►
From YouTube: 2020-11-04 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).
C
C
B
A
A
Yeah,
I
mean,
I
think
we
couldn't
come
to
ground
on
this
on
the
builders
I
mean,
unless
we
want
to
wait
longer.
C
A
A
In
two
weeks
yeah,
I
think
the
plan
I
think
carlos
was
suggesting
last
week
that
we
go
to
a
two-week
release
cycle,
okay
up
until
ga
just
to
get
things,
so
we
can
adapt
quickly
if
there's
issues
or
changes
that
we
want.
The
people
want.
A
A
Ga
burn
down.
We
added
two
more
issues.
We
finished
two
issues.
We
still
have
four
in
progress.
I
think
it
might
be
the
same
four
as
last
week,
so
those
four
are
kind
of
stalled.
A
A
Yeah
well
without
other
maintainers.
Here
it's
hard
to
necessarily
no
carlos
was
gonna,
do
the
release.
A
We
just
got
to
the
agenda
item
about
doing
a
release
about
the
oteno
release,
since
I
think
you
had
volunteered
to
do
that,
release
yes
and
we
had
chatted
a
little
bit
on
somewhere.
I
don't
remember
where
get
her.
D
A
D
Let's
say
afternoon
tomorrow,
european
time.
D
C
D
A
D
Perfect,
okay,
that
makes
sense.
Yes,
that
works
the
earlier,
the
better
yeah,
okay,
okay
between
12
and
two.
That
would
be.
I
can
do
that.
Probably
earlier,
but
yeah
okay
sounds
good
yeah
yeah!
I
I
will
try.
I
I
have
myself
a
purple
of
pr,
so
I
will
try
to
rebase
them
because
they
are
approved
and
likewise
yeah.
Let's
try
to
to
I
mean
I
don't
want
to
rush
things,
but
if
we
are
sure
about
a
breaking
change
thing,
let's
go
for
it
yeah.
A
We
can
postpone
it
yeah,
so
things
that
were
are
potentially
breaking
that
are
prs
that
are
open
the
baggage.
Well,
this
is
actually
only
it's
not
breaking
if
we
get
it
in
now,
because
baggage
has
been
completely
rewritten
for
o10,
so
that
wouldn't
be
an
actual
breaking
change,
because
the
baggage
api
is
completely
new,
essentially
we've
already
completely
rewritten
baggage.
So
this
wouldn't
be
a
breaking
change.
D
Yeah,
I
can
mix
feelings
about
that,
one
because
I
think
it's
as
I
said
there
is
very
clunkier.
You
know
the
way
it
works
now,
but
at
the
same
time
I
understand
that
bagash
is
supposed
to
be
consumed
mostly
by
the
sdk,
like
the
user
is
allowed
to
to
use
it
and
to
grade
it
you
know
and
to
pass
it
around
so
yeah.
I
have
mixed
feelings
about
that.
Obviously,.
A
Yeah,
I
think
my
last
comment
on
this
was
consistency.
So
we
kind
of
have
this
pattern
that
all
of
the
kind
of
open,
telemetry
iterable
we
put
that
in
quotes
interfaces
work
with
this
for
each
consumer,
and
this
would
just
make
everything
work
the
same
way,
which
I
think
will
be
easier.
Even
if
it's
a
little
bit.
D
Okay,
so
I'm
I'm
personally
fine
with
going
with
this
change
as
as
long
as
we
have
an
issue,
so
we
consider
adding
back
the
entry
set
or
you
know
the
list
of
of
the
of
the
values.
Do
we
we
just
consider
that
before
ga
because
yeah,
I
don't
know
how
often,
as
I
said,
users
have
to
use
baggage
directly.
D
A
A
D
Yeah
yeah
yeah
yeah,
that's
correct:
okay,
let's,
let's
yeah,
okay,
we'll
create
an
issue,
so
we
just
consider
this
just
in
case
in
case
we
need
to
add
back
the
collection
and
if
no,
we
can
forget
about
that
yeah,
but
yeah.
I'm
finding
with
this
change.
Carlos.
D
I
will
yeah,
I
cannot
now
I'm
not
at
home,
but
I
will
create
an
issue
in
half
an
hour
or
something
like
that
great
and
then
yeah
perfect.
A
Okay,
what
else
is
there
here
that
we,
so
this
isn't
breaking
one
way
or
another,
this
one
we
can't
merge,
so
this
is
the
kind
of
the
other
one.
I
think
that
is
significant.
A
So
so,
basically
the
idea
is
to
right
now
we
just
have
a
long
time
stamp,
for
example
on
the
to
for
adding
events,
and
things
like
that.
We
just
have
long
with
no
unit,
and
this
change
would
add
units
to
all
of
those
methods
that
take
long
time
stamps
so
that
people
can
use
whatever
units
they
want.
They
aren't
forced
to
use
nanoseconds
and
we
can
convert
to
nanoseconds
under
the
hood.
A
A
C
A
The
sponsor
himself
is
we'll
have
to
figure
out
how
to
deal
with
android
with
the
animal
sniffer.
A
A
A
If
we
really
didn't
want
to
break
the
api,
we
could
deprecate
the
old,
the
the
more
gentle
change
for
0.11.
We
can
add
methods
that
have
the
the.
A
C
I
think
if
it
I
would
prefer
to
have
instant.
So
if
we
can't
do
instant,
then
let's
postpone
that
until
we
figure
out
can
we
do
instant
or
not
because
to
add
these
new
and
then
add
instant
methods,
and
then
we
will
we'll
want
to
duplicate
those
time
unit
methods.
We
will
make
our
life
more
complicated.
A
Cool
all
right.
A
A
A
Okay,
I'm
I'm
totally
fine
with
with
holding
off
until
we
can
get
instant.
So
the
only
other
really,
I
think
significant
potential
breaking
change
is
well
we've.
This
span
contest
inter
context,
interface,
impul
1,
is
controversial
and
build.
I
think,
moving
the
builders
to
top
levels
are
also
controversial,
so
I
don't
think
we're
going
to
get
those
resolved
by
tomorrow
morning.
A
Yeah,
well,
I
know
that
marching
really
wants
to
get
the
milestone
next
milestone
of
sleuth
out
so
yeah
so
hold
off
on
these
two
and
holding
off
on
making
spam
contacts.
E
A
E
Yes,
well
for
that,
I
guess
I'm
a
bad
person,
because
I'm
not
using
getter
anymore,
no.
A
The
work
around
the
work
around
so
there's
some
there
was
a
complaint
that
somebody's
using
an
inappropriate,
username,
getter,
github,
username
and
emoji
and
icon
or
avatar,
and
it's
being
addressed.
A
A
Yeah
so
ted
in
an
absentia
has
put
a
link
to
this
user
research
dog
fooding
guide
that
he
is
trying
to
push
people
to
do,
and
the
idea
is
that
he
wants
people,
both
people
who
are
familiar
with
projects,
and
maybe
people
who
are
more
newbs
in
the
language
to
run
through
the
language,
implementation
and
instrument
some
sort
of
web
framework
and
put
open,
telemetry
support
into
it
and
respond
to
a
survey
about
how
good
things
are.
A
A
Do
yeah
moving
on
vendor
distribution
experience
nikita.
C
Yeah,
so
I
was
trying
so
in
instrumentation
repo.
We
have
that
task
to
provide
different
hooks
for
vendor
distributions,
so
for
people
who
want
to
repackage
open,
telemetry
instrumentation
a
agent
and
provide
some
custom,
sampler
custom
propagator
whatever,
and
so
I
have
tried
to
do
that,
like
click,
create
an
empty
repo
and
try
to
like
add
all
those
extension
points.
C
C
C
C
C
A
A
Yeah
that
does
read
that
way.
Doesn't
it
I
think
that
the
default
context
propagators
is
probably
badly
named,
because
all
that
is
is
just
an
a
default
implementation
of
the
context.
Propagator's
interface,
I
think.
A
I'm
not
sure
exactly
what
you
mean.
I
mean
right
now
if
they
like.
If
you
did
what
you
wrote
here
on
line
18,
they
would
wipe
out
any
other
propagators
that
had
been
previously
configured
and
they
would
just
end
up
with
the
one
demo
propagator.
C
Or
that's
a
good
question
like,
for
example,
if
I
have
again
my,
for
example,
baggage
propagator,
my
own
specific,
I
shouldn't
replace
with
that
all
default.
I
should
add.
A
D
A
A
Oh
for
propagators,
I
thought
that
you
all
were
doing
that
in
the
agent.
Is
that
not
true.
A
Yeah,
I
don't
I
don't
know,
I
don't,
there's,
definitely
not
one
in
the
sdk,
but
maybe
it's
not
it's
not
a
bad
idea.
One
thing,
one
thing,
that's
funny:
I
just
on
the
line
line,
18,
replace
everything
inside
your
set
global
property
propagators
parameter
so
take
everything
inside
there
and
just
replace
it
with
demo
propagator
colon
colon
new.
A
This
was
something
I
discovered
yeah.
You
can
do
that,
which
is
very
because
this
is
a
single
method.
Interface.
You
can
give
it
a
lambda
as
the
implementation,
which
was
very.
C
A
Yeah,
I
think
that
is
correct.
I
also
think
replacing
the
id
generator
is
going
to
be
a
rare.
I
think
it's
going
to
be
a
rare
occurrence,
maybe.
A
C
A
A
F
F
D
F
A
E
Agent,
the
the
current
bridge
that
I
just
wrote,
actually
just
instruments
the
build
method
to
and
then
overrides
it
in
the
build
method.
E
Here,
oh
my
god,
this
is
from
a
bridge
perspective
of
using
it
externally.
Not
I
mean
like
like
we
would
with
open
tracing
per
se.
E
Just
finding
a
clean
hook
on
setting
your
own
tracer
provider
without
having
some
other
hook,
I
guess
I
don't
know
I
mean
how
would
you
want
so?
E
Let's,
okay,
walk
me
through
this
here
I'll
turn,
my
video
on
so
say
a
customer
is
using
open,
telemetry
right
and
they
have
not
configured
they're,
say
they're
only
using
the
api
and
they
only
call
open
telemetry
dot,
get
tracer
provider
dot
and
then
create
a
span,
and-
and
do
that
thing
you
know
just
by
default-
it's
a
no
opt
right
and
those
those
traces
are
not
going
to
actually
get
reported
anywhere
right,
yeah.
Okay,
so
let's
say
they
add
a
java
agent
to
their
application.
E
C
E
So
is
it
registering
via
the
service
loader.
E
So
same
kind
of
thing
for
for
me
like
so
I'm
not
using
the
sdk,
but
I
do
want
to
hook
in
and
provide
my
own
implementation
for
that
to
work
as
a
bridge
and
the
easiest
way
I
could
think
of
is
by
instrumenting
that
build
method
and
setting
my
own
tracer
provider.
C
Yeah,
but
you
we,
I
believe
we
have
a
little
bit
different
use
cases
or
goals.
So
this
is
I'm
a
vendor
who
mostly
it's
totally
okay
with
open
telemetry,
open
telemetry
java
agent,
but
I
want
I
want
to
add
my
my
my
own
piece,
like
my
own
pro
propagator
on
my
own
id
generator.
E
Right,
so
what
I'm
trying
to
do,
I
would
only
want
to
have
done
if
they
have
open
telemetry
so
like
I
can't
presume
that
somebody
has
open
telemetry
on
their
on
their
class
path.
C
E
E
So
so
that's
one
big
difference
between
what
you're
doing
here
and
my
doing
it
as
for
the
bridge
so
but
is
one
so
you
you
said
you
don't
like
me
hooking
into
the
build
method.
Like
I
just
said.
Yes,
do
you
have
a
better
proposal.
F
Yes,
that's
on
my
screen
right
now,
but
that
presumably
I
have
the
open,
telemetry
apis
on
the
class
path.
D
A
A
C
A
A
Anyway,
something
to
think
about,
I
don't
know
exactly
how
it's
going
to
work
inside
the
agent
initialization
code,
but
now
that
we
have
that
capability.
C
A
Please
yeah
yeah.
I
think
that
would
be
a
really.
That
would
be
a
really
nice.
I
think
that
would
be
really
nice,
because
then
you
can
configure
anything
you
want.
You
can
get
the
whatever
set
on
the
current
one.
You
can
modify
it
and
set
it
on
the
new
one
or
but
not
having
access
to
the
the
sdk
tracer
like
it'd.
Be
nice,
though,
that
sdk
the
tracer
sdk
management,
like
that
comes
out
of
the
sdk
specifically.
A
Yeah,
no,
I
I
hear
you
no
absolutely
I'm
just
thinking,
I'm
mostly
thinking
out
loud
that
we
have
more
capabilities
than
we
did
when
this
was
originally
written.
So
maybe
maybe
it's
time
to
kind
of
rethink
the
way.
This
is
all
configured,
and
I
think
this
your
exercise
here
is
excellent,
because
it
is
showing
that
we've
like,
depending
on
what
you
want
to
configure,
we
kind
of
have
a
different
way
to
configure
it
and
it'd
be
nice
to
have
it
all
in
one
in
one
place,.
A
So
yeah,
it
would
be
great
if
you
I'm
creating
the
one
issue
for
the
propagators,
but
when
you've
kind
of
got
something
here
that
you
think
is
will
work
even
if
it's
ugly,
maybe
we
could
create
an
issue
where
you
show
kind
of
here's
the
ugliness,
and
then
we
can
discuss
how
to
kind
of
make
it
into
a
uniform
api
for
configuring
an
instance.
Okay.
E
So
one
thing
I
I
think
you
might
run
into
is:
let's
assume
that
you
have
so
right
now
on
the
the
java
age
of
the
instrumentation
side,
it
looks
like
the
those
tracer
customizers
are
exposed
to
the
service
loader
yep,
so
that
assumes
that
you
can
have
multiple
service
loaders,
but
the
fact
so
when
you
call
opentelemetry.2builder,
I
think
it's
just
getting
a
new
builder.
It's
not
actually
bringing.
F
E
Any
of
the
previous
stuff,
along
with
it
so
each
well
well
to
build.
C
F
C
E
That's
on
the
default
one.
If
you
look
at
the
sdk
instance,
maybe
I
was
looking
at
the
wrong
place
and
I
don't
fully
understand
it.
E
Okay
anyway,
then
my.
A
C
The
only
problem
here
is
that
the
three
final
here.
A
F
D
A
I
think
there's
some
good
opportunities
in
here
and
I
think
you're
gonna.
I
think
if
we
maybe
work
with
the
sdk
directly,
we
can
make
a
really
nice
fluid
way
to
get
to
do
this.
Do
this
work
I
mean
you
could
even
just
say
two
builder
and
then
you
could
add,
like
I
could
imagine,
having
two
builder
and
have
an
ad
ad
spam
processor
right
and
like
on
the
builder
itself.
A
C
C
A
But
it,
but
it
feels
like
that
builder
right.
There
is
the
place
where
we
can
add
all
the
facilities
to
customers
exactly
how
you
want
to.
A
Yeah
anyway,
I
was
very
amused
by
realizing.
I
could
pass
a
lambda,
a
method,
a
constructor
method,
reference
into
that
method
and
have
it
work.
I
was
super
puzzled
because
idea
was
suggesting
to
me
I'm
like
wait,
a
second
that
doesn't
that's
not
a
global
global
propagator.
What
are
you
talking
about?
A
I
thought
that
was
it
was
just.
I
made
me
scratch
my
head
and
realize
that
you
know
the
when
you
any
single
method,
interface,
lambdas
lamp
go
crazy
with
lambdas,
you
can
do
whatever
you
want,
and
then
you
make
it
looking
at
the
eye
we
ran
into
this.
I
think
when
I
was
at
new
relic
when
we
were
working
on,
we
were
working
on
configuring,
our
like
low-level
telemetry
sdk.
A
That
would
send
data
to
new
relic
and
there
was
something
that
was
added
as
a
single
method,
interface
that
you
could
then
do
crazy
stuff
within,
like
you,
look
at
the
method
call,
and
it
made
absolutely
no
sense,
but
if
you
track
through
all
the
method,
references
and
stuff-
and
you
know
it
did
what
it
wanted
to
do,
but
the
method
made
no
sense
the
way
it
ended
up.
Looking
so
it's
funny
all
right
anything
else
before
we
sign
off.