►
From YouTube: 2021-02-17 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
B
Anthony
have
you
taken
a
look
at
any
of
the
milestone
stuff
recently
I
haven't
looked
recently
now:
yeah,
okay
yeah.
He
just
mentioned,
like
I
don't
know
if
you
can
see
the
price
of
the
chatter
about
just
like
rc
timelines,
given
the
announcement
just
went
out
an
hour
and
a
half
ago
for
the
1.0
of
the
spec.
Is
you
know?
The
immediate
questions
is
when
is
go
happening,
which
I
don't
think
there's
actually
like
a
severe
amount
of
pressure
associated
with
it?
B
It's
just
kind
of
trying
to
get
a
story
together,
for
it
is
a
it's
a
good
idea
regardless
and
we
have
like
the
main
project
board
and
that,
like
does
a
pretty
good
job
of
kind
of
giving
overview
of
the
whole
project,
but
I
think
that
we
could
probably
use
this
rc
milestone
to
actually
track
an
rc,
and
in
that
process
I
started
to
like
prune
it
a
little
bit.
B
I
took
out
all
the
metric
stuff
and
took
out
other
things
that
are
probably
not
related
to
anything
in
this
specification,
and
I
tried
to
go
through
and
like
add
anything
that
I
thought
might
need
to
be
there
but
yeah.
I
think
the
goal
of
the
meeting
here
was
to
kind
of
just
go
through
that
and
see
if
we
can
kind
of
remove
things
or
clarify
things
or
close
things,
and
then
I
think
the
goal
after
that
was
just
to
assign
things
if,
if
we
still
had
some
time.
C
Hello,
hello,
hey
hope.
I
didn't
speak
out
of
turn
when
I,
when
I
asked
about
the
coordination
stuff.
I
think
this
is.
This
is
very
much
in
you
guys's
court
and
I
don't
want
to
speak
for
you.
D
B
Sorry
I
mean
I
appreciate
it
regardless,
whether
it's
my
responsibility
or
not,
all
that
kind
of
help
is
really
possible,
so
yeah
great
to
hear
yeah,
and
I
was
kind
of
just
giving
anthony
a
little
bit
of
the
backstory
just
rehabbing,
especially
what
was
said
to
get
her.
My
understanding
was
like
kind
of
one
of
the
things
we
could
do
is
I
think
this
is
your
goal
as
well
correct.
B
If
I'm
wrong,
it
was
just
like
we
can
use
this
milestone
that
I'm
showing
here
this
rc1
milestone
to
really
capture
what
our
rc1
is
actually
going
to
entail.
I
already
went
through
pulled
out
anything
that
was
related
to
the
metrics.
B
I
pulled
out
anything
that
I
thought
could
be
kind
of
like
added
after
the
fact
or
as
an
addition
or
an
enhancement
or
something
like
that
and
I
went
through,
and
I
think
I
did
a
pretty
good
job
looking
at
all
of
the
open
issues
that
we
have
that
are
required
for
ga
and
adding
in
things
that
I
thought
were
related
to
the
specification
or
p1's
or
something
like
that,
but
yeah.
I
think
the
goal
today.
If
I
understood
you
correctly
was
to
kind
of
just
go.
B
Prune
it
and
then
if
we
can
try
to
make
some
assignments,
I
also
see
like
I
know
that
these
should
not
be
assigned
to
krasmir
anymore
he's
not
working
on
the
project.
As
far
as
I
know,
so
I'm
just
going
to
take
that
off,
but
yeah.
I
think
that
I
think
that'd
be
also
really
helpful.
Go
ahead,
yeah.
D
B
Yeah,
it
was
kind
of
a
dump,
is,
I
think,
how
he
said
it:
yeah
got
it
and
yeah.
So
the
back
story
there
is
when
I
was
working
at
new
relic.
We
contracted
out
with
kim
voke,
and
the
contract
ran
up
right
when
he
was
done
with
this
and
he
didn't
get
it.
So
you
just
kind
of
essentially
like
here's.
Here's
what
I
have
like.
I
can't
spend
time
curating
getting
this
in,
but
it's
there
for
people
who
want
to
like
take
a
look
at
it.
It's
cla
sign!
B
C
Before
so
before,
continuing
to
charge
the
issues
just
wanted
to
make
sure
we're
gonna
use
this
rc
one
as
the
as
kind
of
the
the
list
to
run
down
to
get
to
rc,
and
we
will.
We
will
get
keep
adding
new
issues
to
this.
To
this
milestone
right.
B
Yeah,
I
mean
that's,
that's
a
good
question.
Maybe
we
should
talk
about
that
because
I
I
would
before
I
could
say
that
I
really
need
to
go
through
the
specification.
I
think.
D
B
I
think
that
that's
a
really
important
thing
that
I've
done
multiple
times
now
over
the
past
three
months
been
trying
to
get
like
our
spec
compliance,
matrix,
updated,
which
seems
always
also
me
having
massive
conflicts
and
then
making
sure
that
we
actually
are
conforming
to
what
is
needed
for
that
1.0
specification.
I
don't
know
if
we've
captured
everything
in
tickets.
Here
is
my
only
reservation
into
saying
yes
to
what
you
just
asked.
C
So,
at
least,
can
we
make
the
following
statement
that
it's
either
because
it
is
documented
as
a
spec
non-conformance
issue,
or
it
is
a
bug,
nothing,
no,
no
new
features,
so
anything
that's
admitted
to
this
milestone
must
be
us
like
incompatibility
with
spec
or
an
internal
bug.
B
Yeah,
I
think
it's
a
pretty
good
demarcation
yeah.
I
think
I
can
say
yes
to
that,
I'm
thinking
of
a
class
of
issues
that
wouldn't
be
included
here-
and
that
are
things
like
this.
This
too
many
trace
packages
right.
C
And
I
think
that's
fine,
I'm
not
saying
to
remove
things
from
here.
I'm
saying
that
as
a
defense
against
further
like
further
addition
of
such
issues
right
like
fine,
we
we
added
these
things
to
the
milestone.
C
At
a
time
when
we
didn't
know
exactly
what
the
timeline
was
now
that
we're
trying
to
get
to
release,
we
can
say
hey.
This
is
like
our
soft
fields
criterion
and
if,
if
someone
has
like
a
really
strong
argument,
then
they
should
at
least
have
the
argument
like
it's.
It's
setting
the
default.
B
B
Cool,
do
we
want
to
throw
a
date
here
or
is
that
something
we
can
talk
about
later
on.
C
That
feels
aggressive
to
me.
I
would
love
to
put
a
date
here,
but
I
don't
know
what
to
what.
What
would.
B
Yeah
we're
doing
really
good
on
that
one:
okay,
yeah
we
can.
We
can
come
back
with
realistic
timeline
as
well
yeah.
That
sounds
good,
okay,
cool.
I
think
we
got
a
good
understanding
of
that
cool.
Did
we
want
to
go
down
to
the
list?
I
think
that
it
might
be
easier
to
go
from
bottom
up,
given
this
is
a
little
bit
more
recent
stuff,
so
people
might
have
more
context.
I
definitely
know
there's
some
up
here
like
this.
B
One
is
going
to
be
a
can
of
worms,
so,
let's
just
kind
of
I
guess
we
can
just
dig
into
it,
so
the
unifi
attributes
and
labels.
This
is
an
interesting
one
because
in
the
specification,
I
I
think,
there's
a
linked
issue.
Yeah
here
we
go
and
this
was
yeah.
This
was
close.
It's
superseded
by
something,
but
the
idea
is
we
wanted
to
unify
in
the
term
attributes
in
the
project,
and
we
currently
have
labels
in
the
metric
side
and
we
have
attributes
in
the
tracing
side
of
things.
B
That's
not
a
really
big
issue
from
the
tracing
side
of
things.
If
we
want
to
ga
the
tracing,
because
that
is
not
necessarily
going
to
change
the
api,
the
labels.
D
B
Is
going
to
need
to
like
it
could
potentially
need
to
be
renamed,
so
I
included
it
in
here,
but
we
could
also
just
make
a
comment
saying
like
leave
the
labels
packaged
and
then
in
the
metric
api
change.
You
know
the
change,
the
you
know,
the
the
term
or
the
field
name
there
or
something
like
that.
E
One
thing
that
we
did
sorry
for
being
directly
into
the
discussion,
but
one
thing
that
we
did
in
java
was
to
move
these
as
a
part
of
the
metrics
artifact
or
module,
instead
of
being
in
the
common
module
that
was
stable,
just
to
make
sure
that
we
don't
do
breaking
changes
in
that
stable
package.
So
maybe
a
simple
solution
here
is
just
move
the
the
labels
as
a
standalone
package
or
or
something
like
that
module.
C
If
that's
an
acceptable
resolution
for
the
group
I
would
be
happy
to,
I
would
be
happy
to
have
this
assigned
to
me.
A
C
B
Yeah,
it's
just
the
naming
of
of
the
module.
You
know
the
idea
for
that
issue
is
to
unify
on
the
term
of
attribute
or
label
it's
for
cognitive
overhead,
for
new
users
when
they
come
to
like
a
metric
event
and
they're
like
have
to
put
a
label
here
or
they
have
to
put
an
attribute
on
a
span
or
something
like
that,
and
there
was
just
an
ask
to
try
to
unify
on
those
same
terms
because
they're
key
value,
pairs
and
logan's
on
the
call.
B
He
didn't
cringe
that
time
that
I
said
that
so
like
there's,
like
controversy
around
it
as
well,
about
like
how
we
want
to
like
do
translations
for
different
types,
because
there
are
different
types,
but
in
go
like
we
support
like
this
label
concept
and
that's
kind
of
like
implementing
the
whole
thing.
So.
D
B
It
seemed
like
we
wanted
to
just
unify
on
the
term.
Steve
also
was
asking
this:
he
was
a
user
of
this
as
well,
so
the
idea
is
essentially
like
unify
on
a
package
name
or
don't
change
the
package
name
and
just
unifying
the
fact
that
we're
not
going
to
change
it
and
that's
just
what
it's
going
to
be
and
then
the
field
will
be
like
you
know.
In
the
metric
side
of
things,
we
could
change
things
there.
E
One
one
thing
that
we
don't
do
in
java,
for
example,
is
we
don't
use
this
for
for
baggage
or
for
trace
tape,
so
for
baggage
or
for
tradestate?
We
have
an
api
that
simply
accepts
string
string
because
whatever
good
or
bad-
but
we
don't
reuse
this
package,
so
we
were
able
to
to
completely
get
rid
of
the
label
concept
from
the
the.
What
is
what
is
so
far
marked
as
a
stable.
I
don't
know
if
that's
possible
or
not
in
gold,
but
it
may
be
good
for
goal
just
in
case
label.
E
E
B
Yeah-
and
I
think
that's
the
difference
here-
is
that
these
attributes
are
implementations
of
key
values
from
the
labels
package.
Yeah
here
you
go,
this
is
the
and
so
that's
I
think
the
difference.
D
E
But
is
the
value
here
supporting
all
the
things
that
attributes
support
in
the
labels
that
value
like?
Yes,
then,
then,
maybe
a
simple
rename
of
two
attributes
is
is
reasonable.
A
B
I
kind
of
agree
on
that.
One
yeah:
why.
E
Go
ahead,
one
small
thing:
I
saw
this
used
in
the
tray
state
yeah,
if
I'm
not
mistaken,
where,
where
the
only
possible
values
where
we
know
for
sure
are
only
strings
that
may
require
you
to
to
change
the
trace
state,
to
not
use
these
labels
and
to
just
simply
accept
string
string
things.
B
Yeah
right,
okay
and.
B
A
B
C
So
I
have
an
aggressive
question
to
ask
for
this,
for
this
meeting
sure
is
it
okay
for
us
to
say
that
we
should
emerge
with
every
issue
assigned,
even
if,
like
basically,
the
assignee
should
like
be
on
the
hook
for
maybe
you
have
to
go
pay
someone
like
pay
your
pay,
your
neighbor
money
to
learn,
go
and
fix
it,
but
someone
has
the
buck
kind
of
thing,
or
is
that
too
much.
B
I
would
let's
let's
let's
target
that?
Yes,
I
don't
know
if
we'll
achieve
that,
because
I
don't
know
if
I'll
get
through
all
the
issues
today,
but
I
was
kind
of
something
anthony
like
if
that's
the
case,
let's
schedule
something
for
later
on
in
the
week
as
well
to
just
follow
up
on
this
this
one
I
can.
I
can
take
this
one
then
and,
like
you
said,
like
I'll,
find
an
owner
or
I'll
get
it
done.
A
B
B
Actually,
expert
student
takes
the
context,
so
we
could
just
there
you
go.
I
think
that's
yeah.
E
B
Well,
I
mean
it,
it's
the
exporter
interface,
so
you
know
splunk
or
new,
relic
or
google,
whoever
built
their
export
or
have
to
support
it.
I
guess,
but
if
not.
C
B
C
B
Labels
are,
let's
just
say,
not
curated,
well
yeah,
so,
whatever
whatever
we
need
to
do.
Okay,
I
think
we
got
that
there's
some
bogdan.
B
Glad
that
got
him
here,
oh
okay,
yeah!
This
is
a
good
question.
I
still
wasn't
quite
clear
on
this
one,
so
this
is
a
question
about
removing
the
trace
flags
that
we
already
have
supported
for
the
is
deferred
in
the
debug
state
and
bogdan.
Your
point
is:
is
that,
like
in
the
specification,
we
don't
have
the
specified.
E
Have
in
the
specification
that
we
confirmed
to
wtc
standard
for
that
and
okay,
that
is
not
defined
by
even
by
our
specification
or
by
the
standard,
and
I
think
the
problem
that
you
have
right
now
is,
if
you
support
them
and
you
allocate
a
bid
for
them
and
especially
if
you
put
them
on
the
wire
that
will
break
things
when
when
somebody
will
allocate
that
for
different
reasons.
So
so.
B
A
Yeah,
it
explicitly
looks
for
the
sample
flag
there,
the
there's
the
bug
would
be
for,
like
b3,
that
has
a
debug
flag
in
its
propagation
format.
Right.
B
Yeah
yeah
and
like
I
still
I
haven't,
I
have
to
look
at
this
again
but
like
when
I
originally
wrote
it
just
from
the
fact
like
the
span
context
is
not
actually
exposed
to
any
sort
of
of
the
propagators.
It's
like
it's
hard
to.
I
don't
know
of
a
way
to
then
like
transport,
that
information,
because
there's
like
it
can
come
back
as
like
a
context
or
it
can
come
back
as
like
some
sort
of
trace
date,
but,
like
I
don't
like,
there's
no
way
to
actually
like
propagate
this.
E
So
what
we
did
in
java,
because
the
propagator
work
with
the
context
correct,
so
they
accept
the
context
and
they
produce
a
context.
So
injection
the
inject
thing
gets
a
context
and
the
extracting
produces
the
context.
E
So
what
we
did
was
to
allocate
for
for
b3
different
entries
in
the
context
for
these
specific
flags
and
and
then
then,
whenever
you
extract
whenever
you
receive
b3,
we
put
them
in
the
context
and
if,
if
we
are
exporting
v3,
there
will
be
in
the
context,
because
we
propagate
the
context
through
all
the
apps,
so
between
x,
between
b3
propagator
can
read
the
values
from
the
context
and
and
put
them
on
the
wire.
If
that
is
needed,.
A
Yeah-
and
I
think
this
can
go
hand
in
hand
with
the
the
other
issue
that
bogdan
had
opened
about
making
trace
flags
destruct,
because
one
thing
I
do
see
in
the
trace
contact
exporter
is
that
it
just
does
a
bitwise
ant
of
trace
flags
and
the
sampled
flag.
So
if
w3c
adds
more
flags
and
we're
already
using
those
bits,
we
may
run
into
trouble.
B
Okay,
anthony,
I
know
you're
kind
of
in
flux
with
time-wise.
Is
this
something
that
you
could
take
a
look
into
these
two
things?
Could
I
sign
your
choices?
Okay,.
A
B
Right
and
I
never
remember
how
to
spell
your
name-
oh,
it
was
a
captain.
A
B
That's
how
you
say,
aneurysm
right
and
I
feel
like
I
have
an
aneurysm
whenever
I'm
like
trying
to
spot.
Like
god,
I
can't
figure
any
of
this
out.
Okay,
cool
awesome,
making
progress.
B
A
I
I
don't
know
I
I
think
I'd
say
a
reasonable
thing
to
have
on
the
tex
mac
carrier.
I
don't
know
that
it's
going
to
be
impossible
for
any
other
carriers
to
implement,
like
we've
got
custom
carriers
in
the
serama
instrumentation
in
contrib,
but
they
should
be
able
to
enumerate
their
keys
as
well.
A
So
it's
probably
fine.
I
was
just
a
bit
put
off
by
the
the
you
know.
You
must
do
this
when
it's
very
clearly
not
a
must.
B
Okay,
all
right
I'll
be
honest.
I
saw
this.
I
haven't
looked
too
deep
into
it.
Would
you
recommend
that
we
do
this
after
ga
and
remove
this
from
this
milestone,
or
is
this
something
that
we
should
try
to
get
done
for
the
we
need
to
get
done
for
the
release.
A
I
think,
if
we're
going
to
do
it,
it
would
probably
be
good
to
have
it
in
ga
and
it'll
be
a
simple
function
to
add
right.
It's
basically
walk
over
the
the
carrier
and
enumerate
all
of
the
keys
and
return
them
in
a
list.
A
This
was
needed
for
trying
to
remember
what
he
needed
it
for
again.
It
was
something
he
was
working
on
in
contrib,
oh
okay.
It.
E
Is
needed
for
for
the
jager
propagator
and
the
ot
propagator,
because
they
send
every
baggage
as
an
independent,
key
prefix
with
ot
or
younger
dash
baggage
than
the
name
of
the
the
thing.
So
so
because
they
don't
know
the
the
keys
in
the
baggage
they
just
prefix
every
key
with
a
specific
graphics.
So
they
need
to
iterate
over
all
the
keys.
A
Okay,
yeah
yeah,
so
this
seems
like
a
pretty
reasonable
thing
to
add,
and
it's
probably
better
to
do
it
now,
while
we're
already
going
to
be
asking
people
to
make
changes
again,
because
I
think
this
will
be
backwards
incompatible.
A
Well,
it's
is
it
forwards,
incompatible
right,
it's
adding
a
method
to
an
interface
that
things
will
have
to
implement
yeah,
so
I
think
we
need
it
now.
I
think.
C
So
yeah
we
we,
we
have
to
do
it
now.
If
we
want
to
do
it
right.
B
Yeah,
if
you're
up
for
it,
you
can
do
that.
I
can
also
take
this
one
as
well,
but
yeah
if
you're
excited
about
it
and
it's.
C
Well,
I'm
sort
of
round
robin
since
I
it
was
like
I
took
a
couple,
then
you
took
something
you
know
anthony
took
some,
so
I.
B
A
This
is
one
where
we're
dealing
with
what
we
were
talking
about
in
the
sig
meeting
the
other
day
right
with
exception,
error,
yeah
yeah.
So
there
was
an
open
pr
for
this
that,
I
think,
might
just
need
to
be
updated.
B
Yeah
oh
looks
like
that's
right.
Where
was
that
pr
here
we
go
yeah.
It
looks
like
I
need
to
respond
to
this,
but
mother,
I'm
just
gonna
assign
this
to
mothership.
B
First,
package:
okay,
so
this
is
one
that
punia
had
opened
up.
C
Yeah,
I
I
actually
didn't
think
this
was
this
needed
to
be
required
for
required
for
1.0
like
we
get
to
1.0,
and
we
can
do
some
manual
stuff,
the
next
release.
After
1.0,
we
need
some
enforcement,
so
in
my
opinion
this
can
be
relaxed
a
little
bit
this
and
the
other
tooling
one.
They
can
both
be
a
little
bit
more
relaxed
compared
to
just
getting
one
point
out.
B
Okay,
this
one
right
here.
C
Yeah
anthony,
what
do
you
think
I
know
this
was
this
was
your
brainchild.
I
just
want
to
check
in
that.
That's,
okay
with
you.
A
Yeah
I
agree.
These
are
both,
I
think,
about
ensuring
that
once
we've
got
1.0,
we're
maintaining
compatibility,
and
so
they
really
aren't
important
for
1.0
they're
important
for
1-1
or
101..
A
E
A
Have
so
that
was
part
of
this
issue
that
that's
up
right
here
was
improving
our
tooling,
but
I
think
putin
has
done
some
work,
or
at
least
some
thinking
on
what
we
do
in
in
the
interim
to
break
the
the
bits
of
our
tooling
process
that
forces
everything
to
be
in
lockstep,
so
yeah.
I
I
think
we've
got
a
plan
there,
if
not
an
implementation.
C
I'm
gonna
I'm
gonna
socialize
the
plan
also,
but
at
least
I
think
I
have
all
the
pieces
in
my
head
at
least,
and
I've
discussed
with
you
folks
the
questions
that
that
felt
most
contentious.
B
B
Cool
I
just
want
to
like
we
have
30
seconds
left
until
the
top
of
the
hour.
I
can
stay
on
here.
I
actually.
Let
me
double
check.
I
can
stay
on
here,
yeah,
I
can
stay
on
here.
I
don't
know
if
anybody
else
had
other
things
they
need
to
be
doing,
but
I'll.
B
Okay,
that
sounds
good.
I'm
guessing
bogdan,
never
has
anything
to
do
it's
pretty
pretty
open
up
from
what
I've
heard
so
we'll
see,
cool
updates,
fan
status
and
description
setting
to
comply
with
the
specification.
I
have
not
looked
at
this
in
a
long
time,
but
it
looks
like
some
updates
to
the
yeager
status
code
to
comply
with
the
specifications,
so
this
seems
like
a
reasonable
thing
to
keep
in
here.
B
I
think
that
since
I'm,
the
one
who's
also
on
this
issue,
I'll
just
assign
it
to
me
and
take
it
as
an
action
to
find
the
owner
or
to
oh
actually
looks
like
maddie
jay
was
looking
for
some
help.
Okay,
actually
I'm
gonna
sign
to
him
and
then
see
if
I
can
follow
up
and
help
for
direction
on
that
one.
I
don't
know
how
much
activity
he's
actually
got
with
the
project
anymore
might
have
needed
to
drop
off,
but
I'll.
B
B
You
and
me
both
so
yeah.
Let's
see.
E
My
two
cents
here
I
would
do
whatever
grpc
does
to
be
to
be
consistent
with
that,
because
this
is
for
the
grpc
exporter
if
they,
because
they
have
a
with
insecure.
I
don't
I
don't
know
if
they
use
or
not
a
boolean,
but
I
will
do
whatever
they
do.
Just
for
people,
memory,
muscle
memory
to
just
work
like
that,
doesn't
matter
what
it
is.
C
Same
answer
I
mean
I,
I
agree
with
it
for
consistency
and
I
agree
with
grpc's
reasoning
for
doing
it.
The
way
they
do
it
so.
C
Yes,
the
way
I
see
it
is
like
jrpc
wants
to
make
it
inconvenient
for
you
to
pick
insecure.
They
want
to
make
it
conspicuous
and
inconvenient,
and
ugly
they're
like
the
default
is
secure.
If
you
want
to
be
insecure,
then
you
can't
just
say:
well,
it
wasn't
my
fault.
The
customer
just
said
insecure.
I
just
passed
in
the
boolean,
it's
like!
No,
no
you
you!
You
saw
that
boolean,
you
thought
about
it.
You
said
it
was
good.
I
want
to
be
insecure,
cool.
A
A
B
Yeah,
I
really
like
this
idea
of
just
like
the
width
and
secure
because
then
it's
a
secure
by
default
sets
people
up
the
correct
way,
but
I'm
hearing
you
anthony
is
this
something
either
puny
or
anthony.
Would
you
like
to
take
trying
to
respond
to
this,
or
I
can.
C
A
Yeah,
I'm
100
on
board
with
that,
and
we
can't
add
the
other
option
later
if
we
decide
that
that's
an
important
thing
to
add.
So
I
would
be
perfectly
fine
with
saying
we're
not
going
to
do
anything
with
this
now
and
we'll
cross
the
other
bridge
later.
If
we
need
to.
B
Yeah,
that
sounds:
let's
do
that.
I
posted
the
link
in
chat
for
this
one.
So
if
one
of
you
would
like
to
take
responding
with
what
we
just
said,
but
I'm
gonna,
I
think
that
you're
right.
I
think
this
could
be
done
after
ga
as
well.
So
that's
something
up.
B
B
Yeah,
I
think
this
is
the
one
I
would
let
somebody
down
the
wrong
path.
Let's,
let's
I'll,
take
care
of
this
one.
Let's
see
this
is
just
a
export
pipeline
for
the
jaeger
exporter.
B
It
definitely
needs
to
get
resolved
for
the
ga,
so
I'm
gonna
leave
it
in
here,
because
otherwise
it's
gonna
be
a
function
change.
So
I'm
just
gonna
move
past
this
one.
Unless
you
guys
want
to
talk
more
about
it.
B
Okay,
on
my
plate,
I
can
do
that.
Sdk
span
update
should
be
ignored
after
I
just
calls.
Yes,
it
definitely
should
be
okay.
I
wrote
this
yeah.
That
makes
sense.
A
F
B
To
that
as
well,
this
is
an
interesting
one,
so
the
remove
extra
little
value
types
I
feel
like
there
was
an
attempt
at
a
pr
on
this
one,
but
the
idea
is:
is
that
the
label
value
types
we
currently
support
are
the
ones
listed
here,
but
an
attribute
was
a
string,
boolean
double
precision
or
a
sine
64-bit
integer.
So
removing
these
types
was.
I
think,
the
idea
of
this
pr,
oh
and
there's
an
open
pr.
B
And
it's
got
three
approvals:
why
isn't
this
merged?
Someone
didn't
like
this.
A
Yeah,
I
think
there
was
some
concern
with
how
the
you
end
to
end
conversion
was
being
handled,
but
I
think
most
of
those
got
resolved.
D
B
It's
on
the
patch,
though
so
it's
this
is
like
a
hard
one.
I
can
take
a
look
yeah.
E
Yeah,
it
definitely
needs
to
be
on
the
head.
It's
very
important
for
if
you
rebase
on
the
head
that
the
number
will
will
clearly
be
better,
showing
the
reality
versus.
If
you
have
your
branch
on
a
later
like
older
undercommit.
B
Right,
okay,
I'm
gonna!
Actually
I'm
gonna
go
back,
assign
this
pr
to
myself.
So
I
don't
forget
as
well
that
I
need
to
review
that.
That's
my
my
problem,
okay
cool
this.
Maybe
we
have
bogged
in
here.
So
maybe
you
can
have
some
insight
into
this.
B
So
this
was
it
asked
by
morgan
like
in
october,
because
originally
carlos
had
moved
third-party
propagators
like
the
b3
propagator
to
the
contrib
libraries,
and
then
I
think
alolita
was
not
a
really
big
fan
of
that
and
wanted
them
to
move
back,
which
was
kind
of
a
flip-flop,
and
then
morgan
asked
me
to
do
this,
but
he
said
that
he
would
resolve
this
issue
within
like
the
day
or
something
like
that
and
it
never
got
resolved.
So
I
don't
know
what
to
do
here.
There
is.
E
A
there
is
in
the
in
the
specs.
There
is
a
section
explaining
where
these
and
what
maintainers
of
the
the
repo
should
support.
So,
if
you
go
to
the
context,
propagation
and
search
for
b3
or
something
here,
yeah
here
yeah
this
section,
property
distribution,
propagator's
distribution,.
B
Okay,
so
is
there
any
sort
of
additional
properties
that
may
be
maintained
and
distributed?
Oh
extension
packages,
so
all
of
these
should
be
in
the
main,
repo,
essentially
you're,
muted,
but
no.
E
Sorry
so
the
the
the
first
two
says
it
may
be
in
the
api
that
the
other
two
it's
it's
in
the
core
repo
as
different
package,
not
as
part
of
the
api
module
but
as
a
different
package,
and
you
may
have
extra
things.
But
what
I'm
we
are
trying
to
say
is
when
you
depend
only
on
the
api,
you
may
get
the
w3c
ones,
but
nothing
else.
B
B
Where
that
should
be
stored,
but
given
the
fact
that
the
1.0
specification
just
came
out,
I'm
not
really
anticipating
this
changing,
no
okay,
all
right!
I'm
just
gonna
close
this.
A
They're
not
part
of
the
api,
it
says
it
may
be
right.
Okay,.
A
B
I
think
we
should
probably
do
this.
This
is
just
like
changing
to
a
this.
Is
a
review
from
janna
about
the
span
recorder,
actually
being
a
concrete
type.
D
B
Recorder,
what's
fan
recorder,
so
the
spanner
recorder
here
is
essentially
it's
an
extension
to
that.
It's
an
interface
for
the
test
package.
So
essentially,
if
you
want
to
provide
your
own
form
of
how
a
span
recording
is
done
so
every
time
that,
like
you,
use
the
testing
package
and
you
record
a
span
on
start
from
the
from
any
sort
of
registered
spam
recorder
or
on
end
is
going
to
be
used,
and
then
we
provide
a
default
implementation,
which
is
the
standard
span
recorder
which
just
keeps
track
of
them.
B
I
think
in
like
a
list,
so
you
can
get
all
of
the
spans
that
have
completed
all
of
the
spans
that
have
started,
and
it's
just
a
pretty
standard
way
to
like.
You
know,
for
testing
frameworks
and
that
kind
of
thing-
and
I
think
yana's
point
is
just
like
I
don't
know
if
we
need
to
make
this
extensible.
Maybe
this
doesn't
need
to
be
an
interface
like.
Maybe
we
you
know,
users
don't
actually
need
to
be
providing
their
own
spanner
quarters.
In
fact
I've.
Never!
Actually,
oh
that's
not
the
case.
D
B
Span
recorder,
one
time
when
I
was
working
at
urelic,
because
what
did
I
want
to
do?
I
wanted
to
do
key
value
lookups
on
this.
Instead
of
just
getting
a
list
back.
B
So
that
nobody
actually
can
implement
this
outside
of
the
project,
yeah
yeah.
That
sounds
like
a
really
good
idea.
The
interface
from
the
moment,
it's
equivalent
correct.
C
No,
no,
if
it's
an
it's
an
interface
right
now.
The
problem
is
that
tomorrow,
someone
can
say
well
that
used
to
be
an
interface.
I
implemented
it.
Today,
you
made
it
into
a
class.
You
broke
it
into
a
struct.
You
broke
me
you're,
a
horrible
person
yeah
like
no.
I
was
a
horrible
person.
Yesterday
I
already
improved.
A
F
E
B
C
C
I'm
trying
to
take
all
the
issues
that
require
no
conceptual
understanding
of
this
thing,
so
that
you
folks
are
freed
up
to
answer
the
actually
hard
questions.
B
Yeah
right
yeah
now
we're
getting
into
the
weeds
here.
Please
spend
a
relationship
with
potentially
okay.
E
Can
you
can
you
look
at
the
current
state?
What's
the
these
are
the
options
that
you
have
with
new
route?
What
what
are
the
current
options
that
you
have.
B
Where
would
this
be,
I
thought
it'd
be
here.
Sorry.
B
Oh
okay:
here
we
go
yeah
so
with
new
roots
with
record
with
spankines,
I
think.
Are
the
owen.
B
E
Oh
so
so
the
context
is
is
not
in
the
option
is
an
explicit
argument.
As
the
first
argument
correct,
correct.
B
E
E
E
C
It
seems
it
seems
late
in
the
game
to
do
that
right
now,.
B
Yeah,
I
would
have
to
agree,
but
I'm
otherwise.
The
thing
is,
though,
there
was
like
this
guy
bogged
and
he
messaged
me
and
he
was
like
is
the
api
gonna
change
and
I
was
like
I
don't
think
so,
like
that
seems
pretty
stable,
so
I'd
have
to
like.
Have
somebody
deal
with
him?
I'm
not
sure
what
we
would
change
it
too.
That's
yeah,
that's!
I
think
the
other
question.
E
What
I
would
I
would
look
into,
not
necessarily
change,
but
what
I
would
look
into
is
to
remove
the
context
as
an
explicit
argument
and
make
it
as
part
of
the
option,
because
the
context
here,
as
I
explained,
is
not
used
as
as
as
a
normal
context.
C
An
alternative
to
that,
if
we
are
really
considering
alternatives,
could
we
say:
hey
you,
you
extract
like
there's
one,
the
tracer
lets.
You
extract
this
pan
from
the
context
and
that
just
takes
the
context
of
nothing
else
and
then,
whatever
it
gets
out
of
the
context,
the
identifier
of
the
parent
that's
passed
as
an
option,
because
what
seems
odd
to
me
is
the
ability
I
mean
also.
E
Yeah
may
contain
propagation
elements
may
contain
a
bunch
of
things-
maybe
maybe
maybe
another
option
here
to
be
considered
without
calling
that
with
new
route,
maybe
called
with
no
parent
or
with
ignore
parent
or
whatever,
a
better
name
for
that
option.
That
would
suggest
that
this
context
should
not
be
treated
as
a
parent,
but
the
problem
is,
if
you
do
not
treat
that
as
a
parent,
you
have
to
return
a
context
as
a
return
function,
correct
yeah
and
do
you
use
the
the
given?
E
E
A
Yeah,
correct
yeah,
the
contexts
are
immutable,
so
we
don't
really
have
the
the
concept
of
a
span,
that's
created,
but
not
active
in
a
context.
If
you
don't
want
that
space
to
be
active,
you
just
keep
using
the
old
contacts
you
had
before.
E
E
No,
no,
no,
the
spy
start
creates
a
span
is
not
descriptive
enough
about
what
the
heck
is
returning
me
and
what
what
are?
How
are
the
arguments
interpreted
in
this
case,
but
yeah
at
least
that
I
would
say
to
start
with?
E
B
Just
yeah,
this
is
also
another
tough
one,
because
you
know
this
is
an
api
documentation.
B
B
A
C
Yeah,
I
think
we
should
have
a
low
earth
we
should
have.
We
should
be
more
willing
to
accept
docs
changes
because
they
don't
have
knock-on
effects
for
anything
else.
B
Yeah,
so
this
is
the
other
thing
is
so
like
when
you
actually
do
get
the
with
your
root
option.
I
think
this
is
saying
that,
like
we'll
ignore
the
local
and
remote
parents
in
the
context
and
create
a
span
of
the
trace
id,
the
ignore
parents
will
end
up
as
links
with
the
string
attribute,
ignore
on
demand
local
ignore
on
demand
for
local.
C
Do
we
actually
want
to
have
those
links?
I
guess
that
seems
yeah.
The
links
seem
questionable
to
me.
But
again
I
don't
know
enough.
A
I
think
it's
useful
to
have
the
links,
but
it's
probably
also
something
that
maybe
we
don't
want
to
automatically
create,
but
to
have
convenience
methods
to
enable
the
creation
of.
C
A
So
I
think,
I
think,
potentially
one
of
the
the
common
uses
for
this
is
going
to
be
where
someone's
offering
a
service
and
their
consumers
may
have
instrumentation.
That
gets.
D
A
Along
so
it
would
absolutely
be
useful
if
you
could
have
that
link
so
that
if
a
customer
comes
your
customer
support
and
says
hey
something
went
wrong.
I've
got
this
trace
id.
Do
you
have
anything
that
that
correlates
with
it,
but
they'll
be
able
to
do
that?
But
you
don't
want.
You
know
to
pretend
that
there
was
actually
some
missing
parent
because
you're
never
going
to
get
the
spam
data
for
that
got.
It.
B
B
If
we
are
including
links
with
these
attributes
and
then
the
specification
comes
along
and
says
like
this
is
really
useful,
except
instead
of
calling
it
ignored
on
demand,
we're
going
to
call
it,
you
know,
dropped
parent
or
something
like
that
like
that,
then
we'd
have
to
introduce
a
breaking
change
to
be
compliant
at
that
point,
so
I
feel
like
we
should
probably
just
remove
these
for
the
time
being,.
A
Okay,
so
the
safest
thing
to
do
is
probably
to
remove
it,
have
some
documentation
on
how
users
can
create
their
own
if
they
want
to
until
the
spec
comes
along,
and
then
we
do
whatever
the
spec
says.
That
makes
sense.
B
B
Okay,
cool
is
this.
I
think
we're
back
up
to
me
on
the
assignees
right.
B
B
You-
and
this
is
essentially
saying
that,
like
it,
should
only
do
this
if
it
go
ahead.
F
C
C
One
got
in,
but
maybe
they
changed,
maybe
they
changed
the
pr
after
you
approved
it.
That's
what
happened.
Sorry.
A
A
We
should
probably
discuss
at
a
sig
meeting
one
of
these
days
because
I
recently
discovered
github
does
have
the
ability
to
revoke
approvals.
I
want
to
change
and
I
don't
know
if
we
want
to
enable
that
or
not
I
generally
like
to
have
that
enabled,
but
I
can
see
how
it
also
adds
turn.
E
A
Though
we
we
do
already
have
an
implementation
in
set
name
that
runs
a
sampling
decision.
B
C
E
A
I
I
think,
that's
definitely
the
simpler
thing
and
again
keeping
with
the
theme
of
it's
easier
to
remove
functionality
and
add
it
later
that
I
would
definitely
go
with
removing
what
we
have
now.
That
may
be
somewhat
half
baked,
and
we
can
always
add
it
later.
If
we
decide
it's
valuable
yeah
when
in
doubt
leave
it.
B
We
should
probably
I'm
gonna
awesome
pogba
thanks
for
joining
in
yeah.
That
sounds
good.
I'm
gonna
jump
off
actually
as
well,
but
let's
see
where
we're
at.
A
B
It's
a
tough
one,
cool
I'll
finish.
This
comment,
punia
and
anthony.
What
do
you
guys
think
about
meeting
up?
Let
me
check
my
calendar.
Oh
gosh,
tomorrow
looks
really
bad
thursday.
Do
you
guys
have
any
availability,
then.
C
Yeah,
for
me,
so
actually
are
both
of
you
on
the
west
coast.
C
Thursday,
I
mean
I
so
I
can
do
like
actually
late
evening
on
thursday,
like
if
anyone
wants
to
5
p.m-
west
coast
8
pm
east
coast,
if
anyone,
if
I
don't
know
if
that
works
for
anyone,
but
otherwise
you
know
if,
if
the
two
of
you
would
like
to
meet,
that's
one
thing
another
thing
and
please
feel
free
to
like
we
have
the
golang
sig
meeting
on
thursday.
We
can
actually
use
some
of
that
time.
For
this
I
was
just
saying.
B
The
same
thing
yeah:
let's
just
do
that,
let's
just
like
sandbox
like
30
minutes
at
the
front,
and
it's
like,
I
don't
think
we
have
that
much
more
to
actually
sign
and
I
think
we're
doing
a
pretty
good
job
pruning
it.
So,
let's
just
do
that:
let's
just
block
off
30
minutes
as
a
time
box
and
if
we
get
through
it
faster.
Let's
just
do
that
great
cool.
How
does
that
sound
yep
that
works
for
me?
B
Awesome!
Okay,
cool!
Well,
I'm
gonna
finish
up
on
all
the
action
evidence
I
have
for
today
and
then
yeah,
we'll
we'll
do
that
I'll
meet
you
guys
on
that
as
well.
Thanks.