►
From YouTube: 2022-07-13 meeting
Description
Open Telemetry Meeting 1's Personal Meeting Room
A
B
A
A
Yeah,
how
should
we
address
you
I
mean
now
is
jan,
if
I
mean,
should
I
address
you
by
right
yeah,
I
know
I
mean
I
should
be
addressing
you
just
wanted
to
check.
A
Okay,
I
think
probably
I
don't
see
mark
joining
today.
I
think
he
did.
He
did
mention
that
he
won't
be
joining
today.
Tom
did
mention
that
he
would
be
a
bit
late
today.
Devajit
is
not
joining.
I
see
he's
probably
not
there
today.
I
think
we
can
quickly
start
and
let
me
share
my
screen.
A
So
basically,
I
think
when
you
start
the
meeting,
we
just
take
up
the
agenda
items,
everything
which
you
want
to.
A
D
D
C
C
So
the
question
we
want
to
have
is
like
right
now
from
the
github
status.
We
see
the
open
telemetry
still
the
matrix.
I
mean
it's
still
the
experimental
phase.
A
So
I
mean-
let
me
just
go
through
that
if
you
have
gone
through
the
milestone
which
we
are
maintaining
specifically
for
the
telemet
for
the
matrix,
I
think
we
try
to
keep
the
milestone
here
up
to
date
for
the
release
candidate,
but
I
do
foresee
that
this
is
something
which
we
are
trying
to
maintain,
but
I
think
the
dates
which
we
see,
I
think
somehow
this
is
getting
pushed
up
every
every
probably
a
couple
of
weeks.
A
It's
just
that
we,
we
don't
have
enough
bandwidth
from
the
people
to
really
work
on
that
on
the
issues
which
we
have
right
now
pending
for
the
release
candidate.
So
basically,
basically
this
this
is.
This
is
where
we
try
to
keep
up
to
date
to
what.
What
exactly
is
the
pending
pending
items
for
the
matrix
and
what
are
the
dates
which
we
are
planning
as
of
now
for
that
and.
A
Yes,
so
this
would
be
a
release
candidate.
This
means
that
we
feel
that
our
core,
matrix,
sdk
and
api
are
kind
of
stable.
They
can
still
change,
but
they
won't
be
any
drastic
change.
This
does
not
really
mean
that
we
have
feature
complete.
There
may
be
some
of
the
things,
like
example,
and
all
those
things
which
is.
A
Won't
have
any
drastic
api
surface
changes,
and
now
what
we
foresee
is
probably
release
candidate
to
release
would
be
a
couple
of
months,
probably
one
to
two
months
of
time,
frame
between
release
candidate
to
doing
a
release
of
matrix,
hdk,
okay,
okay,
that
that
will
make
some
performance
testing
some
stability
testing
or
more
before
before.
We
do
a
release
for
for
this.
For,
for
the
matrix.
A
Any
further
changes
on
matrix,
even
the
battery
sdk
is
changed.
Specs
is
stable.
We
do
foresee
some
of
the
changes
coming
on
specs,
which
can
have
effect
on
our
release.
Timelines.
E
A
So
we
can't
we
try
to
try
to
keep
up.
A
I
think
it's
it's
something
like
if
we
foresee
any
any
specs
changes
which
will
need
some
changes
in
our
implementation,
feel
free
to
add
item
here,
and
I
think
we
can
feel
free
to
create
an
issue
and
then
we
can
discuss
in
the
meeting.
If
that's
something
should
go
in
the
release
candidate
or
not.
C
I
think
inside
google
we
have
another
delegate
team
working
on
the
open
telemetry.
We
try
to
adapting
the
open
termination
to
google
cloud.
So
we
are
another
team
inside
the
gcp
google
cloud,
which
is
like
building
a
different
solution,
so
we,
I
think
we
I
can
internally
reach
out
to
them
about
them
as
well.
A
I
think
he's
because
of
his
bandwidth
more
going
for
matrix
specs.
I
think
he's
kind
of
not
very
active,
but
I
think
every
now.
Then
we
do
push
him
for
doing
that.
Reviews
for
any
matrix.
A
C
C
A
Okay,
I
think
we
can
just
go
with
the
specs
update.
It's
some
something
I
think
I
just
tried
to
prevent.
Probably
we
can
have
start
discussing
this,
a
generous
text
update
which
had
for
the
couple
of
weeks
and
which
can
have
some,
so
we
may
have
to
do
some
changes
in
our
implementation.
A
Because
of
that
I
think
it
would
be
good
to
you
consolidate
that
and
have
a
discussion,
so
I
just
added
one
section
for
that,
so
one
of
them
is
basically
I
have
the
support
for
the
partial
success
in
a
hotel
to
export
message.
A
A
Joshua
has
approved
it
and
I
think
if
this
gets
approved,
we
may
have
to
do
our
the
changes
in
the
otlp
exporter.
So
it's
just
basically
the
the
status
of
the
export
earlier.
We
just
had
success
in
failure,
but
now
we
are
also
going
to
add
partial
success
as
one
of
the
status.
That
means
some
set
of,
I
mean
I
mean
not.
A
All
the
spans
were
successfully
uploaded
or
exported
that
that's
the
status
which
probably
would
get
added
as
part
of
this
change
and
that
ones
that
get
merged
we
do
have
to
do
changes
in
our
otp
exporter,
both
for
locks,
traces
and
also
for
matrix.
A
There
has
been
otep
for
events
and
lock
cp.
I
think
we
already
had
discussed
it
earlier
also,
and
that
has
been
merged
and
even
the
the
sdk
has
been
updated
for
that.
So
we
do
need
to
use
log
record
and
log
emitter
in
we
do
have
to
modify
our
log
sdk
to
incorporate
log
recorded
log
emitter.
A
I
don't
think
we
have
a
issue
for
this.
Let
me
see
correct
me,
tom
or
assad
I
mean
if
we
have
an
issue.
If
it
is
not
there,
we
probably
have
to
create
a
letter
for
that.
Okay,
yeah,
probably.
A
There
has
been
some
changes
and
some
discussions
happening
for
more
clarity
on
how
matrix
readers
are
in
terms
of
matrix
deleted,
how
the
aggregation
is
doing
as
of
matrix
reader.
We
do
specify
the
aggregation
temporarily
on
that,
but
I
think
there
are
some
more
clarification
required
and
that
may
need
some.
It's
still
not
pr,
but
it's
more
of
a
discussion
going
on
and
it
so
if
this
gets
kind
of
approved,
I
think
it's
probably
may
have
a
changes
expected
in
the
matrix
reader
so
probably
need
to
keep
an
eye
on
this.
Also.
A
So
that's
something
I
added.
I
don't
think
that
would
be
still
complete.
I
probably
we
just
need
to
see
more
specs
changes
which
are
which
are
coming
up
or
which
I've
already
done
in
last
week.
That
would
be
helpful
for
for
us
to
foresee
specifically
for
matrix.
If
anything,
we
need
to
add
in
our
milestone.
A
Basic
changes,
I
think
this
is
something
which
is
smushed
so
sr.
I
think
it
was
merged
once
you're
back
from
vacation,
as
we
were
anticipating
time
to
review,
and
I
think
this
is
this
is
one
of
the
good.
I
think
one
part.
I
think
this
would
be
definitely
helpful
for
both
for
zipkin
and
otlp
http
exporters-
that's
great
and
yeah.
A
We
have
to
remove
the
changes
where
we
were
and
because
there
was
a
pr
for
the
specs
changes
where
it
was
anticipated
that
there
would
be
possible
to
do
simultaneous
export
calls,
so
the
processor
should
be
able
to
do
simultaneous
export
calls
for
doing
the
actual
upload,
but
that
pr
was
not
approved
in
the
same
format,
and
I
think
it
was
not
so
as
of
now
as
of
now
the
process
span
or
log
processor
cannot
make
a
concurrent
export
a
api
cons,
so
that
change
was
reverted
back
before
doing
the
merge.
A
A
Async
functionality
added
in
the
exporter
that
includes
the
async
changes
for
the
girl
library,
which
we
are
using
so
yeah.
That's,
that's,
that's
something
which
is
there
and
I
think
this
is
merged.
We
should
be
good
to
close
it.
A
A
Matrix
progress
yeah-
this
has
been
bit
slow.
I
think,
probably,
let's
go
to
the
time
milestone
once
again,.
A
This
has
been
bit
slow,
partly
because,
probably
you
are
not
there
and
yeah
you
have
been.
You
have
been
doing
lots
of
contributions
on
matrix
and
I
have
been
mostly
involved
in
this
change.
This
was
a
bigger
change
and
I
think
I
am
to
the
point
where
it
looks
like.
I
think
it
should
be
good.
I
I
mean
I'm
in
a
good
state
to
really
create
a
pr
for
the
for
this,
and
this.
This
is
something
which
I'm
in
which
I
was
anticipating.
A
A
And
apart
from
that,
others
are
smaller
pieces.
I
don't
think
this.
Would
this
would
need
to
upgrade
the
information
library
matrix
to
scope,
matrix,
add
configuration
options
for
aggregation,
histogram
min
max
support.
We
don't
really
support
that.
As
of
now
aggregation
temporary
controls,
the
way
we
use
it.
I
think
we
need
some
changes
here
and
we
don't
really
support
the
exponential
histogram
aggregation.
So
I
think
that
is
something
also.
We
need
to
add.
A
Yeah,
if
you
can
create
an
issue
at
least
to
mention
that
whatever
somebody
can
pick
it
up
anytime
sure
yeah
and
this
is
still
not
complete.
I
think
probably
I
just
need
to
go
through
the
specs
once
more
and
to
see
any
more
items.
We
need
to
create
and
add
it
here
so
yeah.
Let's
see
I
just
I
mean,
as
of
now
I'm
kind
of
pushing
that
dude
by
the
date
by
every
15
by
a
couple
of
weeks.
A
A
A
A
I'm
not
sure
the
spelling,
probably
parted
me
for
the
spelling.
It's
not
right
example.
A
A
I
think
that's
all.
There
is
all
there's
no
other
change
apart
from
async,
so
yeah,
and
I
think
it's
probably
good
good
to
have
anybody
good
to
have
a
release
end
of
next
week,
irrespective
of
where
we
are
on
the
matrix.
I
think
let's
try
to
push
for
matrix,
but
I
I
still
am
not
very
confident
as
of
now
if
we
should
be
able
to
deliver
matrix
by
end
of
at
least
this
month.
A
Let's
see
probably
I
mean-
probably
it
would
have
more
clarity
by
end
of
next
week
where
we
are
on
matrix
yeah,
but
I
think
it's
good
to
have
one
release
say
end
of
next
next
week
or
end
of
the
month
as
as.
A
A
Okay,
yes
yeah,
we
should
do
that.
Definitely,
and
I'm
just
thinking
for
the
cleanup
also,
I
think
I
know
mark
has
raised
this
issue
last
week
and
I
was
just
thinking.
Should
we
do
a
cleanup.
A
A
We
can
keep
the
implementation
and
we
can
remove
the
implementation
after
the
release
candidate.
A
We
can
probably
announce
it
and
we
can
release
it.
We
can
remove
it
afterwards,
but
I
think
in
the
ci
it's
good
to
remove.
A
Mean
ui
has
done
some
changes
on
in-memory
exporter,
for
here
he
stated
in
memory
exporter
for
backtracks
and
somehow
that
part
of
code
can
be
used
by
in
memory
exporter
or
for
traces
also.
So
he
just
created
that
issue
to
relate
to
reuse
the
changes
which
he
has
done
for
batteries,
also
in
all
sorts
of
other
traces.
A
Yeah
this
one
for
the
the
changes
which
we
did
in
your
tlp
exporter
for
traces
and
logs
the
same
changes
we
should
do
for
matrix.
Also,
we
should
be
using
instrumentation
library
metric.
We
should
be
using
scope
matrix
all
the
I
mean
the
basically.
We
should
be
adding
the
data
and
scope
matrix
and
not
in
instrumentation,
because
this
this
into
instrumentation
library
matrix,
is
removed
in
the
latest
version
of
proto
or
tlp
protocol.
So
I
think
we
want.
A
A
Okay,
this
is
marcus
created,
provide
builders
to
avoid
exposing
sdk
internals
for
locks
and
metrics.
Here.
A
A
A
Somebody
needs
a
custom
processor,
because
in
the
in
current
in
current
processor,
both
in
simple
processor
and
batch
processor,
we
don't
do
we
don't
check
for
filtration
right,
so
we
don't
need
to
do
any
changes
in
the
exporters,
but
if
anybody
wants
to
really
create
a
custom
processor
which
does
the
filtration
and
want
to
use
this
method
for
a
given
exporter,
let
them
let
them
modify
the
exporter
and
provide
this
method
in
the
exporter
solution.
A
D
C
A
That's
more
of
an
internal
change,
it's
just
that
this
will
make
us
x
compliant
in
terms
that
we
do
sub.
We
do
provide
that
the
processors
can
do
filteration
of
span
data
of
spans
based
on
the
current
attributes
or
any
other
any
other
properties
of
this
band.
They
can
do
the
cultivation
so
with
this,
if
we
start
supporting
it,
and
if
somebody
want
to
implement
a
custom
processor
they
can,
they
can
modify
the
exporters
which
which
for
which
they
want
to
do
the
filtering.
So
yeah.
A
D
A
A
A
A
A
So
so,
basically,
in
start
span
we
can
provide
lift
off
links
related
span
links,
but
we
don't.
A
A
A
We
don't
need
a
vector
right,
otherwise
I
mean.
If
we
support
a
single
link,
then
we
don't
need
to
because
we
don't
need.
We
don't
maintain
the
links
internally
anywhere.
We
just
use
a
recordable
interface
and
we
just
do
it.
We
just
serialize
it.
A
A
Yeah,
that's
just
complicated,
I
mean
there's
a
lot
of
overloading
in
that
data
because
we
have
to
support
anyway.
We
have
to
support
multiple
links,
but
now,
if
we
also
add
another
argument
to
pass
only
single
link,
I
think
that
complicates
that
adds
lots
of
overloading
person.
I
mean
that
because
users
may
want
to
add
provides
only
single
link
or
they
want
to
provide
multiple
links.
So
all
those
possibilities
we
have
to
support
in
the
related
functions
so
that
this
complicates
the
design,
and
I
mean
I
just
looked
into
python
and
javascript.
A
I
A
A
A
A
A
So
basically
he's
talking
about
this
coming
in
examples.
We
do
have
http
and
we
have
this
simple
client
and
server
http
example.
This
the
client
http
client,
which
we
use
is
not
asynchronous
client.
It's
just
a
simple
intention
to
be
client,
a
sequential
kind
of
thing,
and
the
ask
is
that
we
should
also
put
an
example
for
an
asynchronous
http
client.
There
could
be
multiple
http
clients
that
we
can
use
so
yeah.
A
Okay,
this
is
just
for
upgrading
the
grp.
We
cannot
upgrade
to
1.46,
because
1.46
supports
one
for
one
proportion.
Support
only
keeps
spotting.
H
A
A
A
Which
is
something
which
we
didn't
wanted
to
have
it,
but
I
think
that's
so
it's
fine.
As
of
now.
I
think
we
can
improve
it
afterwards,
so
I
mean
I
definitely
asked
him
that,
because
the
in-memory
exporter
should
have
any
dependency
on
the
protocol,
as
also
this
pr
adds
the
dependency
on
photovoltaic,
because
this
is
using
the
protocol.
A
Like
for
traces,
we
use
span
data
which
is
more
of
an
independent
entity
or
object
which
we
have
in
the
traces
before
the
expanded
objects.
But
we
don't
have
any
similar
kind
of
entity
information
so
either
he
is
using
the.
A
A
So
I
think
somewhere,
I
think
I've
been
probably
I
can
clarify
it
here,
yeah,
if
you
see
here
he's
using
it's
using
in-memory
metric
data
as
just
so
smart
with
this
one.
Yes,
we're
using
protobuf,
proto,
matrix
resource
matrix
and
storing
that
by
the
in-memory
object.
That
means
that
the
in-memory
exporter
has
a
difference.
A
A
A
A
This
this
needs
to
be
some
ci
failing,
so
this
has
to
be
fixed.
Yes,
you
pick
the
ci
before
you
can
mark
it,
it's
otherwise
it
could
be
approved
by
oh,
it
does
not
approach.
Sorry,
let's,
let's
see
if
it's.
A
Okay,
fine,
I
think
overt
has
a
question
for
some
in
this.