►
From YouTube: 2021-05-21 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
D
C
C
E
E
Yeah,
it's
actually
really
fast
and
it's
still
got
the
same
engine
right.
It's
still
got
chromium
rendering,
but
it's
weird
when
I
paste
it
into
the
dock.
I
pasted
just
the
link
into
the
dock
for
my
feature
request
and
it
posted
that
whole
description
as
a
part
of
it,
which
I
don't
understand.
It's
very
strange.
D
D
D
B
D
B
All
right
well
we'll
see
how
I
might
fail
a
little
early.
I
got
my
second
vaccination
shot
this
morning
and
I
just
started
feeling
it
about
half
an
hour.
B
Not
flew
down
hard
yeah,
so
yeah,
I
I
think
that's
I
was
not
expecting
until
like
seems
like
later
later
tonight,
or
it's
very
mild
so
far,
but
it's
going
in
the.
E
B
Okay,
yeah
I've
also
had
so
many
people.
I,
the
people
in
my
team,
have
had
some
pretty
strong
reactions
to
the
second
so
just
hearing
about
it
has
me
like
panicking.
D
B
Powerful,
oh,
we
had
a
good
meeting
this
morning.
I
was
pretty
excited
about
the
logging
stuff.
A
couple
new
folks
joined
from
log
z,.
B
And
they
are
doing
some
work
on
logging
want
to
contribute
like
in
that
in
the
in
that
direction.
So
it's
always.
E
Yeah
and
I
think
specifically
or
especially,
an
otop
exporter,
on
top
of
that
cool
yeah,
my
request
was:
if
you're,
going
to
use
what
david
ponslow
started.
Building
please
use.
E
B
At
yeah,
so
that
was
good.
I'm
glad
that
they
joined
to
chat
about
that
also
because
they
were
sort
of
just
going.
D
D
B
Logging
sdk
logging
exporter
thing.
E
So
I
think
there'll
be
value
in
doing
some
analysis
and
taking
a
look
at
the
various
appender
interfaces
across
the
different
logging
libraries
in
java
and
trying
to
find
a
common
some
common
ground
that
we
could
use
so
that
it'd
be
really
easy
for
people
to
not
have
to
worry
about
the
mapping
to
otlp
but
just
map
to
a
more
easy
to
understand.
Yeah.
E
E
I
think
that's
a
piece
that
all
of
them
are
going
to
need
right,
because
all
the
I
think
the
appender
apis
are
just
one
one
log
line
at
a
time,
and
if
you
want
to
use
otlp
with
a
collector,
you
definitely
want
to
have
batching
because
there's
it
doesn't,
I
don't
think
there's
any
streaming.
Implementation
for
the
grpc
in
the
collector.
D
E
E
D
D
Or
yeah
I
mean
if
if
we
just
mdc
is
only
strings,
I
guess.
D
D
E
B
Is
otlp
the
only
I
mean
the
only
thing
we
would
want
to
export?
To
I
mean
that,
like
I
I
wanted,
I
thought
there
would
be
a
a
logging
exporter
sort
of
similar
to
the
span
exporters
that
so
that
people
could
plug
in
their
own
whether
it's
jager
supports
logs
right.
B
I
know
I
would
want
to
export
logs
to
our
back
end
and
we're
not
using
otlp.
You
should
use
otp.
E
E
D
E
So
maybe
for
go,
it
would
make
sense
to
to
try
to
become,
but
for
java
it's
like
nobody's
going
to
use
a
new
another
logging
api.
At
this
point
like
it
would
be
crazy,
so
yeah,
I
think
that
at
the
moment
we're
definitely
punting.
The
the
the
spec
group
is
punting
on
a
logging
api
and
focusing
more
on
like
what
needs
to
be
in
available
for
an
sdk.
D
E
E
D
E
B
E
I
think
logging
instrumentation
would
use
whatever
logging
framework
was
was
adjacent
to
whatever
was
being
instrumented
like.
If
the
library
has
it
most.
Libraries
are
just
going
to
use
slf
right.
So
if
you
oh.
B
I
mean
if
you're
writing
a
the
appender.
D
D
B
So
we
wouldn't,
we
wouldn't
try
to
bridge
the
way
that
we
do.
We
wouldn't
like
yeah.
Somebody
brings
the
append
there.
D
B
Yeah
jonathan,
who
came
and
was
checking
in
on
progress
here
so
just
gave
him
kind
of
the
latest.
He
wanted
to
confirm
that
we
weren't
waiting
for
metrics
api
to
be
stable.
I
explained
how
we
have
the
we
do,
have
a
dependency
on
it,
but
those
methods
are
marked
unstable.
B
Oh
yeah,
then
nikita
talked
about
the
java
agent,
the
extension
stuff,
and
I
was
saying
how
I
liked
how
you
know
we
didn't.
We
weren't
introducing
any
new
api
in
this,
since
it's
all
based
on
the
auto
configure
module.
But
nikita
was
pointing.
D
D
D
D
B
E
It
tyler
had
an
idea
about
extracting
all
of
the
the
bits
of
muzzle
that
are
about
pulling
down
and
processing
various
versions,
like
version
ranges
of
libraries
and
making
that
independent
of
what
you
do
with
those
libraries
once
you
have
them
and
that
it
was
commenting
that
could
be
a
generally
useful,
gradle,
plugin
or
gradle,
something
that
would
let
you
basically
do
provide
callbacks
for
what
you
want
to
do
with
various
version
version
ranges
and
servers
like
security
scans.
E
D
D
Maybe
it's
not,
though,
because
it
does
not
lock
us
into
gradle.
Do
you
have
any
thoughts
like
there
was
that
issue
filed
about
the
maven
plug-in?
I
guess,
and
then
it
just
got
closed
by
the
person
for
some
reason,
which
maybe
I'm
somewhat
happy
about,
because
I
don't
want
to
write
them
even
plug-in.
B
Yeah,
the
shading
is
shouldn't
be
too
hard
to
just
come
up
with
the
same
maven
shade
plug-in
configuration,
for
example,.
D
D
B
D
B
B
B
Because
it
didn't,
I
needed
to
shade
for
shading
resources
in
a
way
that
are
sort
of
more
arbitrarily.
B
B
Yep
yeah
for
the
circular
dependency
mateis
was
describing
his
current
thought
was
just
that
we'd
release.
It
would
always
lag
one
version,
so
the
maven
plug-in
would
depend
on
the
stable
and
that's
where
he
wanted
to
get
one
of
his
motivations
for
getting
the
muzzle
stuff
stable
is
then
the
plug-in
can
depend
on
the
last
stable
release
of
the
stable
release
of
the
muzzle
package
stuff
in
tooling
yep.
B
Yeah
and
I
put
through
this
on,
but
before
you
I
saw
you
already
commented
on
it,
so
that's
cool.
I
just
wanted
to.
B
E
E
E
It's
so
terrible,
300
overhead.
You
know
whatever
random
stuff
people
spew
out
all
the
time
and
of
course
my
dark
deep
fear
is
that
the
problem
is
not
with
the
agent
at
all.
The
problem
is
with
the
sdk
and
there's
a
performance
problem
right
like
in
the
sdk
itself,
and
the
agent
doesn't
have
anything
to
do
with
it.
So
I
really
would
love
it
if
we
had
a
way
to
turn
on
the
agent
with
an
sdk.
That
did
nothing.
E
Perhaps
if
you
had
enough,
you
know
enough
data,
you
could
decompose
where
the
overhead
was
coming
from
and
it
will
potentially
allow
us
to
focus
performance
improvements
on
where
it
actually
is
important,
because
right
now
I
think
we
have.
We
have
it's
a
little
bit
of
a
guess
where
the
where
the
cpu
and
memory
is
being
used
without
doing
some
really
deep
profiling
anyway.
So
this
is
my
idea,
it
was
like
hey,
let's
have
an
agent
option
where
the
sdk
just
gets.
E
D
B
Hey,
let's
see
also
exporting
to
like,
like
having
a
good.
I
guess.
Oh,
we
can
do
export,
not
exporter,
none
already,
because
that's
potentially
one
source
of
overhead
yeah.
E
Definitely
export
is
generally
going
to
generate
a
bunch
of
garbage
during
you
know,
during
the
export
process
you
got
to
do
a
lot
of
conversion
of
certain
parceling
of
types.
E
Yeah
yeah,
I
mean
I
think
we
can
really
break
it
down
quite
granularly.
If
we
get
enough
options
here
could
be
interesting.
D
D
D
D
Right,
that's
true,
so
that
doesn't
help
interesting.
E
E
D
D
Yeah-
and
so
I
mean
I
already
knew
like
my
idea-
is
to
actually
implement
fast
attributes
in
instrumentation
api,
so
don't
use
our
attributes.
Class
use
a
custom
implementation
of
the
interface
because
we
don't
pass
it
to
users,
so
that's
easily
solvable,
but
for
now
it's
probably
adding
a
bit
of
overhead
now
yeah
anyway.
There's
lots.
E
There's
so
many
moving
parts,
it's
going
to
be
nice
if
we
can
isolate
the
moving
parts
and
see
where
the
real
problems
are
anyway
yeah.
This
is
where
what
like
this
idea
would
be
great,
because
we
could
turn
it
on
and
say:
oh
just
using
the
api
without
having
any
sdk
under
the
hood.
It
has
this
level
of
overhead
and
we
can
then
hopefully
isolate
it
to
attributes
or
whatever,
whatever
the
particular
issue
is.
B
Do
you
think
customers
would
be
willing,
like
to
run,
say
a
jfr
jfr
profile
and
uploading
that,
like
sharing
that
publicly.
B
Because
I
was
just
trying
to
think
of
like
going
forward
even
once
we,
you
know,
even
with
these
we're
still
gonna
have
a
lot
of
questions,
we're
still
gonna.
This
is
still
gonna
keep
coming
up,
and
if
we
can,
I
don't
know
like
what
I
I
had
done
previously
like
to
write
some
part,
some
kind
of
custom
parsing
of
jfr
files
to
look
for
glow
root
code
that
was
being
on.
That
was
on
the
hot
path,
and
so
something
like
that
could.
D
E
Well,
if
you
want
to
talk
parsing,
jfr
jason
blum
has
a
lot
of
experience
as
well.
You
also
have
some
experts
at
microsoft.
I
think
recently
acquired
experts.
B
Yeah
and-
and
I
I
don't
have
to
go-
make
friends
with
them
like
action
item
from
last
night.
B
D
E
D
B
B
Yes
remind
me
last
name.
E
Your
audio
is
fading
out
slowly,
I
don't
know.
What's
going
on,
no,
maybe
you're
fading
out,
maybe
exactly.
B
B
All
right
anything
else
to
chat
about.
B
E
D
E
E
B
B
Yeah
pretty
excited
for
we're
now,
four
weeks
away
from
total.