►
From YouTube: 2021-06-09 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).
A
C
C
E
Hello
folks,
I'm
glad
you
pinged
me
on
slack,
I
don't
know.
I
was
just
typing.
B
E
Well,
I
I
live
not
17
miles
from
portland,
but
maybe
like
10
miles
from
downtown
portland,
but
it's
like
in
the
woods
and
we
have
two
goats
and
a
llama.
So
it's.
E
It
is
like
it's
the
woods:
it's
like
a
six
acre
lake.
The
back
part,
is
in
a
swamp.
E
But
we
are
very
close
to
the
suburbs
so
like
in
one
direction,
we
can
go
to
downtown
portland
and
in
sort
of
90
degree
other
direction.
We
can
go
into
the
suburbs.
The
suburbs
are
actually
closer
to
us
than
downtown
portland,
but
technically,
like
our,
like,
our
kids
are
in
the
downtown
portland
school
district.
So,
like
it's
kind
of
this
weird
little
carving
that
goes
anyway,
you'll
have
to
come
to
portland.
So
I
can
explain.
C
E
Oh,
I
know,
but
it's
it's
very
funny
in
portland
like,
like
everybody,
hates
the
blackberries,
because
they
like
overgrow
everywhere
they
are
like
weeds,
but,
like
you
know,
for
a
month
out
of
every
year,
it's
like
you
just
see
people
on
the
side
of
the
road,
picking
and
eating
black.
E
I
put
this
on
just
to
it.
Looked
like
there
was
some
interesting
span,
modeling
discussion,
but
I
didn't
get
a
chance
to
read
anything
today.
E
B
Is
yeah
point
being
so?
Whenever
you
read
the
message,
when
you
consume
the
message,
you
always
have
a
choice.
What
a
spam
for
for
message
processing
will
be.
Will
it
be
like
local
current
span
current
context,
or
will
that
be
the
context
read
from
the
message,
raw
propagated
message,
right
and
so
material?
In
this
pull
request
for
sprinting
integration
chose
to
have
like
local
parent
con
local
context
to
be
apparent
context
a
in
the
majority
of
like
real
messaging
systems
like
kafka
and
rabbit
integrations
that
we
did.
B
E
So
is
this
an
example
where
we
know
like
we,
because
sometimes
we
don't,
we
can't
correlate
it
to
the
processing
right.
We
just
have
a
case.
Somebody
called
receive
and
pulled
a
message,
but
it's
not.
E
B
B
B
E
E
So
that
we
can,
because
in
that,
even
in
that
case,
I
think
we
still
parent
to
the
implicit
context,
because
it's
not
it
is
part
of
a
bigger
transaction,
as
opposed
to
like
a
message:
queue
system,
a
producer
side,
that's
getting
pushed
things
and
we.
B
B
But
my
my
point
in
general
that
I
think
that.
B
B
B
B
C
B
E
B
Correctly,
that's
correct
in
spring
integration
you
more
or
less
subscribe
for
subscribe
your
listener
for
for
incurrent
messages,
which
means
that,
yes,
you
you
have
no
like
business,
so
application
specific
local
context
that
you
could
use
as
a
parent,
you're,
always
called
by
by
framework.
E
Yeah
yeah,
so
that
to
me
is
kind
of
a
difference
like
if
it's
subscribing
to
something
and
you
get
it.
It
seems
to
me
that
the
the
consumer
span
is
both
the
parent
of
all
the
work
and
it
should
parent
the
propagated
span,
whereas
if
it's
a
polling,
if
it's
a
pulling
one,
then
you're
writing
application
code
to
do
the
polling.
So
you
have
an
implicit
parent,
and
that
makes
sense
to
parent
to
that.
B
Any
subscription
is
still
implemented
by
polling
at
some
point.
Yes,
but
that's,
but
but
it
understands
that
that
framework
billion
can
you
you
don't
care
about.
C
Okay,
it's
more
just
like
message:
processing,
it's
not
our
pc,
I
think
so.
I
would
expect
links
for
kafka.
C
E
E
Where
are
you
yeah,
where
you
have
a
listener
right?
You
look
registered.
B
If
you
have
a
listener,
then
for
me
it's
clear
cut.
If
all
that
you
expose
just
listen
to
get
message
and
you
are
not
doing
polling
or
whatever
yourself,
then
you
should
use
propagated
context.
That's
it
in
case
of
of
kafka
or
jms.
When
you
you
actually
call
receive
in
the
loop.
That's
that's
maybe
interesting
and
then.
E
So
jms
has
different.
It
has
both
listeners
and
polling
so
in
our
jms
instrumentation
for
the
listener,
when
we
do
that
we
parent
to
the
the
propagated,
but
on
the
pulling
one
we
don't
we
parent
implicitly,
and
we
do
that
thing
where
we
wait
to
see.
If
anything,
a
message
came
back
before
we
create
this
band.
C
So
the
reason
why
I,
because
there's
a
common
case
of
kafka
or
any
of
these
messaging
things
where,
like
you,
just
dump
your
events
onto
the
topic
and
then
whoever
listens
to
them
does
stuff.
They
could
be
very,
very
independent
processes
like
they
don't
they're,
not
involved
in
the
response
like
they
might
just
have
some
analytics
back.
Then
that's
reading
messages.
E
C
B
E
B
C
B
B
E
I
I
would
like
that
to
be
in
the
spec
first,
because
I
think
that
we're
following
at
least
the
way
that
the
spec
is
written
today
by
parenting,
the
producer
from
the
consumer
in
the
non-batch
case-
yeah.
Yes,
but
I
understand
what
you're
saying
about
the
question
I
I
think
it
sort
of
depends
on
maybe
yeah.
I
guess
it
depends
on
what
the
user
could
want
to
see
in.
C
A
B
E
E
B
B
B
C
B
B
Well,
in
my
manual,
that's
that's
the
problem,
so
I
have
in
examples
where
I
have
like
integration
tests.
I
have
tests
both
for
file
and
folder.
It
does
work.
I
was
unable
to
write
any
like
actual,
like
unity
test.
B
B
And
our
agent
jar
file
can
have
different
names,
so
currently
it's
currently
being
added
into
this
very
same.
I
don't
think
it
makes
any
actual
difference
between
we
still
delegate
to
agent
class
loader.
So
all
agent
classes
are
still
should
be
loaded
from
agent
class,
loader
or
bootstrap,
but
it's
just
one
class
loader
to
delegate
to
him.
B
Probably
this
is
parent.
First,
I
don't
remember
from
top
top
of
my
head.
C
Oh,
we
can
definitely
imagine
it
put
all
the
jars
in
one
stuff.
B
But
what
do
you
mean
shouldn't
do?
What
have
the
agent
are
in
their
list
of
extension
charges?
I
mean
if
you,
if
you
just
point
to
a
folder,
it
totally
makes
sense
that
all
open,
telemetry,
related
jar
files
are
in
one
folder.
B
That's
a
good
one.
Yes,
maybe
the
current
problem
is
that
so
we
we
currently
call
all
these
classes
all
this
slaughter
class,
other
creations
through
reflection,
and
so
I
have
to
pass
that
code
location
through
reflection
as
well,
which
well
I
can't
do
that.
But
I
was
thinking
that
as
a
separate
pull
request,
I
will
try
to
remove
that
reflection.
B
I
make
a
quick,
a
quick,
quick
test
that,
if
we,
if
trust
you
care
to
open
your
id
and
open
agent,
is
that
agent
initializer.
B
B
B
B
C
E
Yeah
cool:
do
you
want
to
get
this
into
our
release.
E
Should
we
plan,
maybe
early
next
week
for
our
release.
B
C
C
E
That's
good
yeah
because
I
remember
when
hearing
about
the.
E
Not
auto-generated
sort
of
like
it
had
to
be
done
by
the
developer,
and
I
was
a
little
worried
about
that,
but
great
that
it
gets.
C
B
E
Oh
nikita,
thank
you
for
finding
that
grails
that
grails,
a
repo
problem
was
driving
me
batty,
like
just
all
the
prs
were
failing,
and
I
I
was
just
like.
E
B
I
don't
know
it's
every
time
when
I
see
that
gradle
fails
to
download
something.
I
try
to
actually
open
that
link,
and
so,
if
that
links
actually
gives
you
four
or
four
yeah,
then
something
wrong.
I
still
have
no
idea
why
that
lucky
hatred
has
changed.
I
was
unable
to
find
any
documentation
about
that,
but
well,
oh.
E
C
Like
so
my
assumption,
when
that
happened
was
that
we,
like
the
palms,
are
in
the
middle
of
publishing,
so
top
level
palm
was
published
and
the
inner
one
wasn't
published.
Yet
so
that's
right.
Of
course
we
just
have
to
wait
a
bit.
You
know
get
fixed.
E
Oh,
yes,
that
was
my
first
analysis
and
then
after
it
still
kept
failing
for
a
while,
I
was
like
so
I
looked
and
I
was
like
these
are
old
versions.
These
are
not
new
and
then
I
was
like.
Oh
man,
the
whole
site's.
E
C
C
The
problem
is
that
simply
in
this
pin
after
the
response
is
consumed,
if
our
tests
don't
consume
the
response,
the
span
doesn't
get
completed
our
miria.
It
sends
headers
before
content.
Unlike
jetty,
I
guess,
and
so
actually
these
client
libraries
were
processing
the
headers
before
the
content
was
consumed,
causing
the
spam
to
not
be
completed.
C
C
C
C
The
net
instrumentation
just
sort
of
sucks,
like
I
don't
think,
there's
any
way
to
generalize
it
as
we've
talked
about,
I
think,
but
in
some
cases
you
do
need
to
wrap
the
response.
In
some
cases
you
don't
it
depends
on
the
actual
framework,
yeah
and
the
context.
Propagation
is
a
mess.
Then,
like
I've,
been
for
the
past
couple
of
weeks,
like
I
looked
through
the
rat
pack,
instrumentation
a
bit
like
I
mean
every
framework
has
their
own
request
response
objects.
C
Yeah
it'll
be
a
work
I
mean
we're.
Gonna
have
to
re-instrument
these
frameworks,
but
it's
definitely
an
important
thing
to
do
at
some
point.
Rat
pack
is
definitely
on
my
mind,
like
that's
the
next
one.
I
want
to
try
and
just
gradually
try
to
get
rid
of
these
right
back
for
these
instrumentations,
because
other-
and
I
mean
the
killer
thing-
is
that
we
don't.
We
can't
trace
http
2
for
any
of
these
coins,
which
is
a
big
bummer
for
a
lot
of
users.
I'm
sure.
C
E
B
C
E
E
For
the
client
for
the
server,
I
think
we
rely
on
nettie
currently,
but
we
could
that
shouldn't
be
hard
to
fix.
B
E
For
my
personal
answer
for
auto
instrumentation
for
the
java
agent
is
yes,
I
care
a
lot
about
the
spring
auto
instrumentation,
because
exactly.
C
B
I
mean,
but
why
I
mean:
if
user
can
just
okay
about
just
spring
framework,
then
yes,
maybe,
but
if
we
already
talking
about
spring
boot,
spread
cloud
thingies,
it's
easier
for
application
developer
to
just
have
print
screws
and
be
done
with
it.
B
And
so
we
kind
of
have
a
duplicated
effort
with
spring
slos.
B
B
B
E
Because
there's
a
lot
of
spring
web
flex
out
there
and
a
lot
of
people
who
can't
or
don't
want
to
crack
open
their
app
and
make
changes
to
it.
B
C
B
C
C
Because
I
mean
is
like
there's
a
spring,
I
mean,
I
think
it
is
fair
to
ignore
non-springboot
users.
I
don't
know
if
those
even
exist
anymore.
So
if
it's
only
spring
boot
users
I
mean
you
just
have
to
add
the
spring
cloud
starter
or
something
like
that
and
yeah.
So
that
might
be
the
extent
of
what
our
agent
does
like
just
adds
those
helper
things
so
that
the
service
loader
picks
it
up.
Ideally,
I
think
that's
probably
the
ideal
situation
for
us
at
some
point.
C
B
C
C
E
E
E
But
I
will
put
it
on
the
agenda
and
ask
our
data
dog
new,
relic
friends.