►
From YouTube: 2021-07-06 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).
B
C
B
C
B
C
All
right
nikita,
why
don't
we
start
with
this
because
matesh
and
I
both
have
prs
out,
and
I
don't
think
we
would
disagree
with
you
here.
Only
that
can
we
please
just.
Can
we
please
fix
what
to
fix
what
we
have
first
and
then,
if
you
want
to
figure
out
a
better
solution,
we
would
love
it
right
now.
Things
are
badly
broken.
A
Yeah,
that's
that's
exactly
one
of
my
problems
is:
are
you
sure
that
our
currently
thinking-
oh,
let's
wait
for
so
yeah?
Are
we
sure
that
our
current
fixes
for
shadow
for
shadow
related
problems
make
things
actually
better.
A
Okay,
I
mean
that
I
I
don't.
I
don't
oppose
to
get
to
current
pull
requests,
yes,
because
exactly
because
they
fix
our
immediate
problems,
but
I
I
see
that
when,
when
we
fix
or
patch
current
problems,
well,
we
have
to
review
everything
regarding
packaging
and
shadow
and
shadow
and
and
whatnot
I
mean
my
let
let
me
for
a
second
start
with
one
simple
question:
in
my
opinion,
simple
question:
how
how
how
many
shadow
jar
for
fat
jar
do
we
actually
want.
C
A
C
C
So
like
the
well
yes,
but
I
guess
to
your
the
you're
saying
aggregating,
we
should
have
two
fat
chars
right.
A
A
A
A
E
A
A
A
So
that
that's
terrible
in
my
opinion,
exactly
for
I
understand
your
point
for
library
and
library,
instrumentation
and
its
transitive
dependencies,
yes,
okay,
but
for
for
extension,
development,
not
having
proper
transitive
dependency,
but
just
one
shaded
jar.
I
think
that
that
makes
complicated.
I
mean
I.
B
C
A
C
Which
also
gives
us
more
flex,
I
think,
is
better
in
the
long
run
for
the
agent.
Also
to
give
us,
you
know,
we've
more
flexibility
if
we
want
to
switch
it
out
with
some
other
catch
yeah.
D
C
Datadog
has
seen
some
issues
with
class
loading
done
by
caffeine,
which
caused
deadlocks,
which
we
were
able
to
convince
ourselves
that
that's
not
a
particular
problem
for
us.
A
A
A
How
will
we
avoid
like
duplicate.
C
C
E
E
E
Embedding
yet
was
that
I
thought
not
embedding
might
keep
our
build
scripts
simpler
until
trust
found
this
problem,
and
then
it
didn't
solve
the
build
script
problems.
So
in
that
case,
embedding
is
probably
it
keeps
the
artifacts
easier
to
reason
about
and
solves
this
implementation
detail
issue
besides
it
from
the
users.
A
Okay,
but
the
mateo's,
your
last
pull
request
that
I
just
commented
some
many
minutes
ago
when
we
had
to
extract
shadow
jar
that
we
have
just
created.
It's
like
in
the
field
of
red
flags.
For
me,
it's
it's
unnatural,
it's
abomination!
Why
do
we
need
to
do
that.
E
By
the
way,
nikita
did
you
find
it
better
before
the
pr
or
is
it
equally
bad,
because
just
to
clarify,
even
in
the
previous
version,
we
are
extracting
it
just
it
happens
to
be
inline
into
that
copy
task
instead
of
a
separate
folder.
But
the
concept
of
extracting
the
jar
has
not
changed
with
that
vr
and
the
reason.
A
A
E
A
Okay,
then,
I'm
still
not
happy
with
our
current
approach,
our
future
approach.
I
I
want
to
try
to
find
a
way
to
make
that
simple,
but
for
that
I
think
we
at
plus
should
have
some
automated
way
to
find
if
we
break
our
publishing
again
so.
A
Apart
from,
let's,
let's
make
a
shell
script,
which
does
maybe
publish
to
made
in
local,
then
find
all
jars
and
or
palm
files,
and
oh
what
else
I
don't
know.
A
C
Yeah,
that's
how
I
found
the
the
the
instrumentation
api
caching
also,
because.
B
E
B
E
E
E
A
E
A
E
A
E
A
B
E
A
C
Did
I
get
which
one
of
these
is
it
testing
common
is
the
one.
A
B
It
is,
but
I
mean
we
are
saying
that
something
was
also
just
about.
E
A
B
A
A
E
A
E
C
B
C
C
Okay,
nikita,
do
you
want
to
open
a
tracking
issue
for
this,
or
do
you
want
me
to.
D
C
Yeah,
I
think
this
is
a
great.
This
is
a
great.
C
Idea
all
right,
everybody
sufficiently
unhappy
about
shading
now.
D
C
Let's
see
I
put
this
on
here,
I
was
looking
to
get
at
the
instrument
a
little
bit
more
doing
a
couple
of
patchy
things,
and
so
I
had
noticed-
and
onorak
had
mentioned
previously-
that
brave
kind
of
this
idea
of
like
extractors
versus
rappers-
that
if
we're
always
going
to
wrap
the
request
object,
maybe
we
don't
need
an
extractor
or
there's
it's
anyway.
Just
wanted
to
bring
that
up
again
see
if
there
was
any
if
I
was
still
sort
of
understanding
that
right.
E
E
E
C
F
E
C
And
then
the
extract
yeah,
the
extractor
was
pretty
dumb,
then
just
always
calling
request
dot
on
the.
E
Yeah-
and
I
mean
so-
I
think
brave
deprecated,
the
non-rappers
but
brave,
is
implemented.
Recently
I
mean
there's
a
bridge.
I
forget
if
the
rapper
calls
into
the
extractor
the
other-
it's
probably
the
other
way
around,
but
of
course
this
can
be
implemented
and
if
you
want
to
just
support
both,
that's
also
something
we
can
do.
E
E
C
Okay
trash
one
logger:
what
can
we
strike
log
because
of
the
the
upper
case?
Okay,
so
yeah
look
at
that.
I
vote
lager
nikita.
E
B
B
A
E
D
E
C
A
C
This
came
up
so
I
added
this-
I
don't
know
if
you
caught
this
discussion,
asking
if
you
know
kind
of,
could
document
more
strongly.
This
memory
leak
issue
and.
B
C
Push
back
a
little
bit
kind
of
just
asking
about
it,
and
so
this
I
think
I
finally
understood
this
comment,
which
is
sort
of
like
okay.
If
we,
if
it
was
a
big
enough
bug,
projects
would
like
release
a
patch
release
right
away
on
the
older
version.
So
when
you
see
the
new
version,
when
you
see
one
three
zero
you
there
might
be
new
bugs
in
there,
but
you
can
be
relatively
sure.
The
last
major
version
didn't
have
any
showstoppers.
C
Okay,
cool
so
yeah,
I
was
kind
of
thinking
about
what
would
you
know
what
would
trigger
us
to
release
patches
and
hopefully,
once
now
that
we
all
the
gradle
changes
settle
down
and
making
patches
and
merging
stuff
will
be
easier?
Well,
I'm
sure
we're
still
churning
on
the
api
and
stuff.
But
in
general
I
mean
security
issues,
of
course,
which
we
haven't
had
any
significant
regressions
like
the
jms
instrumentation
breaking
in
one
three
zero,
even
though
it
even
though
it
didn't
break
the
application,
it
was
a.
E
C
Yeah,
I
guess
regression
would
be
like
I
was
guess.
I
was
thinking
of
things
that
we
don't
necessarily
contin,
consider
stable
like
semantic
attributes
or
something.
B
E
E
C
Memory
leaks
seemed
like
like
a
good
one,
just
because
it's
not
always
apparent
to
somebody
when
they
test
it
out
in
like
in
their
staging.
They
may
not
hit
it
right
away
versus,
say
any
bug
like
when
you
know
like
say
the
grpc
even
like
that
grpc
stack
overflow
bug,
I'm
not
sure
I
would
say
we
would
patch
that,
since
it
was
pre-existing.
B
E
D
D
C
C
Send
a
pr
to
to
document
that
since
feels
like
something
that
people
have
asked
before.
C
Yeah,
so
I
was
gonna
should
do
you
think
we
should
redirect
this
issue
to
the
instrumentation
repo
or.
A
I
think
that
in
long
run,
country
prepo
is
exactly
the
reaper
where
we
want
our,
like
individual
third
party,
contributed
instrumentations
and
we
actually
have
to
move
the
majority
of
our
existing
instrumentation
lower
there.
I
think
in
some
meter,
but
for
that
to
happen,
we
have
to
have
infrastructure
in
place
to
actually
consume
somehow
consume
instrumentation
from
from
that
repo
or
it
could
let
our
customers
to
consume
it.
C
Well,
we
could
kind
of
like
the
collector
does.
I
think,
I'm
actually
I'm
just
guessing
honorable
correct
me
that
the
collector
contrib,
so
we
could
publish
a
java
agent
from
contrib
repo
that
has
the
java
agent
plus
all
the
contrib
things
bundled.
A
B
F
A
I
think
we
have,
I,
I
think
we
have
to
start
saying
some
point:
the
directing
more
attention
to
country,
both
our
intention
and
and
community
attention
yeah.
So
for
that
reason
I
would
not
move
this
issue
to
instrumentation,
that
that
would
be
the
first
one
like
yeah.
Let's
start
discussion
and
you
id
probably
even
if
we
have
some
ideas
in
instrumentation
repo
repo
already
issues.
Let's
do
that
instrumentation
and
we
don't
have
any
capacity
yet
probably
even
move
that
here
at
some
place
at
some
time.
A
C
Did
I
will
add
this
to
the
thursday's
agenda
and
make
call
out
to
other
folks
to.
C
C
Would
you
like
to
be
nikita
and
I
shouldn't
ask
I
shouldn't
ask,
do
you
want
to
be,
I
should
ask,
can
we
make
you
an
approver
and
or
maintain
it,
we
should
yeah,
I
mean
really.
We
should
just.
I
think
contrib
should
probably
be
like
a
super
set
of
everybody
from
both
sdk
and
instrumentation.
E
C
At
least
for
maintainers,
I
would
say
I
don't
see
any
reason
why
we
wouldn't
add
john.
C
C
Luckily,
I
have
your
name,
your
last
name
on
screen.
Here
you
missed
one
s
before
the
z.
B
C
B
By
the
way,
do
we
have
any
steps
or
conditions
online
where
we
would
start
moving
our
instrumentations
there,
because
I
can
imagine
that
we
would
need
probably
both
muzzle
plugins
to
be
extracted
and
our
api
is
very
stable.
But
do
we
have
an?
I
don't
know,
milestone
issue
or
any
talk
about
this
nope.
F
B
A
C
All
right,
we
had
an
agenda
and
we
went
through
our
agenda.
Anybody
have
more
things
that
we
should
chat
about.
C
I
did
what
I
said
I
would
do
at
the
last
minute.
I
reviewed
the
the
attribute
pr,
so
it
looks
good
to
me.
I
think
so
the
question
we
had
an
open
question.
We
need
to
get
back
to
john
and
anarag
whether
to
keep
the
attribute
the
annotations
in
the
one
four
zero
release
or
to
pull
them
until
one
five
zero.
C
C
So
shall
we
just
ask,
we
can
just
ask
justin
what
did
I
somebody
knew
his
real
name?
I
think.
C
But
he's
serving,
I
did
this
last
week,
yes,
ask:
we
can
ask
them
if
they
can,
if
they're
good
with
that
plan,.