►
From YouTube: 2021-10-08 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
No
with
good
with
good
reason,
the
whole
url
becomes.
The
url
encoder
is
the
worst
thing
in
the
world
yep
it
built
in
the
built-in
one.
In
java,
it
decodes
just
fine,
but
its
encoding
is
super
dumb,
like
using
the
plus,
rather
than
using
percent.
C
C
C
D
Like
it
didn't
trigger
me
to
also
send
a
pr
to
the
baggage
thing,
much
less
important
one
well,
I
haven't
seen
yet
like
even
their
examples
aren't
consistent
encoding,
so
we
can
feel
okay
about
not
doing
it
correctly.
Maybe.
A
A
A
Oh
to
the
oh.
E
C
A
I
wanted
to
chat
if
y'all
don't
mind,
bueno
john,
let's
see
do
you
want
to
do,
we
can
do
sd.
The
reactor
stuff
might
take
some
time
so
john,
do
you
want
to
go
over
sdk
stuff?
First.
C
Well,
I
think,
there's
only
the
rule
there's
just
released
to
talk
about
and
I
figured
out
the
way
to
get
my
pr
merge
just
to
make
some
people
who
had
already
approved
it
into
approvers,
and
then
I
can
merge
it
did
that
work
yeah
totally
yeah
yeah
did
you
care
was
that
oh,
it
was
the
updating,
the
symantec
conventions,
which
I
think
for
some
reason
dropped
off
your
radar.
C
C
C
C
C
Well,
so
it
I
mean
it's
just
called,
I
think
it's
just
called
message
id
and
I
think
he
had
suggested
that
they
change
it
to
rpc
message
id,
but
it
didn't
get
or
sorry
because
yeah
rpc
message
or
something
like
that,
but
that
didn't
get
adopted
or
merged.
So
it's
just
message
id.
It
happens
to
be
in
the
rpc
group,
but
that
isn't
ever
a
part
of
any
of
our
semantic
convention.
Yeah.
D
C
D
Yeah
classic
and
I've
been
chatting
a
bit
about
that
how,
as
long
as
our
module
is
alpha,
it's
still
okay,
but
eventually
it's
not
going
to
be
offline.
There
might
still
be
breaking
changes,
so
I
was
wondering
if
we're
supposed
to
have
the
schema
version
or
something
like
that
in
the
package
name,
so
that
we
would
be
able
to
add
new
packages
with
different
version
of
the
spec
in
the
future.
When
we
do
care
about
stable.
D
A
C
Strategy
yeah,
so
I
think
the
the
question,
if
I
want
to
rephrase,
is,
if
somebody's
using
those
semantic
conventions
in
their
instrumentation
and
they
change,
then
we'll
break
their
code.
If
we
get
rid
of
the
old
version,
exactly
yeah
yeah,
I
have
no
idea
what
I
mean
the
yeah,
the
the
schema
will
supposedly
let
you
map
between
versions
when
it
doesn't
talk
about
people
using
backwards,
incompatible
like
from
a
code
perspective.
C
And
that's
a
much
hairier
thing
because
you
might
end
up
with
a
merged
resource.
That's
made
up
of
pieces
from
multiple
schemas,
whereas
you
wouldn't
have
that.
I
mean
I
guess
you
could
have
that
like
there
could
be
two
instrumentation
libraries
that
are
modifying
the
same
span
with
two
different
semantic
attributes
from
two
different
schemas
right.
It
could
happen
right.
D
A
I
don't
know
yeah
and
if
we
we're
like
the,
if
we
just
create
a
new
package
for
every
version
with
everything
in
there,
that
becomes
amazing.
D
That's
definitely
like
most
of
the
versions
don't
have
changes,
so
I
wish
the
semantic
dimensions
were
versioned
in
a
way
that,
like
I
guess,
I
wish
the
reason
simver
and
then,
if
the
major
version
of
the
semantic
conventions
got
updated
on
a
breaking
change,
then
we
could
always
just
add
a
new
package
for
those.
Well,
not
the
minor
versions.
D
A
E
D
C
C
B
D
F
C
D
C
And
I'm
still
still
one
that
one
still
feels
a
little
weird
to
me
for
reasons
I
commented
just
yeah
we're
adding.
I
mean
we're
adding
some
a
weird
thing.
We
have
to
maintain
it's
kind
of
outside
of
the
scope
of
the
normal
stuff.
We
maintain.
D
I
don't
think
we
have
to
rush
this
in
at
the
last
minute
yeah.
I
think
I
think,
waiting
for
nash,
please
just
sort
it
out,
because
I
just
sort
of
just
randomly
did
it.
I
don't
know
because
I
was
like
oh
man
like
I
like
I've
had
in
my
mind,
like
what
can
we
do
about
this
dependency
for
a
while
and
then
just
yesterday,
I
happened
to
just
look
at
it
and
I
was
like
yeah.
We
even
have
it.
It's
not
only
an
implementation
details
in
our
public
api.
This
is
pretty
scary.
D
C
A
Oh,
I
wanted
to
ask
john
about
this
decorating
span.
Processors.
Did
you
mean
like
decorating
like
the
putting
a
decorator
in
front
of
the
batch
span?
Processor.
C
Yeah
so,
for
example,
I
mean
this
doesn't
apply
to
auto
configure,
but
in
the
splunk
android
stuff
we
decorate
a
dashband
processor
and
use
it
to
add.
Well,
I
guess
we
don't
decorate.
In
that
case,
we
just
have
our
own
spam
processor,
that's
running
in
front
of
running
before
the
other
one.
C
It
only
does
on
start,
so
it
doesn't
really
matter
when
it
runs.
I
don't
know
we
just
got.
We've
gotten
requests
for
people
who
want
to
do
like
take
these
as
an
expand
process
and
do
decoration
just
because
the
configuration
right
now
it
implicitly
orders
them.
But
it's
I
mean
there's
not.
I
don't
think
there's
any
necessarily
necessary
guarantee
that
that
order
is
going
to
be.
D
D
I
mean
it
would
be
to
the
existing
sdk
tracer
provider
builder.
Then
the
configure
is
just
providing
that
builder,
which
maybe
we
like
in
adding
a
decorator
knob
to
our
builder,
could
make
it
easier
to
use
in
some
cases
also,
even
though
it's
not
that
necessary
like,
for
example,
our
maria
has
this
pattern,
where
you
can
apply
a
decorator
to
the
whole
server
or
to
particular
endpoints.
So
it
sort
of
feels
like
that.
So
it's
not
the
weirdest
thing.
I.
D
A
Just
set
it
directly
in
here
as
a.
D
C
D
C
Well,
I
know
we
can
do
default,
we
can
add
default
methods,
default,
implementations,
yeah
and
yeah.
No
one's
going
to
be
putting
them
in
a
lambda.
I
don't
think,
but
even
so
even
they
will
still
a
single
method
interface.
They
still
work,
even
if
they
have
a
default.
D
But
I
like
it
has
a
name.
What
do
you
think
the
usability
like?
Is
it
then
going
to
use
the
parentheses
pad
or
something
like
that
or.
C
E
A
E
A
From
so
many
spi's,
but
that
ship
has
probably
sailed
so
I'm
good
with
the
consistency
seems
like
it
would
be
spi
like,
because
the
method
in
here
by
itself
would
be
kind
of
then
add.
D
D
So
now
that
we
have
two
data
points
as
I
have
to
finally
take
that
out
of
the
incubator,
we
need
some
condition
to
take
care
of
the
incubator.
At
some
point
I
think
yeah
I'm
not
opposed
to
it
and
just
forget,
like
it
looks
like
if
two
people
prefer
the
delegator
to
the
builder,
then
it
sounds
like
that's
enough
to
decide
that
the
delegator
is
probably
the
better
pattern
here.
So.
C
C
D
A
Yeah,
it's
a
nice
pairing,
though,
with
the.
If
we
are
adding
that
new
spi.
D
C
C
Don't
think
I
had
anything
else,
fluorescing,
just
kind
of
improver
maintenance
needed
I
want
to.
I
was
disappointed
to
find
out
that
there
were
no
policies
at
all
on
how
we
should
decide
when
people
should
be
removed
from
the
approver
list,
because
we
have
a
very
large
list
of
approvers,
a
half
of
which
never
literally
look
at
a
single
pr
plus
one
for
cleanup.
I
mean
the
flip
side
of
that
is
like
it
doesn't
hurt
to
have
them
there
and
if
we
trust
them,
if
they
came
back,
I
don't
know,
but.
E
A
C
D
E
D
C
C
Yeah,
I
think
I
would
for
people
who
haven't
been
active,
arming
pavel
powell,
just
that
whose
name
I
forget
off
the
top
of
my
head
and
tyler
yeah
and
tyler
really
haven't
done
anything
in
a
long
time.
So
I
would
probably
remove
those
four.
C
A
C
D
A
C
D
C
C
I
was
thinking
that
we
that
open
telemetry
had
first
class
support
for
customizing
tracers
to
add,
like
have
a
custom
sampler
per
tracer,
and
that
that
support
was
basically
we
would
want
to
have
lots
like
potentially
a
separate
tracer
provider
per
library.
A
Sounds
a
little
bit
like
josh
mcdonald's
thoughts
about
sampler,
about
sort
of
the
optimized
samplers
that
I
think
could
target
tracers
or
something
that
could
do
early
efficient,
yeah
suppression.
C
A
C
C
A
All
right,
so,
let's
move
to
the
reactor.
A
I
don't
know
honorable
if
you
had
a
chance
to
look,
but
I
wanted
to
get
your
thoughts
on
primarily
around
the
the
public
api,
what
it
looks
like
and
if
there's
any
other
tests
that
you
think
that
should
be
in
this
first
pr.
A
E
E
Yeah,
just
one
thing:
I've
added
the
one
thing
we
discussed
is
just
putting
context
explicitly,
but
there
are
some
scenarios
with
wisp
and
mono
and
oil
flux
when
it
doesn't
really
work
already,
so
I
kind
of
fixed
them
and
the
problem
that
we
have
we
had
in
the
past,
like
let's
say:
if
we
defer
something
and
then
we
have
we
spend,
then
we
create
a
scope
around
it.
We
may
expand
current
for
this.
E
D
E
F
E
E
D
A
Cool
and
one
thought
that
we
were
not
unsure
about
is
the
splitting
out
of
tracing
operator
and
sort
of
like.
What's
this
entry
point
and
there's
it's
a
little
weird,
because
there's
some
of
it's
static,
it's
a
little
bit
of
a
mix
of
static
and.
A
A
So
we
were
sort
in
some
of
the
other
instrumentations
like
we're,
creating
the
entry
point
as
like
grpc
tracing
or
spring
tracing,
but
yeah
everything
is
and
then
the
grpc
tracing
builder
to
build
that.
But
everything
is
instance
feel
everything
is
instances,
and
this
I
I
mean
I
don't
know.
If
there's
really
anything
we
can
do
about
it.
E
D
D
E
Yeah
I
I
also
saw
that,
like
for
the
anyone
who
uses
this
methods
like
the
convenience,
that
we
will
create,
it's
more
like
that,
the
the
span
builder
right
or
the
the
tracer,
so
it
would
be
very
hard
for
someone
to
create
an
instance
and
then
pass
it
around
or
maybe
they
will
create
it
on
each
tab
when
they
need
to
put
the
context
in
but
having
some
instance
registered,
it's
stateless.
It
doesn't
do
anything.
It
just
creates
a
new
span
every
time
right.
E
E
D
Then
we'd
sort
of
expect
it
to
follow
our
normal
pattern,
where
we
have
like
a
create
method
that
accepts
an
open
domain,
for
instance
right
yeah,
so
for
now
we're
not
doing
that.
So
we
don't
need
it
right
now
and
since
the
api
is
not
stable,
I'm
sure
we
can
make
tweaks
later
if
we
need
to,
but
like
I
sort
of
do,
expect
us
to
expand
functionality.
Also
at
some
point
in
which
case
would
probably
follow
our
standard
model
better
address.
Does
that
make
sense.
A
So
I
run
in
so
like
I
see.
E
A
There
was
another
yeah,
the
core
test.
Yes
right,
right,
yeah,
okay,
so
having
a
convenience
mechanism
to
do
this
monster
yeah.
A
E
This
monster,
but
this
called
it
the
transform
and
the
trace
mono
not
blocking.
Where
is
it
from.
A
A
D
A
E
I'm
not
sure
how
does
it
work
so
this
is
reactors
3.1
and
I
don't
have
a
half
of
apis
that
I
see
in
reactors
3.4
and
then,
but
when
I
use
it
in
my
reactors,
3.4
application,
it
works
fine.
So
I
hope
it's.
I
guess
it's
just
luck,
but
on
their
knees
the
context
is
the
same,
but
the
apis
we
have
are
different
in
application
and
in
shaded
instrumentation
shaded
reactor.
A
A
E
Yeah
they
they
edit,
react.
Sorry,
the
the
context
field
on
the
signal
or
the
method
to
get
context.
A
A
A
Yeah
on
the
the
tracing
operator,
I
mean
this
still
feel
the,
but
it's
not
really
related
to
this
pr,
so
we
can
revisit
the
the
instance
versus
staticness
of
it.
A
A
A
So
but
at
the
same
time,
I
think
you
know
as
long
as
we
have
the
check
there
and
we
can,
you
know
we
can
bring
it
back.
We
can
always
turn
off
this
retry,
this
retry
in
the
future.
If
we
want
to
tackle
that
again-
but
I
was
I
was
okay
with
this,
but
just
wanted
to
run
that
by
you.
B
A
Metrics
exemplars,
so
just
talking
about
oh,
we
didn't
talk
about
this
with
john.
The
question
was:
if,
if
you
all
want
to
disable
exemplars
in
auto
configure
for
the
for
one,
seven
is
still
a
disabled
matrix.
A
D
A
Okay,
that's
good
because
I
mean
it
sounds
cool.
I
would
like
to
I'd
like
to
get
people
using
it,
so
I
was
worried
if
we
disable
it
nobody's
gonna
touch
it.
A
I
asked
if
so
this
is.
This
is
asked
about
kind
of
the
view
state.
This
is
still
needed.
This
horribleness
is
still
needed.
E
A
Josh
was
just
kind
of
talking
about
what
what's
required
to
deal
with
that
cardinality
issue.
A
The
view
api
is
coming,
but
in
this
gonna
be
in
this
release.
Oh.
E
D
D
A
D
D
Okay,
yeah,
that
is
weird,
got
it
yeah,
and
that
is
a
bit
weird
for
us
yeah.
A
D
E
D
B
A
A
A
Uh-Huh
and
have
having
a
a
different
appender
registered
for
the
things
that
we.
E
A
D
A
F
A
Would
need,
I
know,
I
think
we
would
need
auto,
configure
module
to
not
be
shaded,
which
should
be
fine,
because
the
auto
configure
module
needs
to
link
directly
to
the
exporters,
like
it
references
the
otlp
exporters
by
name,
so
that
needs
to
be
in
the
agent
class
later,
but
that
should
be
okay
and
just
the
raw
sdk
in
the
bootstrap
maybe
might
work
hello.
I'm
gonna
put
about
a
65
chance
of
that.
One.
D
D
E
A
D
A
It's
the
same
stuff
that
I
I
put
it
in
the
upstream
originally
and
it
generates
spans
yeah
alone
yeah.
But
then,
but
our
exporter
converts
the
spans
back
and
sends
it
to
our
via
logging.
Our
logging
ingestion.
D
A
A
A
A
D
A
Well,
not
not
the
not
the
hairy
bridge,
but
the
just
the
bootstrap
bridge.
Yes,.
A
Right
so
we
talked
about
that.
Jack
was
thinking
of
starting
to
work
on
some
instrumentation
vlogs.
A
Just
ouch
is
looking
for
feedback.
B
A
A
google
sheets
expert
here:
how
do
I
rap.
A
Ben
talked
about
the
jfr
instrumentation
is.
A
A
A
E
A
There's
an
endless:
this
is
just
the
endless
project
like
there's
no
end
to.
I.
A
Like
all
good
big
projects
are
but
yeah
there's
never
ending
things
that
could
be
interesting,
so
yeah.
He
had
kind
of
a
common
confusion
of
what
do
I
put
for
schema,
url,
instrumentation
name
and
instrumentation.
E
A
A
A
It
not
in
the
I
added
this
link,
because
I
wanted
ben
just
to
look
at
this
because
he
said
he
was
just
starting
to
or
he
was
meeting
with
somebody
next
week
to
talk
about.
They
were
going
to
start
working
on
micrometer,
open,
telemetry
stuff,
and
I
know
that
that
there
have
been
various
iterations
of
that
discussion
over
time.
A
D
D
A
Okay,
thank
you.
Thank
you
for
there's.
A
Oh
no,
okay,
so
yes,
people
are
interested
in
this.
I
don't
have
anybody
from
splunk
here,
but
they
have.
F
A
A
A
So
they
do
have
some.
So
why
didn't
I
find
this
when
I
looked
so
this
was
what
I
thought
was
interesting
that
looked
like
they
were
bridging.
A
Yeah
yeah,
okay,
it
is
there.
A
A
I
was
guessing
they
were
bridging
to
oh
well.
I
guess
I
didn't
really
look.
I
was
hoping
they
were
bridging
to
hotel
doing
what
we
want,
which
is
now.
D
A
D
D
F
D
A
F
A
A
Kind
of
we
do
this
right
when
we
build
the
examples,
would
we
also
try
to.
A
D
A
We
have
to
publish
we're
publishing
the
big
everything.
A
A
The
version:
how
do
these
even
use
the
170
snapshot
version.
A
A
Then
I
don't
stand
a
chance
on
this.
I
will
wait
for
your
your
feedback,
I'll
I'll,
give
it
another
try
tomorrow.
A
All
right,
mccarthy,
we
can
review
yeah
yeah.
A
What
was
what
was
it,
the
splunk
boss
joined.
A
That's
they're,
like
john's
nikita's
manager,.
A
I
think
he's
doing
the
he
was.
He
came
to
the
the
tuesday
semantic
http
semantic
attribute
meeting.
I
think
he's
doing
doing
the
politician
meeting
meet
and
greet
for.
F
A
Cool
anything,
do
we
have
anything?
Oh
I
saw
you
did
some
improvement
to
the
the
the
java
test
messages
looking
forward
to
seeing
it,
because
I've
had
some
trouble
once
I'll
wait
to
upgrade
to
one
seven.
A
D
A
E
A
Yes,
yes,
thank
you
yeah.
I
was
like
it
just
happened
this
week
and
I
was
like
there's
you're,
not
telling
me
what
the
error
is.
Yeah
yeah.
F
F
A
This
was
reported
by
a
customer
yeah,
so
we
definitely
have
customers,
and
I
remember
I
had
a
couple
customers
who
are
on
really
early
versions.
When
I'm
surprised
this
version,
I'm
surprised
this
was
even
working
for
them.
Oh
no,
they
must
have
avoided
somehow
we
had
a
snafu
with
our
ssl
cert
at
one
point
being
signed
by
something
that
was
not
in
like
the
default.
Eight
zero
really
early
releases,
also
so
yeah
yeah.
A
So
I
I'm
fine,
I'm
I
may
just
detect
and
bail
out
or
something
just
so
that,
because
this
is
a
pretty
scary
thing
for
a
user
to
see.
D
A
Because
I
did
I
fixed
this
one
locally
because
it
was
a
method
reference,
and
so
I
just
changed
it
to
a
an
anonymous
inner
class
and
yeah.
And
then
I
hit
another
one
and
I
was
like
no
forget
it.
D
Is
this
related
to
like
compiler,
it's
related
to
lambdas?
Obviously,
maybe
the
runtime
code
generation
like
if
you're
running
the
compiler,
it
would
do
it
at
compile
time,
but
since
we're
an
agent,
it
ends
up
getting
caught
at
runtime
instead
or
something
like.
I
think
I
am
sort
of
familiar
with
this
bug
like
I,
like.
I
I've
seen
many
lambda
compile
failures
on
java
8
before
40,
or
something
like
that.
A
D
E
A
A
All
right,
we'll
have
a
good
one.
Good
luck
with
the
sdk
release
tomorrow,
I'm
sure
it
will
go
better
than
ours.