►
From YouTube: 2023-03-22 meeting
Description
cncf-opentelemetry meeting-2's Personal Meeting Room
A
D
E
B
You
have,
if
you
have
any
things
that
you
want
to
discuss
today.
Please
put
them
on
the
agenda.
B
Okay,
I
guess
we
can
get
started.
There's
the
first
item
on
the
agenda,
I'm,
not
sure
who
added
it.
Storing
and
caching
signals.
F
Yeah,
that
was
me:
okay.
No,
it's
it's
mostly
a
question
because
I'm
kind
of
recently
getting
involved
in
in
hotel,
so
I.
F
So
I'm
working
in
the
for
the
mobile
side
of
the
Android
agents
and
and
at
some
point
in
the
mobile,
we're
going
to
need
to
Cache
the
signals,
the
all
of
the
the
traces
and
logs
and
everything
to
store
them
locally
in
the
devices
and
in
order
to
send
them
later.
I
was
wondering
if
there's
something
that
has
been
discussed
about
it
before
or
if
it's.
If
it's
something
that.
A
B
Do
you
mean
like
in
case
when
the
mobile
device
doesn't
have
access
to
the
to
the
network,
and
you
have
to
wait
until
it
wakes.
F
F
There's
also
there's
also
the
issue
with
slow
internet
access,
especially
when,
when
there's
a
bad
reception
or
it's
kind
of
stuff
that
that
you
know
my
my
make
those
those
export
things
to
fail,
and
so
it
could
also
work
as
a
retry
mechanism,
but
like
for
later,
like
so
yeah.
B
F
B
Was
gonna
say
like
I?
Guess
we
I
guess
it
would
be
something
like
exporter
that
is
aware
of
the
environment
and
can
can
retry
when
based
on
certain
events
that
happen
but
yeah,
it's
not
it's
to
be
defined.
Yes,.
F
F
Think
what
could
be
a
good
starting
point
is
to
find
a
way
to
into
Marshall
the
the
signals
so
that
they
can
be
stored.
I
I
saw
that
there's
a
spec
and
that
transform
to
transform
all
of
the
signals
into
Json
line
and
I
I
I.
Don't
know
that
could
be
a
good
starting
point,
but
I
just
wanted
to
make
sure
if
there
was
something
else
already
worked
ahead,
that
I
could
just
reuse.
B
So
I
know
there
is
a
work
in
progress
by
Splunk
to
donate
their
Android
instrumentation
and.
C
C
For
mobile
ones,
I'm
not
in
the
mobile
game,
but
I
know
that
they
do
have
some
sort
of
storing
on
discontental
connections
over
here
available.
F
B
Yeah
I
think
it
would
be
good
too
there's
a
engineer
that
or
person
who
from
Splunk,
who
sometimes
joins
and
I
think
he
works
on
the
Android.
Instrumentation
is
Jason,
Jason
plum,
so
maybe
like
maybe
I
can
reach
out
or
you
can
reach
out
to
him,
and
you
know
discuss
that
with
him.
E
Yeah
I
would
think
it
would
be
the
exporter's
role
to
do
this.
I
know
for
node
I'm,
pretty
sure
we
have
built
a
file,
persistence
layer
there,
but
I
I,
don't
know
what's
already
in
the
batch
Explorer
or
anything
else
like
that.
F
Yeah
but
I
guess
that's
as
Martin
mentioned,
these
exporter
will
have
to
know
about
the
the
environment
that
it
seemed
right.
So
correct,
yeah,
yeah.
E
F
E
They're
talking
about
automatic
retries,
it
came
up
in
a
spec
or
maintainers
meeting
a
couple
months
ago.
I
think
it's
more
for
the
server
back
end,
so
it
doesn't
cause
duplicate,
Telemetry,
okay,.
D
E
I
I
I
never
drilled
into
the
the
details
around
it.
If
you
go
searching
the
spec
repo
for
retries,
you
might
be
able
to
find
it.
F
I
I
saw
a
link,
it's
just
sharing
in
the
chat,
okay
cool.
Thank
you.
Take
a
look
at
it
as
well.
Yeah.
B
Okay,
this
is
the
second
item
on
the
agent
does,
is
I
put
it
there?
It
was
a
question
that
I
had
on
the
you
know.
So
Santosh
has
the
has
defined
the
browser
events
semantic
conventions
and
one
of
them
is,
is
HTTP
request.
Timing
and
I
was
just
wondering
how
why
do
we
need
that?
If,
if
that's
also
captured
as
part
of
resource
timing
as
I
won't
remember,
because
I
don't
actually
see
it
in
either
of
the
docks,
as
listed
as
a
separate
type
of
event,
it's
just
in
that
again.
E
I,
you
know
I
I
think
we
can
do
them
as
the
same
thing
and
if
there's
a
type
field
that
we've
got
on
the
request,
so
the
resource
timing.
That
would
probably
suffice,
because
yeah
they're
the
same
set
of
fields.
B
B
Okay,
well,
that
kind
of
leads
leads
to
the
next
topic
that
I
that
I
added
I.
So
yesterday,
I
created
I
created
issues
in
the
sandbox
or
all
the
different
instrumentations
implementing
the
different
instrumentations,
based
on
the
events
defined
in
the
semantic
conventions,
and
so
as
I
was
going
through
them.
B
D
B
Yeah
so
I
think
just
just
like
the
issue
assign
it
to
yourself
and
their
I
think
I.
Think
the
math
correct
me
if
I'm
wrong,
but
the
the
workflow
would
be.
You
know
fork
for
this
repo
from
the
from
the
auto
and
base
and
then
based
on
the
other
instrumentation
POC
Branch,
create
your
own
branch
and
then
create
a
create
a
PR
to
you
know
to
to
merge
into
the
auto
instrumentation
POC
Branch.
Here,
okay,.
D
E
That's
actually
part
of
my
update
so
effectively.
There's
the
last
item
in
the
list
is
a
PR
which
is
still
building
because
I
only
push
it
in
just
before
the
meeting,
which
does
exactly
that.
E
So
yesterday,
I
actually
found
some
time
to
work
on
it
and
found
the
the
reason
the
build
was
you
know,
although
the
merge
was
breaking
is
git
was
being
a
bit
too
clever.
So
when
I
did
a
git
merge
from
a
change,
it
ended
up
duplicating
the
relative
path
for
for
an
eslint,
so
I
had
to
drag
over
some
code
that
I
had
from
the
merging
in
the
repo
that
once
it
actually
does.
E
The
git
merger
then
compares
the
source
location
that
I
merged
from
into
the
branch
that
I'm
merging
into
and
effectively
just
overwrites
files,
if
they're
not
the
same
because
of
merge
problems,
so
that
seems
to
have
fixed
it.
So
nice
test
PR
passed
fine.
E
E
B
And
once
once
the
events
API
is
in
in
the
main
branch,
then
the
instrumentations
only
have
dependency
on
the
events.
Api.
That's
all
yeah.
E
I
had
a
couple
of
comments
for
your
events,
SDK,
which
I
thought,
which
probably
good
to
talk
about
here,
a
couple
of
things
so
I
realized
you
couldn't
do
some
of
this,
because
it's
the
API
wasn't
there
yet
the
first
one
is
the
package
name.
We
probably
want
to
like
prefix
it
with
sandbox
to
be
consistent
with
the
other
ones.
Okay,
yeah,
that's
just
for
consistency.
Now
you
can
call
it
something.
Also,
if
you
like
to
the
other
one,
is
for
this
to
actually
build.
E
You
need
to
add
it
to
the
rush
Json
in
the
root.
So,
if
you
go,
if
you
go
look
at
Maine,
we
can
show
open
us
the
same
thing.
If
it's
not
there,
Rush
doesn't
know
about
it
and
doesn't
drag
it
in
so
in
the
rash
station.
This
is
the
the
file
that
affected
every
project
that
the
sandbox
builds
has
to
be
listed
here.
So
you
just
added
no
particular
order.
E
This
should
publishes
the
flag
in
terms
of
how
Rush
detects
stuff
it
resolves
all
conflicts.
It
looks
at
all
package
jsons
that
are
listed
and
you
know
figures
out
what
what
the
build
order
should
be,
but
yeah
the
package
name
should
be
the
package
name
that
you've
got
in
your
package
Json
and
then
the
other
critical
bit
is
the
path.
So,
okay,
it's
as
easy
as
that,
and
then
your
npm
run.
Compile
from
the
root
should
then
use
it.
E
E
So
in
fact,
I've
got
Auto
limiting
turn
turn
on
as
part
of
compiling.
So
you
will,
when
you
do
a
build
for
the
first
time,
because
I
don't
run
lint
as
part
of
the
merge
it,
it
will
modify
files
automatically
on
you
and
you
go,
but
I
didn't
change
that
it's
because
effectively
did
a
lint
fix.
It's
part
of
building
I.
B
E
E
A
E
B
E
Okay;
okay,
if
you
comment
on
one
of
those
that
should
add
you
to
the
thing,
so,
if
you
just
say
hello
or
Please,
assign
to
me
or
something
like
that,.
D
B
Okay
and
I've,
been
at
I,
can
do
the
same
for
you,
I
think
you're
working
on
the
HP
one
right.
B
Okay,
I
guess:
that's
all
the
things
on
the
agenda.
Does
anyone
else?
Does
anyone
has
any
other
things
to
talk
about.
E
This
is
the
pr
to
bring
in
the
API,
so
one
of
the
advantage
of
having
open,
Telemetry
build
that
is
I,
can
sign
it
off
and
then
push
it
straight
in
so
yeah.
A
E
Yep,
cool,
okay,
so
yeah
we're
just
waiting
on
the
Windows
tests
to
run,
which,
for
some
reason,
take
longer
I,
don't
know
why?
Maybe
they
don't
have
in
git,
they
don't
have
or
GitHub
they
don't
have
as
much
beefy
hardware
for
their
Windows
environments
or
something
okay.
It's
been
running
for
like
a
minute.
Almost
12
minutes,
I
think
it's
the
same
for
the
JS
repo
as
well.
E
E
One
future
bit
of
work
that
I
might
have
to
do
is
the
auto
merging.
If
you
look
at
the
history
of
the
the
branch
looks
awful,
at
least
in
my
local
one,
and
that's
because
in
fact
it
creates
two
local
commits
before
it
does
the
merge
to
create
the
pr
every
time
it
runs,
plus
the
main
emerging
from
the
Json
Country
Water
also
has
a
commit,
so
it
becomes
a
bit
of
a
spider
web.
So
hopefully
it
doesn't
cause
you
too
much
grief
Martin,
bringing
that
in
so.
E
Yeah
effectively,
every
branch
that
we
create
in
this
repo
will
get
the
automatic
open,
Telemetry
stuff,
because
it's
the
the
GitHub
restrictions
are
defined
as
star
star.
So
yeah
you
will
need
a
prta.
Do.
B
A
git
merge,
okay,
so
yes,
once
I'll
do
that
as
soon
as
the
this
prism
I'll
do
that.
E
Yeah
at
some
point,
I'll,
probably
we'll
probably
have
to
we-
can
talk
to
Daniel
about
it
as
well
request
access
to
figure
out
how
we
give
people
ownership
of
individual
branches,
because
that
would
be
the
preferred
method.
So
we
haven't
got
to
worry
about
managing
each
branch
ourselves.
B
D
B
B
Great
thanks,
everyone.