►
From YouTube: 2022-06-08 meeting
Description
cncf-opentelemetry meeting-2's Personal Meeting Room
A
A
C
Okay,
let's,
let's
start
looking
a
minute.
C
Oh
sorry,
yeah
thanks.
So
yesterday
yesterday
we
talked
about
about
semantic
conventions
a
little
bit
more,
so
I
can
give
a
quick
overview
to
anyone
who
wasn't
there,
and
I
can
also.
I
can
also
show
you
the
project
project
board.
I
added
some.
I
added
some
draft
items
on
that
board
and
we
can
go
through
them
if
you
want.
B
Okay,
so
yeah,
I
think
you
know
we
were
internally
talking
about.
You
know
this
topic
about
sending
for
for
base
pages
space,
initial
web
page,
sending
the
transparent
header
in
the
meta
tag
or
in
the
server
timing,
header
and
then
accessing
through
you
know
some
apis
about
that
is.
Is
that
the
solution
you
know
we
are
standardizing
on?
If
so
are,
are
there
any
ways
to
automate
it
on
the
server
side?
B
You
know
that
insertion
of
our
generation
and
insertion
of
you
know
this
header
is
the
apm
agents
anyway
helping
with
that
process,
or
is
it
entirely
on
the
customers?
B
D
E
B
Do
you
know
which
languages
do
that
today.
D
B
Is
there
any,
could
you
send
any
links
to
it
anything
related
to
it?
Martin.
D
B
Yeah
yeah,
I
think
that
will
be
helpful,
because
what
we
also
want
to
know
is,
is
it
by
it's?
We
have
to
enable
it
through
some
flag
right,
it's
not
always
going
to
be
happening.
B
I
I
also
want
to
know
how
do
you
detect
when
to
send
it?
So
so
there
might
be
some
more
clarity
we'll
need
there,
so
I
can
get
into
it.
A
Is
this
link
nav
that
you've
posted
so
that
this
is
a
code
out
of
application
insights
that
does
exactly
this?
It?
It
looks
for
a
trace
parent
or
a
server
timing
with
a
trace
parent
name.
I
recently
moved
it.
It
was
now
common
core.
I
obviously
moved
to
our
shared
core
so
that
we
could
use
it
more
readily.
It's
wrapped
up
in
a
function
called
find
the
trace
parent,
because
by
default
it
doesn't
do
it.
B
That
we
need
in
the
client
no
actually
my
question
is:
this
is
on
the
client
side
in
the
agent
side,
but.
B
Back
end
server
side,
who
generates
it
and.
A
And
in
which
cases
and
there's
actually
a
follow-up
question
as
well
in
terms
of
the
trace
parent
that
it
writes,
does
it
use
the
span
id
I.e,
the
parent
id
of
the
hosting
environment,
or
does
it
create
a
new
one
so
that
the
web
page
has
its
own?
That's
it.
A
So
you
probably
got
the
trace
id
to
be
the
same.
But
do
you
want
the
span
id
to
be
the
same,
and
I
think,
as
martin
and
I've
already
said
it,
it
really
does
depend
like
a
lot
of
our
agents
live
as
plug-ins,
so
we
don't
have
access
to
the
runtime
that
the
application
is
running
in.
A
So
we
don't
actually
do
this
today
by
default,
because
you
need
to
do
it
if
you're
like
in
a.net
app,
you
need
to
have
it
as
part
of
your
razer
template,
so
it
burns
it
onto
the
return
to
page.
A
The
reason
for
the
server
timing
is
so
that
it
gives
a
little
bit
more
access,
but
if
you
don't
have
access
to
the
context
by
either
create
id,
you
don't
know
what
values
to
go
in
there.
You
can
burn
anything
you
like
and
that's
part
of
the
issue,
it's
a
case
of
if
it's
in
proc,
it's
easy.
If
it's
not
in
prop
it's
like
well,
where
do
you
get
the
values
from.
B
I'm
not
sure
I
fully
understood.
Are
you
talking
about
the
reliability
of
the
the
span
id
in
the
in
the
trace
id
talk
about
two
things.
A
So
so
to
answer
your
question,
the
the
first
part
was
where
who
writes
this
and
the
who
writes
this
depends
on
the
application,
but
simplistically
it
has
to
be
in
the
application
space
so
in
in
the
in
the
server
side.
Sdk,
because
that's
the
point
at
which
it
has
access
to
the
trace
id,
which
is
the
major
component
of
the
the
transparency
on
the
right.
So
that's
really
going
to
be
in
the
for
the
sdks
on
the
server,
so
that
has
access
access
to
that,
depending
how
it
passes
around
the
trace
id.
A
The
other
question
is
because
I've
got
an
internal
discussion
going
on.
This
is
when
you
burn
that
trace
parent
onto
the
page,
either
as
a
header
or
as
a
meta
tag.
A
You
want
the
trace
id
to
be
the
same,
but
do
you
want
the
parent
id
the
id
the
span
id
to
be
the
same
as
the
span
id
that's
of
the
request,
that's
generating
the
page,
or
do
you
want
to
create
a
new
one
to
say
this
is
now
the
page
and
you
want
the
page
to
inherit
that
and
the
other
side
of
that
is,
if
you
have
it,
as
the
the
server
writes
the
span
id
of
itself
on
the
client.
Do
you
want
to
say?
A
B
Yeah,
and
on
top
of
it,
I
think
mike.
My
question
is:
how
does
it
work
when
there
are
intermediate
reverse
proxies
right,
like
typically
on
on
the
back
end,
you
know
between
your
application
or
in
front
of
your
application
server.
Let's
say
there
is
a
you
know,
an
nginx
or
an
unwave
proxy
or
something
right.
You
know
they
might
add
their
own
span
in
between
and
they
might
not
be
able
to
modify.
You
know
this
header,
so
if,
if
they
were,
then
you
know
that
also
comes
into
consideration.
A
A
But
they
should
be
hello.
Well,
they
may
not
be
able
to
do
the
meta
tag.
That's
definitely
the
headers.
The
headers
are
fast
and
clear
they're,
not
in
the
in
the
html
page
itself.
B
Okay,
okay,
so
one
aspect
you
are
suggesting
is
at
least
we
should
in
the
in
the
standard
somewhere.
We
should
at
least
specify
you
know.
What
is
the,
how
is
the
span
id
specified
in
the
trace
id
is
supposed
to
be
used?
Is
it,
though,
exactly
the
two
points
you
consider
right?
Is
it
supposed
to
be
considered
the
parent
or
is
it
supposed
to
be
the
one
that's
the
client
would
use
yeah?
Is
it
the
generating
span
or
the
one
that.
F
C
D
A
A
Correct
yeah
not
all
browsers,
have
that
api
to
access.
A
A
I
think
the
instrumentation
the
document
load
instrumentation
that
bart
created
in
the
javascript
sig
also
effectively
looks
for
the
traceparent
meta
tag.
Only.
I
don't
believe
it
looks
for
the
server
timing.
C
A
C
There
was
at
some
point
there
was
a
discussion
about
having
a
response
header
into
the
w3c.
The
yeah,
like
you,
have
like
the
trace,
trace
state
and.
C
A
C
So
is
there
a
response?
Header
for
ajax
calls.
C
And
it's
it's
called
it's
transparent.
C
Right
right,
but
it's
for
I
guess
I
guess
it's
a
different
use
case.
Yeah.
C
A
B
C
A
B
That
is
not
a
standard
right,
so
so
the
it
is
not
part
of
any
standard
for
the
proxies
to
look
at
the
server
timing
header
or
to
look
at
the
meta
tag
to
replace.
A
It's
not
not
standard
today,
yes
on
the
implementation,
but
they
would.
They
do
technically
have
access
to
it,
of
course,
depending
on
the
language
they're
they're
implemented
in
and
the
runtime.
A
Robust
solution
to
this
problem,
probably
better
saying
it's
an
incompletely
defined
problem,
so
I
think,
trace
context
is
well
defined.
It's
just
how
that's
passed
to
and
from
the
client.
That's
the
undefined
portion.
D
A
Browsers,
specifically,
for
other
environments
that
are
doing
this
themselves,
like
the
browser
itself,
could
do
it,
but
the
javascript
runtime
or
any
other
runtime.
That
effect
isn't
a
vm
of
some
kind
that
doesn't
do
the
networks
calls
itself,
but
doesn't
really
do
that.
D
A
It's
a
bit
like
the
ajax
requests.
It
has
to
be
in
the
allow
for
javascript
to
get
access
to
that
returned
header
from
a
particular
cross
argument
server.
So
it's
not
like
automatic
either
it's
in
the
same
domain.
A
A
Is
problematic
yeah
before
trace
application
insights
had
this
thing
called
request
id
and
it
would
use
a
meta
tag.
So
that's
long
before
me.
C
So
do
we
do
we
think
that
this
should
be
defined
as
a
spec
for
the
backend
instrumentations.
B
I
I
I
I
don't
know
hundred
percent,
because
even
if
we
do
remember
that
this
is
not,
this
is
not
going
to
completely
solve
the
problem
in
all
for
all
use
cases.
So
so
I
I
don't
know.
B
It
no
no,
but
I
think
martin's
question
is:
should
the
standard
include
some
text
to
indicate
how
to
interpret
this
transparent
header
for
on
on
exactly
the
the
point
that
you
said:
never
whether
that
span
id
included
in
the
transparent
does
it
refer
whatever.
A
Yeah,
I
think
that
we
should
probably
define
that,
in
terms
of
the
client
run
in
terms
of
how,
if
we
find
a
meta
tag
or
a
server
id,
how
we
interpret
that
or
whether
we
have
an
option
to
change.
B
A
Yeah
I've
had
two
different
conversations
with
two
different
ways
around
so
so
far.
It's
50
50.
B
Actually,
it
should
be
the
the
client
spans
id
right
so
that
the
server
can
say
that
hey
for
my
span.
I
have
I
have
used
this
as
my
parent
and
therefore
you
know
you
better
generate
your
spam
with
this
id.
A
Yeah,
so
I
that
was
my
preference,
like
I've
had
a
pr
out
for
ages
that
I
still
haven't
pushed
in
because
of
this
discussion.
My
original
implementation
did
exactly
that.
It
pushed
the.
I
don't
know
if
you
guys
have
heard
of
the
pet
clinic
example.
I
think
hacked
that
and
made
pat
clinic
return.
The
the
java's
server-side
span
id
on
the
basis
that
the
client
can
then
choose
what
it
wants
to
do.
A
A
A
My
argument
is,
it
gives
flexibility
to
the
client,
so
it
depends
on
the
application.
The
application
may
just
be
oh
well
it
just.
It
starts
up
and
shuts
down
account
together.
B
B
Is
it
will
be
a
dangling
span
right?
It
will
be
a
span
in
the
trace,
but
you
don't
know
what
its
child
or
parent
spanish
in
terms
of
linking
yeah.
A
B
Just
stick
to
that
yeah:
okay,
martin,
you
may
want
to
you,
know,
create
an
issue
and
to
track
it.
A
Probably
one
of
the
other
things
that
we
talked
about
yesterday
for
those
of
you
not
here
it
is
in
the
notes.
So
last
week
I
talked
about
creating
effective
sandbox
repo.
I've
got
the
request
out
on
the
basis
that
request
will
eventually
get
approved
at
something
created.
At
some
point,
I
I
do
have
a
local
repo,
where
I
am
slowly
merging
things
in
I
had
everything
compiling
and
linting.
A
I
did
eventually
get
the
tests
working
or
some
of
the
tests
working
yesterday,
I.e
by
removing
all
the
node
and
the
webpack
stuff,
and
then
I
decided
okay
well,
there's
been
a
release.
Let
me
go
and
resync
that
so
it's
part
of
reseeking
that
I
realized.
A
I
was
a
bit
too
aggressive
in
my
in
my
reordering
of
things,
so
I'm
spending
some
time
fixing
that
up
so
that
it
will
be
a
little
easier
to
resync
because
it's
a
manual
process
to
resync,
because
I'm
changing
the
package
names
from
like
open,
telemetry.js
api
to
opengls
sandbox
api,
so
that
it's
all
local
and
so
in
fact,
I'm
prefixing
everything
or
with
sandbox
in
there,
because
that's
the
whole
purpose
of
this
report.
A
Hopefully
that
will
get
approved
shortly
and
then
we
can
go
from
there.
The
idea
I'm
thinking
is
we
can.
We
can
have
main
as
effectively
the
main
one
where
we
use
to
sync
the
real
code
in
and
then
we
can
go
and
create
forks.
A
A
A
On
martin's
comment
earlier
about
the
semantic
conventions,
so
hopefully
I'll
get
a
chance
later
in
the
week
to
effectively
sync
with
ram
and
we'll
start
grabbing
what
we
can
and
fleshing
it
out
a
bit
more,
at
least
from
the
microsoft
side
of
things,
and
then
we
can
merge
that
with
everyone
else's
views
as
well.
B
Yeah,
no,
I
added
one
more
topic
to
the
agenda.
We
could
talk
about
that
briefly
before
you
go
over
the
project
board.
Okay,.
B
And
this
is
the
topic
I
think
we
discussed
maybe
last
week
about
the
data
attribute,
having
an
object
as
a
its
value
and
never
had
a
you
know,
issue
asking
about
that.
Support
and
tigran
replied
that
log
records
allow
attributes
to
have.
You
know
nested
values,
so
we
we
should
check
whether
your
our
sdks
have
support
for
that
and
and
then
you
know,
just
use
it.
A
Spain
events
according
to
the
spec,
also
support
it
that
that's
part
of
the
reason
I
raised
the
issue
was
there
is
inconsistencies.
So
if
you
look
at
the
the
common
attribute
definition,
it
says
it
they
take
primitives
but
further
down
the
page.
It
talks
about
attribute
collections
and
specifically
states
that
those
attribute
collections,
including
spain
span
events,
span,
links
and
metrics
resources
and
log
records
can
have
nested
objects,
services.
A
A
I
would
like
to
see
the
semantic
inventions
that
we
create
for
events
being
okay.
This
is
what
an
event
looks
like
and
it
can
live
as
a
span
event
or
a
log
event.
A
That's
in
terms
of
how
we
get
it
into
those.
I
think
it's
a
little
bit
open.
I
like
the
log
emitter
concept
that
tigran
and
the
log
sig
have
been
talking
about,
and
if
we
follow
that
and
have
it
as
a
separate
api,
we
can
say
we'll
ask
the
logs
for
a
lot
for
an
event.
Emitter,
and
we
can
ask,
you,
know,
extend
spans
and
now
ask
spans
for
an
event
emitter,
because
the
existing
ad
span
on
spam
event
on
a
span
for
adding
a
span
event
only
takes
a
name.
C
A
Or
something
like
event,
emitter
yeah,
so
I
think
you
you
ask
the
destination
for
where
the
event
is
going
to
be.
Give
me
a
thing,
so
I
can
go
and
start
adding
events
to
it.
A
And
I
think
most
of
them
most
people
do,
I
know
santosh
doesn't,
and
I
think
then,
a
lot
of
people
are
on
the
fence
about
which
way,
but
my
preference
definitely
is
to
keep
them
separate.
A
But
that's
why
I
started
looking
into
this
event
and
found
this
inconsistency,
because
ted
was
telling
me
that
no
no
all
attributes
support
nesting
and
then
tigran.
Is
that
saying?
No?
No,
it's
only
logs
that
support
nesting
and
when
you
read
it,
I
can
see
why
both
people
are
saying
that,
like
I
started
with,
proto
proto
supports
nesting
for
all
attributes
and
then
the
spec
is
depending
which
bit
you
want
to
read.
A
That's
fine
in
the
case
of
the
javascript
sdk
and
api
today,
a
span
attribute
is
an
attribute,
so
it
does
support
nesting
today,
because
fan
attribute
there's
now
an
alias
of
attribute.
B
A
B
So
what
is
our
requirement,
though?
Do
we
have
use
cases
where
we
want
nested?
I
mean
objects
for
span.
Events
too.
E
C
So
very
thinking,
basically
specifying
that
all
attributes
on
events
should
be
nested.
B
So
so
maybe
the
next
step
could
be
nev.
I
think
you
should
maybe
reword
your
issue
saying
you
know
we,
we
you,
we
only
care
about
the
span,
events
and
events
right.
I
I
think
we
are
not
so
much
concerned
about
the
spans
themselves.
Having
you
know
this
requirement.
A
B
A
But
the
spec
allows
it
in
one
place
and
doesn't
allow
it
another
that
should
be
fixed,
yeah.
Okay,
I've
got
that.
That's
what
my
spec
issue
is
about.
It's
okay,
so
my
preference
is
the
fact
that
it
it
is
allowed
in
the
in
protobuf
and
attribute
collection,
specifically
called
that
as
a
my
preference
would
be
to
we
just
allow
it
like
it's
defined,
but
we
just
it
just
needs
to
be
clarified
if
they,
if
it
ends
up
being
clarified,
saying
okay,
attribute
collection
should
not
be
on
spans
or
span
events.
B
A
No,
I
think
it's
a
collection
of
the
keys
of
such
collections
are
unique.
B
B
Primitive
values
or
an
array
of
primitives
or
an
actual
attribute
itself,
yeah
yeah,
so
so.
E
B
Attribute
value,
the
value
can
only
be
one
of
the
primitive
data
types.
That's
what
it
says
at
the
top.
A
Yeah-
and
this
is
where
it's
like,
if
you're
looking
at,
I
think
it's
basically
yeah
so
in
attribute.
It
actually
already
says
array
of
primitive
values,
so
attribute
collections
are
are
well.
I
was
reading
that
as
nested,
so
this
is
where
the
yeah,
the
inconsistency,
because
it's
already
called
out
that
an
attribute
is
a
primitive
or
an
array
of
primitives.
So
why
would
they
then
need
to
say
an
attribute
collection?
B
Yeah,
I
think
I'm
also
worried
whether
the
sdk
implementations
would
be
able
to
handle
this
correctly,
where
some,
since
this
is
attributes,
is
part
of
the
common
library.
You
know
how:
how
do
how
do
you
ensure
that
is
the
span?
Attributes
do
not
take
objects
as
attribute
values,
whereas
the
log
record
attributes
you
know
can
can
allow
it
yeah.
A
And
that's
part
of
the
clarification
it
needs
to
be
that
they
support
it
or
they
don't.
My
concern
is
more.
The
exporters
like
if,
if,
if
as
defined
attributes
are
nested,
the
sdk
should
support
that.
B
Okay,
yeah:
let's
follow
up
on
this,
then
yeah.
A
Which
is
why
I
raised
it
in
the
spec
meeting
this
week
because
it
was
allocated
to
josh
and
I
haven't
seen
josh
on
them
on
the
course
recently.
I
think
he's
at
the
moment.
C
Okay,
so
this
might
be
a
detail,
but
I'm
wondering
if,
if
sending
the
attributes
as
nested
is
something
that
is
done
by
individual
instrumentations
or
if
it's
optimization,
like
in
the
exporter.
A
A
I
like
the
nested
objects
in
there
and
we
will
deal
with
it
going
out
the
door,
so
I
know
that
that's
going
to
be
a
requirement
that
we
need
to
have
as
an
optional
field
on
all
events,
otherwise,
that's
going
to
be
unusable,
which
is
why
we,
I
think
logs,
was
chosen
originally
because
logs
was
the
only
thing
that
supported
unless
that.
Well,
we
believed
blogs
is
the
only
thing
that
supported
nested
attributes.
A
That
is
the
other
part
of
it.
Yes,
there
is
a
hard
requirement
that
we
need
that
we
need
to
allow
customers
to
do
it,
but
because
of
the
support
that
as
an
optimization,
I
think
we
should
also
divide
out
to
find
our
semantic
conventions
as
such,
like
within
microsoft,
we
do
have
our
events
defined
as
in
object
hierarchy,
so
yeah.
C
You
know
I'm
thinking,
like
all
all
attributes
now,
in
the
semantic
conventions,
have
name
spaces
so
like
exception.something,
right
and
three
attributes
so
like
you
could
you
could
potentially
remove
the
exception,
name
space
being
repeated
so
yeah
and
that
could
be
done
automatically.
That's.
A
A
C
Okay,
so
I
want
to
go
over
the
the
project
board,
but
before
I
do,
I
noticed
that
there
are
a
few
new
people
in
this
in
this
group,
and
I
should
have
mentioned
it
in
the
beginning
anyway,
if
you're,
if
you,
if
you're
here
for
the
first
time,
welcome,
let
me
know,
let
us
know
if
you
have
any
questions
or
any
topics
you
want
to
discuss.
A
I'm
just
gonna
add
that
welcome
and
if
you've
got
mobile,
app
experience
very
welcome
because
we're
we're
very
heavy
browser
at
this
point.
So
we
would
definitely
like
input
from
other
client-side
runtimes.
G
Hi
there
I
just
wanted
to
say
hi,
my
name
is
purvi.
I've
recently
joined
honeycomb
and
one
of
the
things
I'm
working
on
the
telemetry
team.
I'm
sure
that
you
folks
have
maybe
come
across
some
of
the
other
telemetry
team,
folks,
maybe
through
the
javascript
sync,
but
I'm
I
will
be
focusing
on
the
browser
along
with
my
co-worker
winston
as
well.
So
I
just
wanted
to
say
hi
sorry
for
the
lack
of
mobile
experience,
but
I
do
have
some
react
native
stuff
going
on
too.
So
I'm
just
really
pumped
to
be
here.
C
Okay
yeah,
so
let
me
just
share
the
project
really
quick.
C
So
we
we
talked
about
this,
I
think
last
week
and
the
intent
is
just
to
have.
I
have
a
place
where
we
can
see,
what's
which
things
are
in
progress
and
which
things
are
being
worked
on
right
in
progress
and
things
that
we
want
to
backlog.
The
things
you
want
to
work
on,
I
have.
I
have
added
a
number
of
draft
things
in
the
icebox
column.
C
These
are
not
necessarily,
I
think
we
should
go
through
these
and
and
make
sure
that
they
make
sense
to
everyone,
but
then
they're
currently
only
drafts,
so
they
need
to.
They
should
be
converted
to
issues
if
they
decided
to
actually
do
those.
So
from
my
perspective,
the
the
top
thing
that's
going
on
right
now
is
is
the
events
api
which
is
being
discussed
with
btgrand.
C
Once
once,
we
just
have
a
decision
on
what
that
api
should
look
like
we
can,
you
can
hopefully
start
implementing
like
the
experimental
version
of
it,
and
then
then
we
can.
We
could
potentially
start
working
on
instrumentations
that
use
it.
So
so
I
think
the
the
top
thing
in
my
mind
for
this
group
is
to
to
discuss
the
semantic
conventions
from
you
know
for
the
next
near
future.
C
So
I
added
you
know
the
top
four
items
are
about
semantic
conventions,
so
one
is
just
defining
the
conventions.
The
first
one
is
for
defining
conventions
for
for
events.
What
what
what
actually
constitutes
an
event
and
then
the
next
three
is
errors,
so
errors
from
browser
semantic
permissions
for
interactions
and
semantic
commissions
for
timing
events-
you
can,
I
can
see
these
as
the
top
priority
ones.
B
C
You
can
just
at
the
bottom
click
on
add
item
and
just
you
search
for
it.
You
can
paste
in
the
url
and.
C
B
C
B
Them
we
can
add
a
label.
Okay,
yeah.
C
Okay
and
then
so
beyond
those
we
have,
I
added
the
implementation
of
the
experimental
events
api
in
the
in
the
sdk.
This
other
for
ephemeral
resources
is
something
that
ted
is
going
to
be
working
on.
We
need
that
for
session
id,
and
then
we
talked
about
having
two
additional
tabs
sometime
in
the
future,
which
which
outline
kind
of
like
a
guideline
for
what
an
sdk
for
browser
should
look
like
when
an
sdk
for
mobile
should
look
like
for
for
js
for
js
we're
gonna
need
an
otop
exporter.
C
I
believe
that
the
js
sake
is
soon
gonna
be
working
on
logs,
so
maybe
they
maybe
we
don't
need
to
work
on
this,
and
then
I
also
have
a
ticket
here
that
I'm
actually
gonna
create
issue
for
this
one
implement
the
browser
attributes
in
the
js
sdk,
which
is
actually
part
of
the
spec.
Now.
C
So
if
you
think
of
anything
else,
please
add
it
edit
here
and
then
I
think
for
for
next
next
week
we
can,
we
can
maybe
start
defining
this.
The
semantic
conventions
a
little
bit
a
little
bit
further.
A
Could
you
paste
the
link
to
this
board
in
the
in
the
dock
so
that
we've
got
it
there?
I'm
not
not
seeing
it
just
probably
at
the
top
there
in
the
work
in
progress
that
way,
everyone
got
access
to
it
more
easily.
B
I
I
was
told
earlier
somebody
mentioned
that
the
json
format
is
is
still
not
part
of
the
stable
spec.
Is
that
still
the
case.
A
Currently
yeah
they
came
up
in
both
the
maintainers
and
the
spec
feeding
this
week
as
well.
So
there
were,
things
are
slowly
pushing
forward.
There
was,
I
think,
in
this
spec
meeting,
I
think,
martin
you
were
in
that
one
where
bogdan
was
talking
about,
should
the
case
of
enums
effectively
frees
both
the
integer
value
as
well
as
the
name
of
the
enum.
A
I
my
preference
would
be
yes,
I
think
of
who
others
were
as
well,
but
yeah
at
the
moment,
the
it's
it's
getting
there,
but
it's
not
quite
there
yet,
which
is
also
why
the
js
jason
exporters
are
still
an
experimental
because
it's
not
frozen
okay.
So
then,
martin.
B
B
A
Just
so,
we've
got
a
russians
there,
so
we
know
whether
they
well,
what
the
state
of
the
json
exporter,
which
is
really
the
project
definitions,
are
and
any
related
items
we
find
that
might
be
a
column
in
the
board.
We
want
to
put
in
stabilization
items
or
something.