►
From YouTube: 2023-01-18 meeting
Description
cncf-opentelemetry meeting-2's Personal Meeting Room
A
B
I,
don't
have
anything
specific
today.
Yesterday,
I
was
just
working
on
getting
the
sandbox
going.
I've
now
got
the
API
building
and
the
node
test
running
and
I'm
just
working
on
getting
the
the
browser
test
running
which
don't
currently
exist
for
the
API.
B
This
is
all
happening
from
the
automated
script.
Create
these
configs,
so
it
looks
like
I
can
still
get
this
hit
Friday.
So.
C
A
Now
from
the
explosion,
yesterday,
I
didn't
catch.
One
item
like
what
was
the
argument
from
the
existing
JavaScript
maintenance
about
keeping
both
together
browser
and
node.
B
The
way
it
was
originally
created,
it
was
just
going
to
be
JavaScript,
so
you
know
because
they're
language
Basics,
not
Target,
you
know
and
environments,
so
not
not
node
versus
browser,
and
you
know
even
like
prior
to
me
joining
that.
That
was
the
approach
but
yeah.
It
really
was
the
case
of
like
it.
Originally
Bart
was
one
of
the
maintainers
and
he
was
heavily
focused
on
the
browser.
B
He's
sort
of
you
know
step
back.
Well,
probably
this
like
18
months
ago
now,
so
things
changed
a
little
bit,
but
it
was
really
brightening.
Myself
were
really
the
only
two
focusing
on
browser,
but,
for
example,
like
I've
got
the
bundles
building
now
out
of
the
sandbox,
and
the
API
on
its
own
was
48k.
B
And
that's
just
the
API
like
it's
like
that's
just
that's
just
done
on
its
own
is
too
big
I'm,
just
gonna
find
what
the
minified
version
of
that
was.
That
was
48k
here
with
all
the
comments
removed
box.
B
The
packages
epi
build
bundle,
okay,
so
the
UMD
yeah,
so
the
API
is
48k
and
the
minified
version
of
that
is
still
19k
so
and
that
that's
based
on
version
1.3,
so
I
haven't
actually
synced
in
a
little
bit
because
I
want
my
script
to
go
through
the
the
full
update
and
that's
based
on
es5.
So
if
I
did
an
es6,
Builder
that'd
probably
be
a
little
bit
smaller.
B
Of
course,
the
esthetic
I
don't
have
any
expand
up.
Okay,.
B
So
IIs
it's
still
48k
19k.
So
how
close
to
14K?
Okay,
so
it's
47.4,
uncompressed
minified,
it's
18.7,
so
yeah.
B
A
B
Yeah
once
we
target
a
particular
Mr,
although
the
fact
that
that's
the
full
that's
what's
being
fully
exported
from
the
API
yeah
anyway,
once
I
get
the
bundles
there,
we
can
then
play
with
them
so
I'm,
concentrating
on
having
each
individual
component,
bundled
because
I
want
to
create
size
tests
so
that
we
can
validate
what
the
sizes
are
but
yeah.
When
we
come
to
create
a
distro
or
a
full
SDK,
we
should
be
on
the
tree
shake
a
bunch
of
stuff,
but
that's
that's.
You
know,
48k
seems
large
yeah
you're
right.
B
The
metric
stuff
is
not
insignificant,
but
six
files
in
each
file
is
between
two
and
six
K.
So
yeah,
it's
18.2k,
Road
type
type
script.
So.
C
B
So
yeah
the
way
the
script
is
bundled
I
just
have
a
config
that
says,
move
auto,
merge,
JS
API
into
this
folder,
and
then
the
scripts
go
and
create
all
the
other
stuff
around
it.
To
automatically
generate
this
stuff.
Could
roll
up
so
I
found
in
my
earlier
tests
at
the
end
of
21
I
can
create
small
abundance
with
roll
up
than
I
can
with
webpack,
which
is
when
I
first
did
all
this?
That's
why
I'm
using
roll
up
to
to
generate
these.
A
B
First
then
Force,
an
update
from
the
the
JS
and
Contra
repos
into
Auto,
merge
and
then
run
that
through
and
make
sure
that
my
I
can
bring
the
API
over
without
getting
merge
complex
or
they
can
make
some
nice
script
that
can
resolve
the
merge,
complex
and
then
I
was
going
to
start
bringing
in
the
other
ones
playing
changing
my
config
to
include
like
somatic
conventions
and
any
other
package
that
we
need,
which
isn't
a
lot
at
this
point,
but
we
can
grow
them.
B
Originally.
I
was
trying
to
not
bring
the
auto
merge
folder,
which
is
the
folder
that
gets
created
in
the
merge
Branch
into
Main,
but
that
was
causing
grief
when
I
run
a
second
pass,
because
I
had
to
delete
it
from
my
staging
one
that
we
tried
to
delete
it
so
Maine's
going
to
have
the
auto
merge,
folder
as
well.
Just
it'll
have
the
stuff
that
it
wants
to
play
with
pushed
into
it
into
its
package
structure
for
building,
but
it
means
we
can
then
do
a
git
move.
B
I
saw
you
updated
the
thing,
but
no
I
haven't
had
a
chance
since
I
was
going
right.
Apart
from,
like
you
know,
several
hours
of
meetings
I
was
trying
to
be
trying
to
focus
on
the
sandbox.
A
I'm
now
looking
for
closing
any
open
items
from
that
dock,
I
want
to
wait
before
I
raise
a
pair
to
the
spec.
You
know,
I'm
I
was
thinking.
Maybe
we
should
also
start
doing
the
prototyping
and
then
you
know
we'll
have
something.
A
Yeah
but
I
can
open
the
pr
for
the
resource
attributes
that
we
have
identified
so
far.
That's
not
going
to
change
yeah
the
attributes
inside
the
event.data.
We
said
we're
not
going
to
have
semantic
conventions
for
them,
so
so
then
it
will.
There
will
be
two
PRS.
You
know
one
for
the
semantic
conventions
for
the
resource
attributes
and
this
overall
document.
B
I
haven't
been
monitoring
the
status
of
the
the
experimental
resources
which
is
Ted's
one,
that
session
ID.
We're
gonna
have
to
deal
with
and
tigrants
one
okay,
it's
in
meetings
and
then
concentrating
on
trying
to
get
us
unblocked
so
like.
If
we
had
the
sandbox,
we
could
be
doing
prototypes
in
there
already
without
you
know
a
lot
faster
than
we
can
in
the
Js
see.
A
Other
than
that,
this
is
more
for
the
lock
sick,
but
briefly
I
reached
out
to
the
cloud
events
folks
asking
them
to
join
one
of
our
six,
so
they
came
back
saying
that
they
are
already.
One
of
them
is
joining
a
messaging
sync.
A
There's
a
Thursday
8
A.M,
Pacific,
Time
messaging
Sig,
where
I
I,
don't
I,
think
it's
also
a
bunch
of
folks.
Like
us,
I,
don't
see
anyone
from
the
TC
members
there
and
even
that's
been
going
on
for
a
while.
B
Yeah
I
saw
that
one
start
up
a
while
ago,
but
at
the
time
I,
just
I'm
already
overload
with
meetings
actually
yeah
the
only
one
now
at
eight,
because
there
was
the
the
sampling
and
the
there's
another
one
that
looks
like
they
run.
Bi-Weekly
at
the
airport
tomorrow
also
happening
on
Thursday,
but
I
was
going
through
for
a
bit.
Okay.
A
So
yeah
I
asked
the
cloud
events
folks
to
I
mean
basically,
the
agenda
was
to
understand
like
how
they
produce
events,
since
you
know,
in
our
case
also,
we
have
been
discussing
about
apis
yeah
to
produce
events,
API
and
the
data
model,
and
they
came
back
saying
you
know
it's
not
necessarily
the
right
way
to
look
at
it.
You
know
they
Cloud
events
is
more
of
a.
How
do
we
expose
existing
event?
Data
in
a
common
location,
people.
A
B
A
I'll,
try
and
join
the
messaging
stick
tomorrow
to
understand
more
about
it,
but
so
far
it
looks
like
we
may
not
need
to
consider
them
in
the
in
our
events,
design,
yeah,
but
eventually,
if
there
are
use
cases
to
turn
hotel
events
into
Cloud
events,
then
we'll
need
to
define
a
mapping.
B
Yeah
and
I
think
your
basic
structure
works
anyway,
like
if
the
event.data
is
their
data,
the
attributes
they
have
attributes
so
yeah
it
works.
The
only
problem
I
would
see
is.
If
we're
turning
it
into
a
cloud
event,
then
we
have
to
have
some
sort
of
additional
transport,
so
that
would
give
us
poor
connections
leaving
a
box
if
you've
got
like
tracing
metrics
logging
in
Cloud
events.
It's
like
that's
excessive.
B
B
I.
Think
Ted
was
the
only
one
who
raised
it
two
weeks
ago,
three
weeks
ago
and
yeah
there
wasn't
too
many
people
going.
Yes,
we
need
that
they
say.
Why
do
we
need
that
yeah.
A
I'm
not
bringing
it
up
much
now
we
have
other
things
to
do,
but
eventually
I
think
that
will
be
required
for
the
ramsig,
both
mobile
and
web.
A
A
A
Yeah
yeah,
it's
it's
complex
to
set
up
yeah
yeah.
A
And
it
also
needs
a
good
alignment
that
you're,
you
know,
multiple
requests,
your
trace
and
the
logs
exporters
have
to
export
around
roughly
the
same
time,
yeah.
A
No
but
they're
they're
periodic
timer
also
have
to
align.
B
Yeah-
and
there
is
open
discussions
about
that
as
well
in
terms
of
what
are
the
defaults
now,
what
should
be
the
difficult
periods
of
time,
logging
that
that
came
up
a
couple
weeks
ago
in
a
spec
seek
in
terms
of
I
think
Jack
was
driving
the
fact
that
five
seconds
was
too
long
and
they
wanted
more
like
100
milliseconds,
which
you
know
for
a
browser,
is
not
going
to
work,
I,
sort
of
said
that
yeah,
it's
fine
to
have
defaults
as
long
as
they're,
based
around
language
and
runtime
yeah.
B
What
the
runtime
environment
is
for
a
browser,
100
millise.
That
can
be
way
too
long,
considering
it's
going
to
take
several
monthly
seconds
to
even
construct
the
outbound
juicing
packet
based
on
the
full
otlp
definition.
So
now,
if
you've
got
lots
of
events
in
there,
it's
going
to
take
even
longer
and
that
well,
unless
you,
unless
we,
we
start
having
the
the
SDK,
do
it
in
a
worker
thread.
It
blocks
the
UI
as
part
of
that.
B
So,
as
you
say,
to
work
a
thread
or
sorry,
a
web
worker
I
just
think
you
must
think
of
a
missed
trades
yeah.
It's
getting
out
the
browser
metric
support
today
is
only
modern,
browsers,
I
mean
that's
also
an
issue.
C
D
Okay,
sorry
about
that
yeah,
sorry
for
being
late,
some
family
duties,
yeah.
A
No
problem
yeah,
we
didn't
have
much
on
the
agenda,
so
we're
just
chatting
in
general
and
I
wanted
to
see.
If
you
have
any
things
to
discuss,
we
can
briefly
go
over
it
and
then
wrap
up.
D
Yeah
I
know
the
the
only
thing
that
I
have
is
some
I
just
opened
draft
PR
in
the
Js
repo
this
morning
to
separate
the
events.
Events
API
from
the
logs
API
just
you
know,
so
please
take
a
look
at
it
and
if
it
looks
good,
then
it's
not
it's
in
draft
mode
right
now,
so
I,
just
kind
of
the
main
things
are
are
there,
but
like
I,
haven't
updated
the
tests
and
all
that
stuff,
so
yeah.
If
that
looks
good
to
you,
please
review.
D
A
So
the
events
API
does
not
refer
to
the
logs
at
all
right.
D
A
D
D
They
they
said
they
they
do
have
an
implementation
of
the
logs
SDK
and
they
said
they
would
open
a
PR
again,
but
there
was
that
it's
been
a
week.
So
if
they
don't,
then
I'll
I'll
ping
them.
D
A
Okay,
hey
I'm
I'm,
looking
at
the
the
API,
your
PR
and
I'm
I'm
wondering
if
we
should
I
mean
we'll
we'll
I
I
want
to
bring
this
up
in
the
Rocks
again
as
well,
where
we
should,
you
know,
consider
the
event.data
in
the
API
design.
In
the
sense
you
know
the
Amity
event
or
MIT
event
should
not
just
take
name
and
event
or
currently
Amity
event.
A
Just
takes
a
full
event.
It
doesn't
even
take
the
name,
so
the
name
is
inside
the
event:
object,
okay,
so
I
I
think
the
the
event
data
model
that
you
have
is
the
event
interface.
It
might
can
have
you
know,
data
and
other
attributes.
As
you
know,
two
separate
set
of
attributes,
one
that
goes
to
event.data
and
the
other
one
goes
at
the
top
level.
D
Yes,
that's
that's
right,
I
was
I
was
wondering
if
that
should
be
a
separate,
PR.
I,
don't
know
if
it's
going
to
be
something
that
could
hold
up
this
thing.
This
separation,
but
I
can
combine
them
into
one
yeah.
A
No
before
you
do
that,
I
think,
let's
find
out
if
you
know,
if
there
is
agreement
on
that
approach,
yeah.
A
Maybe
now
you
know,
should
we
open
a
issue
so
that
you
know
I?
Think
people
like
John
Watson,
you
know
they
don't
join.
He
doesn't
join
the
these
six,
so
it'll
be
good
to
get
his
views
too.
Yeah.
B
He
commented
quite
a
bit
on
the
thing:
yeah
yeah
effectively
raise
our
spec
issue.
Changing,
although
or
do
we
want
to
like
prototype
it
and
settle
first,
so
you
know
let
Martin
through
the
definition
and
then
we
say
okay.
This
looks
like
it's
going
to
work
and
then
go
change.
The
spec
either
way.
B
A
Yeah
yeah:
let's
bring
it
up
today,
get
some
initial
views
and
then
we
can
follow
up
by
an
issue
if
required,
to
get
more
comments.
D
Also
I
can,
if
I
can
show
you
something
really
quick
hold
on.
D
You
can
use,
can
you
see
my
yep
so
I,
so
I
haven't
touched
the
API
itself
in
this
PR
like
it's
it's
the
way
it
was
before,
basically,
I
just
separated
them
into
a
different
package,
but
the
way
it
was
so
there's
this
event,
emitter
that
now
has
a
mid
function
here.
That
takes
this
event
interface.
D
So
that's
that's
the
same
as
it
was
before.
So
the
event
interface
has
things
like
you
know.
All
the
things
that
you
would
might
want
to
to
collect
on
the
event
like
attributes,
name
timestamp
so
like
one
option,
is
to
would
be
to
add
the
data
as
a
property
on
this
interface.
C
D
C
B
That's
the
way
it
looks
like
yes,
yeah,
it's
probably
nice
to
have
both
like
I,
can
see
the
advantage
of
just
saying
Emit,
and
then
you
a
pack
of
you,
give
it
a
JavaScript
object.
D
Yeah
and
the
reason
is
this
way
like
we,
the
the
original
blogs
API
PR,
there
was
a
discussion
about
what
what
exactly
it
should
look
like
in
JavaScript,
and
so
so.
This
is
what
we
settled
in,
based
on
like
having
like
a
single
interface.
A
Now
I
think
this
format
is
fine,
except
that
here
you
know
the
line
38.
You
know
there
could
be
two
of
them
one,
you
call
it
as
data
and
the
other.
You
call
as
like
other
attributes:
yeah
yeah,
okay,.
B
C
A
So
Martin,
what
did
you
say
about
the
logs
SDK?
Somebody
else
is:
is
going
to
raise
a
PR?
Is
it
yeah.
D
So
there's
a
person
who
actually
had
a
PR
open
with
the
SDK
at
some
point:
they
closed
it
but
I
so
and
then
they
tried
to
make
changes
to
the
event
to
the
logs
API
and
there
was
a
discussion
about
basically
closing
closing
that
PR
for
now,
because
there
were
changes,
other
changes
coming
and
they
said
they
said
in
that
discussion.
I
can
link
it
to
you.
No.