►
From YouTube: 2019-10-15 Spec SIG
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
B
C
C
C
C
C
C
Good
morning,
everybody
its
aspects-
a
sick
meeting-
also
plenty
faces
here,
so
we
don't
have
many
items
on
agenda.
If
you
have
anything,
I
know
like
a
few
people
will
try
to
Norse
pool
requests,
so
this
is
still
ongoing
and
we
still
need
to
have
active
participation
in
pool
requests,
reviews
and
stuff
like
that.
C
C
We're
gonna
propose
the
last
time
that
we
want
to
be
aggressive
on
this
milestone
because
it
was
like
way
short.
So
maybe
we
can
postpone
most
of
this
issue
to
zero
point
three.
So
let's
maybe
discuss
what
issues
is
the
address
to
this
milestone
and
then
everything
else
I'll
just
go
and
we'll
declare
victory.
C
D
Don't
know
see
zero
point
two,
it's
supposed
to
be
the
end
of
this
week
in
the
past.
Okay,
well,
I
mean
that's
still
a
review
I,
don't
think
we
have
a
lot
of
outstanding
agreement
about
most
of
it.
But
there's
are
some
questions.
I
think
that
are
legitimate
about
the
so-called
observer
metric,
so
I
think
we
could
probably
get
most
of
that
document
merged
today.
If
we
took
out
a
driven
metrics
and
continue
discussing
those,
they
were
generally
agreed
upon,
but
have
some
details
that
are
clearly
asking
questions.
I.
C
D
There
are
probably
that
there
needs
to
be
one
more
document,
because
the
meter
interface
has
changed
a
little
bit
and
but
it's
pretty
much
the
easiest
part
I.
Think.
If
we
leave
out
observer
metric,
we
can
have
an
agreement,
but
we
can
have
a
doctor
a
false
set
of
documents
by
the
end
of
this
week
for
sure
yeah.
C
So
other
issues
here
I
see
some
issues
that
just
require
some
qualification
or
discussion.
I
think
there
can
be
easily
planted
like
where
the
spawn
started,
separation
instead
active
fun
or
doesn't
say
the
active
span,
so
I
think
that
one
specify
that
start
time
operation
does
not
detective.
Can
you
move
to
0-3
yeah
and
previous
one
multiple
run
times?
Multiple
tracers
can
also
be
a
MOOC,
yes,
yeah
I
think
everything
else
can
be
moved
to
the
next
milestone
and
among
entire
species.
Issues
with
no
milestone.
I
also,
don't
see
anything
critical
for
this
milestone.
C
This
is
alpha
version,
so
nothing
is
critical
in
just
our
decision
to
put
as
much
as
we
can.
Okay.
So
then
I
think
the
decision
is
to
move
everything
else
beyond
this
matrix
to
the
next
milestone
and
you'll
get
an
agreement
on
coolly
quests
that
are
currently
in
flight.
By
end
of
the
day,
we
will
also
take
them
yeah.
D
C
Okay,
I
had
one
topic,
but
I,
don't
think
I'm,
not
sure
who
will
be
interested
in
that.
So
one
of
the
issues
in
a
specification
repository
is
how
to
do
binary
formatter,
but
now
I'd
only
have
one
parameters
analyzed
to
one
field
and
then
WC
group
we
discussing
like
is
it
one
fielder?
It
should
be
two
kind
of
fields
how
to
represent
header
to
represent
like
propagation.
C
Like
I
see
Daniel
here,
who
also
part
of
phone
TT,
see
working
group,
maybe
Henry
I
can
also
participate
in
this
discussion.
We
really
need
to
close
on
how
standard
you'll
see
it
and
what
kind
of
puzzle
protocols
you
use.
My
name
for-
and
my
question
was
if
anybody
else
was
called
like-
has
a
very
big
stake
in
binary
protocol
for
like
for
other
technology,
like
wants
to
implement
I,
don't
know
like
some
custom
RPC
protocol,
so
something
like
that.
I
think.
A
The
main
interesting
thing
for
many
of
us
is
that,
eventually,
when
the
binary
protocol
is
finalized,
we'd
like
to
be
able
to
natively
receive
the
binary
article
right
on
our
end
right.
So
we
don't
have
to
write
exporters
for
every
single
language,
so
I
think
that
I'm,
not
at
least
for
honey.
Those
perspective
we're
not
fussy
about
what
the
binary.
F
C
In
general,
my
two
cents
on
this
every
time
when
you
have
a
binary
protocol,
you
don't
have
a
clear
key
value:
separation
or
sometimes
you
may
not
have
a
key
value
separation.
So
it
may
be
hard
if
you
have,
if
you
provide
a
lot
of
keys
to
to
keep
track
of
them
and
everything
it
depends
on
the
binary
protocol
that
is
implemented,
but
sometimes
some
of
them
may
not
be.
We
work
very
well
with
a
lot
of
keys.
B
F
C
I
think,
michael,
I
think
we're
talking
about
the
context
propagation
part,
not
the
data
serialization.
So
so,
when
it
comes
with,
how
do
you
send
data
to
your
back-end?
That's
up
to
you,
but
I
think
we
are
mostly
referring
to
to
be
in
contact.
I
I
mean
between
processes,
context
propagation,
which
is
happening
on
the
critical
fire,
not
not
as
an
exporter.
C
G
Think
maybe
you're
right
if
we're
not
maybe
not
speaking,
binary
protocols
there,
because
queue
implementations
over
also
very
specific.
Sometimes
they
have
fields.
For
me,
the
data
already
built-in,
so
it's
most
probably
more
of
a
sensor
or
like
an
issue
of
how
to
instrument
those
cue
protocols
or
queue
drivers.
C
C
Yes,
there
is
a
problem
sent
me.
There
is
a
problem
of
how
do
you
put
the
types
and
stuff,
but
there
is
also
the
concrete
problem
that
we
already
know
that
we
have
to
transfer
some
informations
between
producer
and
consumer,
correct,
yeah,
I'm,
saying
like
it's
a
different
problem,
you
just
it
says
it
is
that
nobody
looking
into
that
right
now,
because
it's
a
little
bit
more
advanced
scenario,
I
think
yeah.
C
So
my
my
in
general,
my
main
concern
with
having
multiple
fields
is
the
following,
so
there
may
be
binary
protocols
where
we
may
have
to
add
explicitly
into
their
definition
a
specific
key,
so
the
keys
may
be
statically
defined
somewhere
and
adding
more
keys
may
may
bother
people
and
may
ask
why
do
you
need
multiple
kids?
So
imagine
somebody
has
a
concrete
definition
of
these.
Are
the
fields
that
I'm
always
propagating
and
now
you
go
and
add
two
of
them
to
the
user
instead
of
one?
Does
it
make
sense
yep?
C
H
F
Yeah
I
guess
the
question:
there
is
kind
of
the
degree
to
which
it
makes
sense
to
just
have
one
binary
propagator
when
there
are
these
different
use
cases
and,
for
example,
like
gr,
PC
versus
custom,
R,
PC
versus
thrift
versus
n,
QP
p,
or
what
I'd
ever
remember
the
exact
acronym?
Are
they
sufficiently
different
from
one
another.
A
A
Okay,
oh
one
other
thing
for
binary
vertical
I'm,
realizing
that
there
may
be
a
need
like
I'm
thinking
of
projects
like
what's
it
called
a
slack,
has
a
thing
where
they
shove
events
through
a
Kafka
pipeline
right
like
that,
not
everything,
maybe
gr,
pc-based,
so
I
think
that
may
be
worth
thinking
about.
How
do
we
encode
individual
events
and
messages.
C
E
You
have
another
about
propagators
in
general
and
so
that
there
is
in
the
SDK
specification.
There
is
no
mention
at
all
about
these
propagators,
so
I'm,
not
sure
if
this
is
intentional
or
just
yeah
just
to
be
done,
because
I
think
we
should
document
it
in
the
same
way
as,
for
example,
we
do
for
span
processors
right
now.
C
A
A
C
I,
don't
know,
I
would
say:
I,
don't
have
a
concrete
answer,
but
adding
more
approvals,
I,
don't
think
he's
gonna
solve
a
problem.
We
try
to
go
with
the
30
or
40
percent
approval
rate
for
the
attacks,
so
adding
more,
not
necessary
will
solve
the
problem.
Unless
you
reduce
the
power
of
how
many,
what
percent
of
the
profits
to
the
right
I
mean,
we
just
gotta,
extend
the
problem
more,
but.
C
C
Maybe
like
specific
time
in
a
week
like
maybe
half
of
this
meeting
will
take
them
like
actually
like
going
through
pool
requests
and
like
one
by
one
and
I
probe
in
Xiamen
like
this
question,
was
a
working
issues
are
and
making
not
trying
to
be
more
decisive
like
this
in
or
out
and
like
this
I,
don't
like
in
a
face-to-face
meeting.
The
only
risk
of
that
is,
it
will
left
people
out
of
the
discussion,
because
the
discussion
will
be
in
very
specific
time.
A
C
If
you
have
an
idea,
what
is
actually
working
out,
we
lost
it
here
like
horse
and
you
lost
it.
You
just
let
us
know
we
need
to
discuss
it.
One
one
thing
that
we,
it
may
help
approving
gotab's.
Maybe
maybe
we
can
use
half
an
hour
of
this
session,
starting
with
next
week,
the
like
later
half
an
hour,
the
second
half
two
for
every
author
of
the
author
to
present
the
the
the
changes
and
some
reasoning
that
may
help
with
a
proper
process.
C
Yeah,
okay,
I
think
I
think
we
should
try
next
week
and
see
how
it
goes
reserve.
The
second
half
of
the
meeting
for
Day
perfect,
maybe
evening
before
that,
like
Monday
evening,
we'll
have
lockdown
on
chapter
3
discussion
and
don't
try
to
notify
people.
Yes,
that
is
fine.
I
mean
we
can
put
some
work.
It's
okay!
The
two
of
us
can
prepare
like
this,
isn't
gonna,
be
the
next
thought
that
we
discuss
and
then
everyone,
the
kinder.
C
E
Yeah,
could
we
go
back
to
the
topic
of
propagators
for
one
last
minute
or
so
I
think
my
my
last
question
was
I,
don't
know,
maybe
it
was
the
audio,
but
were
you
committed
on
it?
So
I
was
asking
about
the
absence
of
documentation
in
the
expect
of
propagators.
Whether
this
is
intentional
or
not
so
this.
C
C
C
C
C
H
C
A
Where,
for
some
reason,
a
decision
was
made
to
not
allow
ordinary
community
members
to
triage,
which
I
think
maybe
it
was
a
mistake
so
bright
now,
only
approvers
can
triage
issues
and
assign
tags.
So
you
can
put
it
in
the
title
of
your
quest
and
that
will
at
least
make
sure
that
you
can
edit
it.
But
in
the
longer
term
we
may
need
something
more
sustainable,
yeah.