
something
like
that,
that
might
be
something
to
either
just
detach
from
go
and
go
like
we're
going
to
do
what
we
got
to
do
in
order
to
be
able
to
do
the
api
for
metrics
and
sdk
for
metrics
or
collaborate
with
them
and
like
what
is
it
that
you
guys
found
as
the
weakness.
B
Where
I
think
the
problem
is
going
to
come
up
okay
anyway,
I
just
wanted
to
put
it
on
your.
I
know
that
you
know
carlos
and
josh
have
been
talking
about
this-
also
hey
carlos,
but
I
thought
it
worth
putting
on
a
more
of
a
manager
make
sure
that
manager
manager
is
aware
that
we
we've
got
a
definite
disconnect
between
the
two
implementations.
B
B
I
think
most
people
would
probably
say
no,
but
that's
kind
of
where
we
are
like
this.
The
sdk
spec
is
being
put
together
in
such
a
way
that
it
doesn't
look
anything
like
what
java
has
so
actually
anyway,
I
tried
to
push
josh
towards
specifying
functional
requirements
on
the
sdk,
rather
than
very
specific
structural
requirements,
but
he
was
not
particularly
interested
in
going
down
that
route.
A
A
B
I
don't
know
where
that
stands,
hey
and
all
three
of
you
work
together.
So
maybe
you
could
sure
anyway.
So
this
is
it's
been
praying
on
my
mind
and
I
don't
I
want
to.
I
want
to
try
to
head
it
off
at
the
past
early,
so
we
don't
get
into
the
same
situation.
We
got
with
the
java
context,
implementation
where
we
kept
kicking
the
can
down
the
road
for
a
really
long
time,
and
then
suddenly
we
had
to
do
a
bunch
of
late
work
to
rewrite
our
to
do
a
new
context.
So.
C
B
Well,
but
I
think
this
actually
is
a
little
bit
political,
because
we
have
an
implementation
written
by
one
maintainer
that
doesn't
line
up
with
other
implementations
right,
yeah,
good
point
so
anyway,
I
just
wanted
to
bring
it
up,
at
least
with
you
andrew.
So
you
knew
that
this
was
something
that's
kind
of
lurking
in
the
background
for
java
and
I've,
I'm
feeling
uncomfortable
saying
that
we
have
a
metrics
implementation
when
it
doesn't
isn't
going
to
look
like
the
spec
at
all.
B
C
C
C
However,
the
expectation
is
that
this
will
be
done
till
by
the
end
till
the
end
of
january.
Being
realistic,
I
think
the
specification
ready
yes
metrics
in
the
metric
side
so
and
then,
of
course,
we
need
to
still
you
know,
allow
some
more
time
for
actual
implementations
to
be
compliant
so
yeah,
so
I
guess
yeah.
So
it's
gonna
take
a
little
while
yup
agreed.
B
Yeah,
no,
no
just
metrics
in
our
tracing
sdk
is
very
much
in
alignment.
I
mean
100
aligned
with
the
specification.
So
no
worry
there.
It's
just
metrics.
B
Anyway,
I
have
a
hard
stop
at
10
and
I
think
probably
a
lot
of
people
are
at
kubecon,
so
all
right
a
light
meeting
today
I
don't
have
to
man
the
splunk
booth
until
friday,
so
I
get
to
I
don't
have
to
kubecon
today,
but
I
have
other
things:
I'm
working
on
so
ga
burned
down.
We've
got
nine
p1p2
issues
open
one,
it's
it
says,
plus
zero,
but
we've
actually
had
a
couple
closed
and
a
couple
opened
there's
still
three
sitting
in
progress.
B
I
believe.
That's
still
true.
Let
me
take
a
look
yeah.
The
batch
observer,
we're
still
ken
finnegan
and
I
are
still
kind
of
trying
to
figure
out
what
that
api
should
look
like.
So
we
can
put
in
some
recommendation
into
the
api
specification
yeah
and
this
confusing
configuration
story.
This
is
what
I'm
kind
of
mostly
focused
on
and
actually
caused
me
to
open
up
a
p1,
this
two
builder
being
dangerous
yesterday
so
anyway.
B
So
that's
where
we
are
most
of
the
open
issues
and
to
do
are
not
are
kind
of
non-functional
non-functional
requirements,
there's
a
couple
that
are
bugs
that
need
to
be
worked
on,
but
it's
pretty
good.
I
haven't
put
p3s
on
this
in
this
project,
yet
I
should.
I
should
spend
some
time
taking
a
look
at
what's
in
the
p3
bucket
and
see
whether
the
things
we
really
need
to
do
for
ga
and
not
yeah
cool.
B
So
that's
where
we
are,
I
mean
I
think,
we're
continuing
to
make
good
progress
close
to
eight
issues
this
week
from
the
ga
burn
down.
I
think
that's
pretty
decent
and
honorag,
and
I
are
continuing
to
spend
a
significant
portion
of
our
time
doing
some
api
cleanup
and
now
working
on
sdk
configuration
cool.
A
B
D
That
you
submitted,
how
do
you
plan
on
or
what
what
do
you
have
a
proposal
for
how
to
fix
that.
B
Yeah
we
so
we
had
onorag
and
I
had
an
extensive
discussion
last
night.
I
put
the
comments
actually
on
the
issue
itself,
which
are
the
first
step.
We're
going
to
do
is
we're
going
to
remove
the
two
builder
and
require
building
instances
of
open
telemetry
from
scratch,
rather
than
doing
clones
of
existing
ones.
B
And
then
we
will
go
back
circle
back
and
try
to
figure
out
what
the
what
the
right
solution
is
going
to
be.
But
for
now
I
want
to
just
remove
the
thing.
That's
that
I
think
is
dangerous
are
as
dangerous
as
maybe
overstating
a
little
bit,
but
certainly
could
lead
to
unexpected
behavior
with
the
shared
state
that
would
be
shared
across
instances.
So.
D
B
Yeah,
we
absolutely
had
this.
This
was
the
extensive
discussion
we
had
last
night.
I
think
we're
not
clear
we're
not
sure
the
the
use
case
that
honorable
put
together
originally
for
the
for
the
two
builder
and
for
cop
doing
these
copies
was
to
enable
having
different
instrumentation
libraries
to
use
different
propagators.
B
That
was
that
was
the
original
use
case,
but
I
think-
and
so
what
we're
thinking
of
initially
is
maybe
address
that
use
case
specifically
not
and
not
have
a
like
a
a
very
large
scale,
two
builder
on
open
telemetry
itself.
B
His
use
case
is
that
you've
got
two
different
back
ends.
You
need
to
talk
to
and
one
of
them
talks,
jaeger
and
one
of
them,
or
one
talks
b3
and
one
of
them
talks,
w3c
and
the
you
could
just
configure
it.
So
you
inject
every
single
propagation
format,
but
that's
wasteful
and
it's
a
lot
of
extra
bytes
being
flung
around.
B
So
so
the
way,
so
I
think
what
he
wants
to
do
is
he
wants
to
be
able
to
have
the
user
be
able
to
to
actually
have
the
instrumentation
choose
a
propagator
or
choose
an
open
telemetry
instance
based
on
ip
address
or
some
sort
of
hostname
pattern
or
something
along
those
lines,
and
potentially
you
might
want
to
say
hey
this.
I'm
using
this.
This
instance
of
http
client
should
use
this
propagator.
This
other
instance
of
http
client
should
use
the
this
other
propagator.
C
I
do
remember
listening
to
this
argument
a
lot
of
times.
I
never
saw
that
in
practice.
So
that's
why
I'm
curious
about
whether
somebody
actually
implemented
this
and
it
seems
complicated
because
it
it
sounds
like
every
instrumentation
piece
has
to
support
this.
So
I
am
very
curious
about
how
practical
it
is.
I
remember
in
python
I
think,
which
isuke
who
mentioned
this,
but
it
was
never
implemented.
I
don't
yeah,
I
don't
know
I
I.
B
Have
a
doubt
yeah,
it
might
be
more,
it
might
be
more
theoretical
honorable
seem
to
have
a
use
case
for
it.
I
think
that
brave
supports
this,
which
is
one
of
the
reasons
why
I
think
he's
been
pushing
for
it,
because
if
we
want
to
support
brave
use
cases
in
a
in
a
clean
implementation
in
brave,
we
might
need
to
have
it
depending
on.
What's
going
on
there.
D
Okay,
what
if
the
property,
what
if
the
propagator
call
accepted
like
a
properties,
class
instance
or
something
or
some
map,
or
I
don't
know
making
it
so
that
you
have
a
little
bit
more
information
that
you
can
go
on
to
select
that
but
less
of
a
maybe
this
falls
into
your
proposal
of
like
let's
solve
that
separately.
B
Right
exactly
like
the
first
thing
I
want
to
do
is
I
want
to
really
say:
let's
get,
let's
get
something:
that's
safe
and
it's
not
going
to
be
surprising,
and
then
we
can
try
to
address
this
use
case
and
decide
whether
it's
something
that's
important
to
address,
because
the
other
way
is
you
could
just
in
like
the
instrumentation
could
just
have
propagators
directly
injected
into
it
right.
You
don't
have
to
get
those
off
the
global
or
off
of
an
open
telemetry
instance.
B
So
there
are
multiple
solutions
to
this
anyway,
and
so
for
this
p1
issue,
I'm
going
to
work
on
just
you
know,
making
it
making
it
unsurprising
and
safe
to
start
with,
and
then
we
can
continue
to
work
on
the
configuration
story
from
there.
B
All
right
are
we
going
to
release
this
week.
That's
the
next
actually.
B
I
think
we
should,
because
we
said
we
were
going
to-
I
am
very
busy
this
week.
I
might
have
time
to
try
something
this
afternoon.
I
can
try
to
do
that
myself
again.
I
already
did
it
once
cool
yeah.
If
you
could,
that
would
be.
That
would
be
great
and,
and
hopefully
all
of
the
wrinkles
are
worked
out
from
last
time.
C
Yeah
yeah
yeah.
Actually
we
need
to
you
know
I
you
know.
I
think
that
idea
is
to
support
spring
in
a
very
smooth
way,
so
we
need
to
persuade
marching
to
join,
to
join
these
meetings.
You
know
so
he's
in
the
loop
in
case
we
decide
to
break
some
like
we
decide
to
change
something
that
unexpectedly
breaks
his
code
and
he's
aware
of
that.
You
know,
or
he
can
give
his
input
like
as
soon
as
possible.
You
know.
B
So
yeah
he
he
asked
whether
we
had
a
github
group
that
people
could
be
notified
with
I've,
never
done
that
before.
Like
had
github
groups
for
notification,
I
have
been.
I
don't
know
if
people
have
been
noticing,
I
have
been
keeping
the
changelog
up
to
date,
putting
in
prs
every
other
day
to
keep
the
change
log
going.
So
it's
very
clear
what
hasn't
been
released
yet
so
there's
another
one
in
this
morning.
B
B
All
right,
andrew,
we
talked
a
little
bit
already
about
the
that
whole
span
attribute
ordering
that
has
been
removed
from
the
spec,
which
probably
just
needs
to
be
removed
from
the
compliance
matrix,
so
that
people
don't
think
it's
a
requirement.
B
B
C
B
B
B
Next
release
after
today,
what
is
we're
going
to
be
releasing
0.11
today?
So
I
think
if
we
could,
I
think
it's
a
goal,
certainly
to
have
a
0.12.
The
api
release
candidate,
I
wouldn't
say
sdk.
Quite
yet,
we've
got.
We've
still
got
configuration
issues
to
work
out
that
are
going
to
take
a
little
while
I
suggest
okay,
but
the
the
changes
have
definitely
slowed
down.
If
you
look
at
the
release,
notes,
there's
way
fewer
changes
than
there
were
no.10.
Oh,
that
10
was
a
almost
a
complete
rewrite
of
the
api
layer.
B
B
A
Cool
I'd
love
to
see
you
guys
like
leading
the
pack,
you
know
setting
the
was
it
pleasing
the
trails.
B
Yeah
it's
one
of
those
things
where,
once
you
start,
I
mean
we're
going
to
get
user
feedback
soon
too,
and
then
you
know
all
hell
could
break
loose
on
what
the
api
like
on
changes
to
the
api.
If
people
are
like
this
doesn't
make
any
sense,
we
need
to
completely
rewrite
this.
It
probably
won't
be
that
bad,
but
I'm
sure
they're
going
to
be
changes
requested
at
the
api
layer
once
we
actually
get
users
using
it
extensively.
A
Yeah,
but
that's
no,
that's
just
software
engineering
right,
like
gotta,
get
the
users
to
use
it
more
users.
Try
the
different
use
cases,
kick
the
wheels,
harden
it
and
then
develop.
You
know
yeah,
it's
just
the
way
towards
stability.
B
B
Cool
all
right,
yeah
and
then
you're
just
some
notes
on
sdk
requirements.
Yeah
and
I
don't
know
if,
if
the
light
steppers
could
powwow
and
think
or
get
together
and
chat
about
what
you
think,
the.
A
We'll
definitely
be
checking
yeah
lightsabers,
definitely
talking
together
with
it,
but,
of
course
I'll
bring
this
to
the
metrics
a
metric
sick
meeting.
So
that
way,
it's
a
public
discussion.
People
understand
what
the
understand
what's
going
on
absolutely.
B
Yeah,
I
don't
want
to
keep
this.
This
shouldn't,
be
anything
secret.
That's
all
I
wanted
to
bring
up
here
as
well,
and
there's
also
an
issue
by
the
way,
andrew
that
I
logged
in
java
the
java
repo
down
here,
2033
just
to
track
whether
we
do
need
to
do
this
or
do
significant
work.
B
All
right
well,
as
I
said
I
at
the
top,
I
have
a
hard
stop
a
meeting.
I
need
to
be
different,
a
splunk
meeting.
I
need
to
be
in
it
at
10.,
so
unless
there
are
further
issues,
I
will
see
y'all
later
and
hopefully
see
some
folks
around
kubecon.
B
Yeah
yeah,
I
assume
it
will
be
a
very
light
week
next
week,
I'm
I'm
still
working
monday
through
wednesday,
but
but
yeah.
I
don't
think.
There's
gonna
be
actually
there's
a
question
right
this
morning,
andrew
about
whether
we're
having
a
metrics
sig
meeting
this
week
because
of
kubecon,
because
if
we
don't
that
means
it's
we're
not
going
to
have
one
next
week,
because
it's
on
thursday.