►
From YouTube: 2020-08-12 .NET Auto-Instrumentation SIG
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
B
A
I
just
put
the
the
dark
for
the
meetings
on
the
chat,
so
if
anyone
once
you
sign.
A
A
C
A
E
C
Cool
so
hi
guys,
maybe
we
can
get
started.
C
I
think
I
just
wanted
to
say
hi
to
people
and
then
hand
over
to
the
x-ray.
Folks
is
the
extra
folks
already
here,
yeah
but
you're
here,
right,
cool,
yeah,
hi
everyone
and
I
think
we're
also
recording
so
that
we
can
have
like
point
people
to
any
information
technical
information
about
later.
That's
actually
really
cool,
so
hi
folks,
and
also
maybe
we
can
have
a
quick
round
of
introductions
for
those
who
were
not
here
before.
C
I
have
folks
here
from
datadog
who
are
on
the
tracer
team
and
who
have
worked
on
the
code
base
for
longer
than
my
like
longest
of
us,
so
they
would
be
great
feedback
for
anything
that
we
do
provide
feedback.
So
maybe
we
can
do
a
real,
quick
round
of
introduction,
although
we
did
it
last
time
as
well.
Maybe
really
really
quick.
F
I've
been
here
for
a
year
and
so
have
a
lot
of
experience
with
our
crew
base
happy
to
participate
in
this
meeting.
D
Hi,
I
am
lucas.
I
work
with
greg
and
zack
on
the
net
tracer
at
datadog.
I've
been
here
for
a
little
over
two
years,
I'm
the
oldest
engineer
in
the
team,
which
just
means
that
all
the
bad
code
that's
been
there
for
a
long
time
is
all
mine.
A
Ahead:
okay,
so
I've
been
working
for
splunk
and
with
the
signalfx.net
implementation,
I'm
trying
to
push
the
out
experimentation
for
some
time
for
dotnet
and
open
telemetry.
I
work
on
the
collector
and
I'm
being
kind
of
cleaning
up
my
plate
that
you
start
to
really
work
on
the
auto
instrumentation
for
dotnet.
C
And
we
have
folks
from
x-ray.
G
Oh
hi,
this
is
bullpen
and
I've
been
working
in
aw618
for
like
half
a
year,
and
I
was
mainly
working
on
the
donate
agent.
H
Microsoft,
working
on
the
dot
net,
four
agent
team
with
mikhail
and
just
just
joined
microsoft,
six
seven
months
back.
C
Okay,
cool
which
specific
team
like
we
haven't,
met,
but
hi.
H
Yeah
we
talked
once
saw
the
phone
with
mikhail
noah.
I
work
with
mikhail
cool.
G
C
Okay,
mickey
has
problems
with
audio
as
usual.
Maybe
james
can.
H
C
C
B
Okay,
yeah,
so
hello,
everyone,
I
am
sijo.
I
work
in
microsoft
and
I
work
mostly
in
the
open
elementary
sdk
and
I
thought
it's
best
if
someone
from
the
sdk
repo
joins
the
auto
instrumentation
efforts
as
well,
so
I
won't
be
actively
contributing,
but
I
will
try
to
like
actively
follow
what's
going
on
here,
yeah,
that's
it
cool.
C
I
E
C
And
then
we
also
have
james,
who
is
also
from
micros.
He
had
just
a
sound
issue
earlier.
We
couldn't
hear
anything
but
he's
also
from
microsoft
and
is
also
on
the
same
team
as
mikhail
and
or
like
a
sister
team
of
the
same
organization.
Anyway,
I'm
exactly
sure.
C
Anyway,
welcome
everybody
so
let's
hand
over
to
the
x-ray
fox
and
they
can
talk
about
their
agent.
I
would
really
love
to
learn
more
about
the
architecture
and
how
you
guys
collect
things.
Do
you
want
choose,
mr
stop
sharing,
so
you
can
share
your
screen.
G
G
Oh
okay,
so
so
first
I
have
to
enable
the
screen
sharing.
So
I
have
to
create
and
redraw
again.
C
A
Yeah,
so
just
a
clarification
how
this
is
organized
on
the
ripples,
because
I
I
actually
had
a
bit
confused
in
my
mind
when
I
I
created
actually
open
telemetry,
already
create
the
teams
that
are
gonna,
be
assigned
some
of
their
rights
and
that
are
the
ones
allowed
to
take
some
actions
on
the
repo.
So
right
now
as
maintainers,
it's
me
greg
and
seizure.
A
We
are
the
ones
that
have
the
rights
to
merge
and
together
if
they
are
approvers,
because
the
approvers
can
of
course
approve
prs.
But
the
three
of
us
are
the
ones
that
can
merge.
A
Open
telemetry
has
kind
of
very
define
very
strictly
who
can
be
approvers
or
maintainers,
but
the
requirements
are
kind
of
a
bit
strong
for
teams
that
are
being
just
starting.
A
So
it's
cheap
in
this
case
that
if
you
have
the
experience-
and
you
do
a
few
prs
who
you
can
become
a
member
approver
as
long
as
you
have
the
knowledge
or
you
have
people
that
are
already
part
that
are
backing,
that
you
have
the
knowledge
and
that
you
are
going
to
contribute.
A
So
I
think
for
people
that
wanted
to
be
really
involved
in
in
the
product
this
in
the
project.
This
is
the
the
right
time
to
kind
of
join
on
board
with
us
and
start
to
make
the
request
to
kind
of
join
and
participate.
A
My
experience
with
the
collector
is
that,
depending
on
how
busy
things
are,
we
need
a
lot
of
approvers
because
the
the
number
of
reviews
can
be
pretty
high
depending
on
the
cycle
of
the
product,
but
we
also
want
to
keep
a
high
quality
and
the
product
cohesive.
So
it's
it
requires
a
bit
of
time
to
dedicate
to
be
an
effective,
approver
and
maintainer,
but
for
people
that
are
really
willing
to
participate
and
are
going
to
contribute
to
this
project.
A
This
is
a
moment
that
you
have
the
opportunity
or
not
having
that
to
satisfy
that
criteria.
I
don't
remember
exactly
the
numbers,
but
it's
like,
I
think,
for
a
established
repo.
You
have
to
have
like
30
prs,
approve
it
and
merge
it
to
become
another
approver.
You
know,
so
this
time
is
is
a
very
good
time
for
people
that
really
want
to
get
involved
in
the
project.
A
I
have
a
pr
opening
to
set
up
the
initial
things
and
I
would
like
kind
of
to
start
with
that,
and
then
we
keep
adding
people
as
we
we
do,
because
we
need
to
have
the
owners
before
we
start
to
to
merge
code
on
the
ripple.
So
there
is
our
red
apr
open
for
that,
and
I
encourage
people
to
kind
of
reach
to
us
to
kind
of
use
this
moment
to
be
if
they
want
to
really
be
involved
in
the
project
to
be
approvers
and
eventually
maintainers
too.
C
Yeah
exactly
I,
I
have
the
same
thought
we
so
I
think
the
entire
datadock.net
tracer
team
should
be
approvers.
C
C
I
don't
know
whether
so
I
would
definitely
like
to
be
someone
else
from
datadog.
It
should
be
also
maintained
is
simply
because
I
know
that
I
cannot
scale
to
this
completely
otherwise,
so
I
I
don't
want
things
to
be
blocked
on
datadog,
but
they
might
become
if
I'm
the
only
person,
because
if
I'm
critically
busy
with
something
else,
then
you
know
so
so
far.
We
have
besides
me,
two
people
who
have
agreed
to
be
maintainer
if
we
need
that.
So
that
is
something
I
guess
for
us
together
to
decide.
C
If
we
need
so
many
maintainers,
if
we
do,
then
there
is
lucas
and
and
tony
who
said
that
they
up
for
it
and
tony
is
also
in
the
european
time
zone.
So
that
would
allow
us
a
little
bit
more
kind
of
addressing
in
different
time
zones.
C
If
we
believe
that
we
shouldn't
have
so
many
maintainers,
then
I
I
don't
know
like
it's
it's
up
for,
for
you
guys
to
to
say,
then
we
can
just
pick
one
person
besides,
but
I
think
definitely
everybody
from
datadog
should
be
approvers.
A
I
I
I
want
to
say
that
I
I'm
giving
chances
to
other
people
speak,
but
from
from
my
part,
I
think
that's
a
a
very
good
idea.
You
know
we
need
the
the
people
with
the
experience
in
the
code
base
and
helps
keep
the
project
moving
in
the
right
direction.
So,
for
my
part,
I
think
it's
great.
You
know
I
I'll
be
looking
to
sponsor
both
lucas
and
tony
to
to
be
members
and
and
maintainers,
but
if
other
people
have
some
some
something
else
to
add
or
some
different
opinion,
please.
E
I
think
let's
wait
maybe
a
bit
so
right
now
we
have
a
series
of
presentations,
so
we
had
a
new
relic
last
time
and
we
are
having
aws
today.
E
E
The
idea
is
just
to
understand,
like
what's
going
to
be
the
participation
from
each
company
and
see
then
like
what
could
be
the
right
action
for
maintenance,
but.
C
E
C
You
know
on
a
technical
level
for
sure
I
just
we
would
need
if,
if
you
want
to
create
a
different
base
for
this
like
a
whole
different
base,
that
means
you
need
to
go
back
to
essentially
going
through
the
motions
of
like
both
a
technical
as
well
as
an
organizational
process.
A
A
I
I
think
the
this
has
been
a
long
discussion
on
this
topic,
and
that
is
also
the
side
of
the
open,
telemetry
itself
that
basically
already
have
a
kind
of
path
with
datadog.
To
do
this
process
you
know,
so
I
think,
as
greg
said,
we
are
interested
and
we
are
welcome
all
the
other
parties,
but
the
decision
was
kind
of
to
follow
this
path
with
datadog.
E
Well,
I'm
saying
that
we
got
a
new
input
right.
This
is
just
newly
released
by
the
new
relic
justice,
but
we
didn't
have
this
data
before,
since
we
haven't
pushed
data
dock
to
open
telemetry,
yet
I'm
just
suggesting,
let's
give
it
a
shot,
and
I
understand
that
we
are
almost
closed
of
a
data
dog
and
choosing
the
base
agent,
but
let's
give
it
a
shot
just
in
case
if
it
will
save
us
a
couple
of
three
months
later
on
in
15
issues
which
we
know
exists
in
that
documentary.
E
Imagine
I'm
just
saying
like
let's
be
open-minded,
let's
not
rush
it
too
much.
Let's
give
it
time
frame
time
box
it
and
we'll
pass
it
from
there.
C
That
makes
sense,
but
what
are
the
specific,
because,
let's
just
beyond
listening
to
the
technology,
and
they
should
to
time
boxes,
discussion
now
as
well,
so
that
we
have
enough
time
what
is
the
and,
by
the
way,
the
whole
maintainers
and
approvers
saying
we
can
still
do
it.
It
shouldn't
be
blocked
on
on
on
on
online.
C
E
C
What
I'm
saying
is
like
what
what
specific
answers,
because,
let's
just
wait
and
think
I
I
don't
know
that
that's
just
a
delay.
But
if
you
have
a
specific
question
that
you
would
like
to,
I
have
answered
that
would
make
sense.
E
E
Folk
also
are
interested
in
trying
new,
relic
and
comparing
it.
This
is
just
to
understand,
like
will.
We
save
more
time
we're
moving
with
new
relic
agent,
potentially
or
whatever
our
agent
if
a
ws
has
different
one
right
or
if
we
will
actually
be
better
off
just
moving
forward
with
data
dog
one
and
fixing
issues.
C
So,
okay,
what
is
the
time
frame
that
you
would
like
to
look
at
this.
E
I
don't
need
a
week
not
moving
a
week.
I
think
we
should
be
enough
and
also
I'm
very
interested
in
discussion
and
open
discussion
having
let's
say
next
week.
E
I'll
have
my
opinion
from
next
week
as
well
and
have
a
new
reality
folks
and
discuss
what
what's
in
best
interest
for
them
as
well,
so
just
to
like
have
better
better
community,
and
I
don't
want
to
say,
hey
new
relic
you've
been
too
late
right,
that's
not
the
case.
We
just
formed.
We
seek
and
I
see
homes
they're
like
what
do
you
really
want?
Do
you
want
to
move
to
that
way?
I
seen
that.
E
A
C
A
C
Yes,
you're
right:
let's,
let's
do
that
so
either
way.
As
a
summary,
I
think
michelle,
I
agree
that
we
should
be
thoughtful
about
this.
I
just
what
I
want
to
avoid
is
because
we
can
always
keep
delaying
endless
new
information.
So,
from
my
perspective,
if,
as
long
as
we
time
box
this
and
a
relatively
like
more
or
less
clear
way,
this
is
a
good
discussion.
I
just
don't
want
to
be
be
here
in
a
month's
time
having
the
same
conversation,
so
I
think.
C
Point
but
let's,
let's
not
have
this
for
too
long.
J
I'll
just
comment
really
quick,
I'm
sorry
I
hopped
on
late,
I'm
alan
from
new
relic
chris,
who
presented
last
week,
is
not
here
today
the
from
new
relic's
standpoint
you
know
in.
We
simply
want
to
participate
in
what
seems
best
for
the
community
and
understanding
that
that
maybe
the
data
dog
agent
or
you
know
I
mean
that's
the
direction
we're
headed.
That
is,
you
know
we
want
to
support
the
community
decision,
but
absolutely
any
other
questions
about
the
new
relic
agent
we
are
here.
J
We
can
answer
those
questions
to
the
note
about
approver,
though,
regardless
of
direction
that
the
community
ends
up
going.
We
very
much
are
interested
in
being
a
very
engaged
member
of
this
project.
So
that's
awesome.
C
One
thing
that
michelle
you
would
need
to
answer
with
eleanor
and
his
colleagues
is
so
it
is
not
the
same
to
open
source
things
and
to
donate
them
to
open
telemetry.
It's
I'm
not
a
I'm,
not.
E
C
I
Can
I
squeeze
in
one
more
introduction
here:
we've
got
dave
on
my
team,
so
even
though
you
guys
may
think
of
me
as
the
I-corps
profiler
person
really
for
the
last,
I
don't
know
two
years
or
so
I
david
mason,
there
has
he's
been
doing
all
the
day-to-day
stuff
you
may.
Some
of
you
may
have
talked
to
him
on
on
email.
If
you
had
profiler
issues
so
yeah
he's
basically
the
one
running
the
show,
I
don't
know
dave.
If
you
want
to
say
anything,
I
may
have
stole
the
thunder
there.
I
K
Yeah,
that's
that's
pretty
much
it!
I
I
do
the
day-to-day
stuff
and
I
core
profiler
right.
You
know,
write
features,
test,
communicate
with
people
figure
out
issues
noah's
doing
a
lot
of
the
driving
on
the
like
open
telemetry
but
yeah,
I'm
I'm
the
icor
profiler
guy.
G
So
hi
everyone,
so
my
name
is
photopin
and
I've
been
working
on
the
aws
x-ray
donor
agent,
and
so
I'm
gonna
briefly
introduce
our
architectures
and
the
features.
So
our
agent
consuming
consists
of
two
parts.
One
part
is
the
yeah
auto
instruction
library
and
that
part
is
the
provider.
G
So
the
all
these
jump,
teaching
libraries,
we
just
apply
the
diagnostic
listeners
to
intercept
the
information
during
the
request,
response
lifecycle
and
we
integrated
with
aws
x-ray,
donate
sdk
to
like
generate
trace
context
and
then
further
send
it
to
our
daemon.
So
here
we
are
using
the
four
types
of
diagnostic
nesters
which
are
provided
by
the
donaire
runtime.
G
We
have
the
sp
net
core,
the
hp
handler
circle
kind
and
the
different
work
and
the
the
initialization
of
this
diagnostic
listener
is
very
straightforward.
You
just
simply
call
the
diagnostic
subscribe,
then
you
can,
then
there
you
go,
and
here
we
are
combine
the
all
of
the
initialization
of
this
type
of
all
these
diagnostic
listeners
into
one
initialization
method,
which
we
call
as
the
hooking
method.
G
So
if
customer
wants
to
using
the
auto
instrumentation
libraries
they
just,
they
only
need
to
call
this
function
in
their
application
so
that
their
application
can
get
instrumented.
G
But
we
don't
want
the
customer
to
make
any
code
change.
So
here
we
are
using
the
provider
to
automatically
insert
this
hooking
method
in
the
application
in
the
wrong
time.
So
the
provider
is
also
very
straightforward,
so
we
just
call
using
the
providing
api,
and
so
we
call
the
gip
compile
it's
not
called
it
would.
G
When
the
jt
compiles
data
method
is
called
which
in
sp
network
application
should
be
the
main
function,
and
then
we
call
the
get
iel
function
body
from
the
main
function,
and
then
we
just
insert
the
il
code
of
all
the
instrumentation
hooking
method
by
coding.
This
method
set
io
function
body
so
that
the
ir
code
of
user's
application
is
modified
and
our
the
the
authentication
network
has
been
inserted
into
the
application
so
that
their
applications
get
is
automatically
getting
structured.
G
So
this
is
the
architecture
of
our
agent
for
tracing
espn,
core
application.
So
here
we
are
using
the,
and
here
we
inject
injected
tracing
handlers,
which
is
the
the
hooking
method
through
profiling,
api
into
user's
application
and
when
user's
application
is
running,
the
diagnostic
listeners
will
capture
all
of
these
metrics
informations
and
our
x-rate
on
sdk
will
further
generate
trace
context
and
listen
to
daemon.
The
daemon
will
further
send
this
trace
context
to
our
x-ray
back-end.
G
So
this
is
our
mechanism
object
of
using
provider
and
our
auto
instrumentation
to
automatically
instrument
the
user's
estimate
core
application
in
the
runtime
and
so
for
sp
donate.
We
have
like
a
differentiation
mechanism
for
for
it.
We
don't
use
provider
in
this
case
and
we
are
just
using
the
http
module.
So
we
use
an
hp
module
and
install
this
http
module
into
user's
application.
G
So,
on
this
hp
module,
we
initialize
the
http
diagnostic
listeners
and
circling
one
listener
and
the
entity
framework
core
entity,
framework
command
interceptor
and
we
further
inject
this
tracing
module
into
user
applications
so
that
user
application
can
get
like
automatically
instrumented.
G
G
And
then
our
extra
extraordinary
sdk
will
capture
this
handling
generators
contacts
and
then
send
to
x-ray
team
and
then
further
to
extra
back-end.
G
Yeah,
so
that's
the
general
mechanism
of.
Are
we
using
how
our
agent,
like
trades
for
sp9
and
update.
A
One
question:
you
said
that
not
for
asp.net,
but
you
you
inject
the
profiler
via
a
managed
call,
or
I
misunderstood,
that.
F
So
then,
to
clarify
in
asp.net
applications,
then
you
just
add
that
instead
of
making
that
call
by
modifying
io,
you
just
make
that
call
on
your
when
you
initialize
your
module.
That's
the
only
difference.
Okay,.
G
G
Yeah
so
as
we
are
only
targeting
the
beta
version,
so
there's
not
more
for
us
to
do
for
the
ga.
So
currently
it's
only
like
windows
compatible.
We
have
to
make
it
comfortable
to
run
on
the
next
for
the
ga.
Also,
if
users
want
to
use
our
agent,
they
have
they
have
to
reference
the
opening
strategy
library
so
that
our
provider
can
find
the
hooking
method
and
the
hpv
module
can
initialize
the
interceptors
for
the
application.
G
So
and
this
day
we
are
trying
to
remove
this
step
for
the
ga,
because
we
want
the
agent
to
come,
pre-install
this
assemblies
and
its
dependency
of
symbols
for
customers
so
that
they
don't
have
to
reference
just
on
instruction,
never
anymore
and
yeah.
So
this
is
the
feature
we
are
supporting
so
far
for
using
our
agent,
so
we
do
support
the
espn
experience
core
and
http
and
sql
client
and
entity
firmware
core.
The
database
request.
A
Regarding
the
profiler,
do
you
disable
any
optimization
like
in
line
or
there
is
any
specific,
because
this
is
one
of
the
things
that
we
always
look
when
we
look
at
the
profilers.
A
So
do
you
disable
optimizations,
like
jitting
line
engine
is
typically
disabled.
This
kind
of
thing
when
you
inject
the
profiler.
G
So
far,
we
don't
support
that
the
provider
is
very
straightforward
and
simple.
Just
inject
this
method.
C
So
so,
basically-
and
please
correct
me
if
I'm
wrong-
but
this
basically
means
that
essentially,
this
is
the
this
deals
with
the
supports
libraries
who
have
already
like
activities
built
in
and
if
a
library
does
not
instrument
itself
with
activities.
It's
currently
not
yet
supported
right.
All
right.
A
A
So
for
aws
the
the
users
in
this
case
they
they
yet
are
read
the
the
profile
of
their
app
on
their
vms
or
they
have
to
to
do
that
setup.
G
A
I
mean
like
if
I
was
outside
aws
and
somebody
wants
to
install
a
profiler
they
they
need
to
basically
configure
the
vm
and
do
that
that
work
do
does
aws
red
provides
these
in
the
in
their
dotnet
images.
G
G
J
I
haven't
looked
into
this
too
much,
but
with
dotnet
core
there's
something
called
startup
hooks
and
it
may
not
serve
the
purpose
that
you
need
here,
but
my
understanding
of
them
is
that
it
enables
the
ability
for
a.net
core
application
when
the
clr
is
bootstrapping
it
that
it
allows
some
arbitrary
code
to
run,
which
might
allow
might
be
another
hook
for
getting
your
module
in
place
and,
ultimately,
the
method.
I'm
just
curious,
if
that,
if
that's
something
we
all
had
researched
at.
G
All
we
haven't,
so
we
haven't
gone
that
far
as
well
as
targeting
the
beta,
but
we
definitely
will.
J
A
Yeah,
I
I'll
be
curious
about
that
too.
If
I,
I
think,
noah
left
the
meeting,
but
if
dave
knows
something
about
please
let
us
know.
E
J
Great
thanks
and
just
some
additional
context
from
from
new
relic
in
our
experience
with
aws
that
you
might
be
interested
in,
is
that
we
that
we
have
a
separate.net
product
for
monitoring
lambdas.
J
It's
basically
just
an
sdk
for
that's,
built
upon
open
tracing
that
enables
people
to
basically
wrap
their
methods
for
instrumenting
and
sending
up
data
to.
J
One
of
the
things
that
we
had
originally
kind
of
prototyped
was
just
using
our
agent
running
in
a
lambda
environment
and
we
actually
got
that
working,
but
the
agent
our
agent
product
turned
out
to
be
too
heavy
weight
in
the
context
of
a
lambda
because
of
the
startup
time
required.
J
K
Just
as
a
tangent,
so
I
did
some
work
and
it's
either
three
or
three
one
to
address
the
the
startup.
So
the
issue
with
lambda
is:
if
you
start
up,
then
you
can't
use
one
of
the
pre-warmed
up.
You
know
in
lambda
they
they
have
processes
just
sitting
there
waiting,
and
then
they
insert
your
code
into
it.
K
E
And
do
you
guys
know
what
what's
is
used
in
aws
by
new
relic
in
case
of
java?
Are
you
I
mean
I
I'd
imagine
same
called
startup
time
if
you
would
be
arrested
right
by
using
the
profiler
api?
Sorry
in
java,
it
should
be
whatever
bytecode
stuff
or
are
also
suggesting
sdk
for
now,.
E
Okay,
I'm
asking
like
for
java
keys
in
iwps.
Are
you
using
bytecode
or
are
you
using
sdk
as
well.
J
For
java,
new
relic
also
has
an
sdk
based
solution.
J
E
Yeah
because
we
are
facing
similar
issues
whenever
we
work
on
arrow
functions,
quite
similar
concept
right
and
cold
startup
time
is
critical
for
us
as
well.
So
I
would
love
to
learn
like
what
can
we
do
in
dot
net
core
and
in.
J
J
I
don't
know
if
you
enter
if
you
ever
cross
paths
with
john
watson,
but
he's
he'd
be
a
great
person
to
talk
to
from
new
relic
in
relation
to
java.
E
And
I
wonder
if
from
aws
side
itself,
we
can
do
much
about
call
startup
times
like
on
agent
sites.
You
can
do
rigid
right,
but
maybe
pools
could
be
provisioned
with
some
agents
or
some
I
don't
know
it's
just
something
offered
up
for
discussion.
E
And
from
aws
side
guys,
do
you
feel
any
limitations
that
you
don't
really
use
that
much
of
a
profile
functionality
or
so
far?
This
is
just
a
new
feature
and
you
didn't
get
that
much.
Customer
pressure
for
advanced
instrumentation.
G
Yeah,
so
we
do
very
much
like
too
low,
or
we
can.
You
know
using
the
because
the
the
profiler
needs
to
find
the
cookie
method
right
and
the
customer
has
to
reference
this
library
so
that
the
prophetic
can
find
the
hooky
method.
So
we
are
very
curious
about
if
the
leading
way
that
we
can
say
set
a
path
for
this
assembly
is
for
the
booking
method
and
the
profiler
can
find
not
in
user's
application.
But
in.
E
E
G
G
Okay,
so
yeah,
that's
what
transitions
are
about
and
you
guys
have
any
like
last
questions
cool.
Thank
you,
okay!
Thank
you.
Thank
you
guys.
E
C
C
C
Here
you
go
so
yeah
in
terms
of
agenda.
I
think
it's
kind
of
related
to
the
discussion
we
had
before
before
the
aws
presentation
about
goals
and
what
not
so
I
think
we
can
touch
a
little
bit
more
on
michael's
point,
so
I
think
I
think
a
technical
evaluation
of
what
of
the
all
of
the
new
stuff
is
really
good.
C
Something
that
michael
you
need
to
consider,
though,
is
that
from
microsoft's
perspective.
Of
course,
you
would
like
to
ever,
and
you
should
definitely
invalidate
evaluate
everything,
but
from
the
perspective
of
many
other
people
who
are
here
as
well
as
why
the
community
wanting
to
consider
is
that
people
who
wanted
to
base
themselves
on
non-non-proprietary
text
tech,
they
are
already
running
on
datadog,
so
splunk
has
made
investments
in
datadog
based
based
technology.
C
Well,
we
anyway,
but
then
also
a
whole
bunch
of
other
community
members
for
smaller,
smaller
companies
who
choose
to
to
run
based
on
open
source
technology.
They
are
also
running
on
datadog.
So,
while
from
microsoft's
perspective,
specifically,
it
might
make
sense
to
choose
something
that
right
now
is
most
aligned
to
what
microsoft
needs
right.
Now
you
need
to
consider
that
as
a
group,
we
need
to
balance
effort
on
yeah.
C
We
know
that
right
today,
datadog
has
some
performance
limitations
that
that
that
need
to
be
addressed
in
a
you
know,
and
I
was
a
very
high
priority,
but
for
everybody
else
who
wants
to
use
an
open
source
based
solution,
the
consideration
is
different
than
microsoft,
because
microsoft
didn't
use
anything
in
the
past.
For
microsoft,
is
you
know
which
of
the
possibilities
is
the
fastest
to?
You
know,
address
the
issues
and
start
using
it,
but
for
everybody
else
is.
C
This
is
a
a
consideration
versus
all
the
investments
that
have
already
made
into
an
existing
open
source
technology,
and
so
that
is
something
I
think
it's
good,
that
that
we
take
a
little
bit
of
time
to
look
at
things,
but
it's
something
to
consider
you
know.
So
if,
if
you
conclude
that
you
need,
you
have
now
multiple
open
source
options
and
you
can
get
what
you
need
from
one
of
them
of
them.
C
You
know
a
few
weeks
faster
that
doesn't
mean
that
everybody
else,
because
everybody
else
who
wants
open
source
already
started
using
open
source.
So
that's
that's
something
to
consider,
as
as
as
as
you
look
into
this.
E
Yeah,
I
don't
object
anything
for
what
you
said.
That's
all
good.
What
I'm
saying
is
like,
let's
not
the
past
of
like
to
define
our
future
right.
Let's
make
sure
we
analyze
our
options
properly,
yeah.
J
A
Just
on
the
technical
level,
we
have
something
very
very
core
to
this.
That
is
basically
intercept
the
the
either
or
the
call
the
target,
but
we
want
to
intercept
the
method
in
the
end
and
have
a
way
for
people
to
write
their
instrumentations,
that's
kind
of
supported.
So
I
think
that
there
is
a
different
set
of
integrations,
ready
and
sometimes
even
the
same
integration
for
big
targets
like
asp.net,
but
that
they
have
generate
slightly
different
data.
A
But
I
think
the
point
is
that
we
want
to
it's
gonna
require
effort,
but
I
think
it's
one
of
goal
of
all
of
us
is
kind
of
to
be
welcome
to
anyone
interested
in
working
this
kind
in
this
project
you
know,
so
we
can
satisfy
the
needs
and
the
things
that
aws
have
that
new
relic
has,
and
what
I
think
we
want
is
to
leverage
the
experience
that
people
have
you
know.
So,
whatever
was
learned
in
that
code
base,
if
we
get
the
participation
of
new
relic
and
the
aws
folks,
microsoft.
A
Folks,
we
are
leveraging
out
the
experience
that
this
is
brought
in
together,
on
top
of
whatever
data
dogs
splunk
already
have
with
their
with
the
data
dog
implementation.
You
know
so
in
a
sense
of
course,
we
we
should
look
and
take
that
step
carefully,
but
in
a
sense
in
the
end,
these
must
accommodate
all
those
instrumentations.
You
know
all
those
integrations.
I
mean.
E
C
Yeah
and
one
thing-
and
I
don't
know
whether
ellen
on
the
call-
whether
you
can
answer
it
or
not,
is
also
the
or
maybe
this
is
some
other
folks
on
your
end
who
need
to
answer
this?
I
just
went
through
this
process
myself
on
data
dock
side.
I
I'm
an
engineer,
so
I
was
thinking
okay
well,
our
stuff
is
open
source.
Can
I
just
go
press
a
button
and
it
becomes
and
merge
it
and
show
open
telemetry
turns
out.
C
No,
there
was
a
whole
process
on
the
back
that,
where
our
product
guys
worked
with
with
open,
telemetry
people
board
or
whatever,
and
it
is
apparently
far
from
trivial
and
require
all
sorts
of
legal
and
product
alignment
to
to
make
this
possible
and
new
relic
might
very
well
be
up
for
it.
I
I
don't
know
like
absolutely.
All
I'm
saying
is
that
making
things
open,
source
and
and
doing
like
becoming
base
for
open
telemetry
is
not
the
same.
C
C
So
it
was
a
whole
process
for
us
to
understand
whether
we
can
make
such
commitments
and,
at
the
end
of
the
day,
we're
here.
So
that
is
definitely
a
thing
to
work
through.
I
don't
know
what
the
answers
are,
but
if
you
want
to
consider
this,
then
it's
something
you
can
choose
to
kind
of
address
in
one
way
or
another.
J
Yeah
absolutely,
I
can
definitely
appreciate
that
and
based
off
of
mikhail's
question
last
week
I
actually
did
sort
of
just
poke
and
ask
what
that
might
involve
if
the
community
does
decide
that
they're
interested
in
it
even
like
just
even
a
part
of
the
agent
and
I've
gotten
at
least
some
initial
answers
that
there
should
be
no.
There
should
be
nothing
that
would
block
that,
and
could
it
could
be
streamlined
pretty
pretty
quickly.
J
Just
like
you,
so
so
you
know
yeah,
I'm
I'm
I'm
asking
questions
so
if
people
do
end
up
expressing
any
interest,
absolutely
we'll
do
our
best
to
streamline
it
as
fast
as
possible,
because
I
mean
being
an
engineer
myself,
my
ideal
would
be
like
hey.
You
know,
like
here's,
an
idea
from
such
and
such
project,
you
know
whether
it
be
new
relic
or
not.
C
Yeah
this
is
great
and
thank
you,
so
I
think
I
think,
on
on.
I
would
let's,
let's
time,
boxes
to
the
the
coming
week
and
decide
next
week
we
can
yeah.
We
can
close
this.
I
think
for
for
now
part
of
the
for
now
we're
just
going
to
I
don't
know,
I
don't
think
we
need
to
decide
anything
right
now,
paulo.
What
would
what
do
you
think.
A
I
I
I
think
we
we
don't
need
to
decide
right
now,
but
I
I
I
agree.
We
should
time
box
and
not
only
time
box,
but
we
should
say
what
what
are
the
nexus
steps
then.
So
we
can
come
to
a
decision
and
kind
of
move
ahead.
You
know
so
are
we
gonna
looking
more
details
into
the
the
new
relic
are
gonna?
Do
some
comparison?
A
Are
you
gonna
bring
something
kind
of
comparing
this
to
or
you
know
so
you
need
to
be
concrete.
C
E
As
I
said,
I'm
just
learning,
what's
the
open
source
right,
so
I'm
trying
it
out
I'll,
see
specific
details
on
what
optimizations
have
been
turned
off.
What
rigid
current
story
is-
and
I
already
know
from
a
previous
presentation
that
new
relic
is
instrumenting
target
method
rather
than
call
side
which
is
helpful
right
and
I'll
just
see.
If
I
find
anything.
E
E
Let
me
I'll
try
to
start
initial
comparison.
E
And
we
can
discuss
it
next
week,
like
whatever
dance
disadvantages,
and
if
we
enough
to
use
new
relic
versus
data
and.
E
As
I
said
like
I,
don't
know
what
I
don't
know
today,
so
I'll
be
investigating
I'll
post.
My
findings
whenever
I
find
something
into
the
guitar
channel
so
feel
free
to.
E
A
I
would
even
propose
that,
as
soon
as
you
do
that
we
try
to
schedule
a
meeting
with
the
parties
that
are
really
interesting
participate
of
the
conversation,
so
we
can
speed
up
this
process.
A
You
know,
I
think,
from
my
perspective,
I've
been
trying
to
get
this
started
since
april,
and
the
four
reasons
that
I
discussed
in
that
document
kind
of
it
took
up
a
bit
more
time
than
that,
and
I
think
that
my
goal
from
from
splunk's
perspective
in
open
telemetry
is
that
we
really
do
want
to
have
a
open,
telemetry,
auto
instrumentation,
offering
and.
B
A
Would
like
to
avoid
long
delays?
That's
my
big
concern!
So
as
soon
as
you
are
able
to
kind
of
publish
that
in
the
guitar
channel,
we
should
try
to
schedule
a
meeting
even
before
the
next
week.
Two
weeks
meeting.
C
Yes,
I
agree,
and
I
think
to
ellen's
point
as
well,
so
either
way
whatever
basis
we
choose,
I
think
we
will
all
our
own
joint
interest
is
to
have
the
open,
telemetry
think
as
as
good
as
possible.
So
I
think
us
wanting
to
move
forward.
C
I
would
like
you
know
either
way
if,
if
we
go
with
the
basis
we
have
today-
and
we
discover
that
there
is
some
possibilities-
some
things
that
we
want
to
copy
from
the
new
relic
folks
for
code
base
into
open
telemetry,
then
I
I
really
appreciate
ellen's
and
new
relics
offer
who
will
and
it
shouldn't
take
a
long
time
since
the
fox
know
the
code
base.
C
We
know
the
code
base,
we
can
do
this
relatively
quickly
and
I
just
would
like
to
be
in
a
board
where
we
are
all
having
a
open,
telemetry,
sync
and
contributing
and
working
off
of
it
rather
than
having
another,
more
sort
of
planning
discussions
and
so
on.
I
think
we
in
terms
of
the
things
that
you
called
out
like
the
the
target
method-
instrumentation,
that's
really
a
no-brainer,
so
this
will
happen
anyway,
regardless
of
what
we
take
there.
C
This
will
be
addressed
since
we
all
talk
about
it,
so
it's
important
for
all
of
us
and
then
the
other
things
will
also
be
eventually
addressed
according
to
priorities
right.
So
I
think,
once
we
got
going
working
together
with
ellen's
team
to
take
a
certain
feature
and
massage
it
and
to
and
put
it
into
open
telemetry.
What
do
I
shouldn't
be
a
big
deal
for
the
engineers
you
know
once
once
we
move
forward
there,
so
I.
E
Agree
please,
all
I'm
saying
is:
let's
make
sure
that
we
are
open
enough
right,
so
that
new
relic
and
our
companies
can
participate.
E
I
know
that
both
plunk
and
data
docker
data
dock
agent
based
heavy
today,
so
let's
not
sway
it
out
and
let's
not
wait
to
break
we're
coming
into
your
photography
right.
Let's
make
sure
what
we
are
driving
forward
in
a
common.
C
C
Think
we're
all
I
think,
we're
all
agreeing
we're
all
open.
I
I
just
I
just
don't
wanna
like,
as
I
said,
I
just
don't,
wanna
wait
for
for
something
that
is.
E
That
look
has
been.
E
E
And
it
doesn't
have
any
data
documentation
anywhere
right.
Does
it
have
any
data
doc
mentioned
anywhere?
Oh.
C
It
has
it
has
locked
in
everything
else.
It
has
lots
of
them
like,
as,
as
we
discussed
in
the
roadmap,
we're
creating
a
fork
without
actually
not
having
a
github
fork,
but
just
you
know
a
a
a
git
fork
so
to
say,
since
right
now
we're
not
going
to
ship
it,
we
will
all
work
from
the
common
code
base
and
cross
merge
and
then,
when
we
get
close
to
to.
A
Yeah
there
is
a
rational
for
that
in
the
road
map.
We
want
to
kind
of
collaborate
and
we
want
to
avoid
kind
of
cosmetic
changes
in
this
very
beginning.
That
makes
impossible
to
walk
with
the
two
code
bases
concurrently,
so
we
want
to
evolve
the
things
and
do
this
kind
of
rebranding
cleanup
of
namespaces
api
is
kind
of
in
a
very
deliberate
manner.
A
C
Right
now,
people
who,
like
right
now,
we
essentially
again
consider
the
fact
that,
from
the
community
standpoint,
we've
been
driving
that
open
source
base
and
people
have
been
taking
integrations
so
right
now
it
has
this
particular
layout,
and
we
want
to
continue
doing
this
for
open
telemetry,
but
having
continuous
merges
with
just
renamings
will
slow
everybody
down.
So,
as
we
come
close
to
actually
shipping
open
telemetry
efficiently,
then
of
course
this
needs
to
be
cleaned
up,
but
right
now
this
would
be
just
making
every
merge.
E
I
I
see,
I
think
we
didn't
use
this
approach
with
our
dot
net.
Sdk
right.
Do
you
know?
Do
you
remember
any
particular
reason
why
we
choose
not
to
put
any
code?
Was
it
a
requirement
of
open
telemetry,
seek
to
make
sure
that
there
is
no
attributions
like
comedy
0?
Do
you
know
any
details
for,
or
you
should
exactly.
A
Eventually,
we
need
to
transition,
headers
and
notes
is
to
the
appropriate
thing
for
a
patch
license,
but
we
are
not.
As
far
as
I
know,
I
I
can
double
check
that
I
checked
this
some
time
ago.
We
are
not
required
to
do
that
at
the
very
start.
You
know.
A
I
I
would
say
that
whenever
we
make
available,
then
is
when
we
need
that
to
be
to
be
really
done.
You
know
we
need
that.
We
will
need
the
proper
notices
for
any
copyright
and
this
kind
of
thing
and
also
the
surface
api
type
names
file
names,
everything
you
know.
E
So
it
will
be
a
big
change
potentially,
but
I
can
change
for
companies,
and
you
guys
would
love
both
plank
and
network,
would
love
to
defer
it
for
what
exact
reason
just
not
to
change
your
code
now
and
you're
ready
to
change
it
later
or
what's
a
reason
not
to
do
it
now.
A
Yeah
for
from
from
my
perspective
that
that's
the
way
to
go
because
I
think
I
think
the
collaboration
is
going
to
be
easier
fro
for
if
I
took
only
splunk
side,
I
would
say:
hey:
let's
do
it
now,
but
then
the
collaboration
with
datadog
becomes
very
hard.
You
know,
so
I
think,
and
even
in
the
interest
of
the
project,
let's
say:
zach
goes
and
fix
a
bug
after
we
have
the
initial
merge.
A
E
I
don't
think
we
will
be
doing
this
all
right,
so
we
open
telemetry
we'll
have
to
do
this
name
anyways,
so,
like
all
the
interesting
parties
who
want
to
use
it
we'll
have
to
upstream
it
anyway.
So
this
argument
of
some
particular
with
making
change
and
we
need
to
compatible
with
party
repo
will
not
be
applicable.
A
E
A
A
I
bet,
if
we
put
the
code
there
and
don't
provide
this
guidance,
the
first
we
are
going
to
get
a
bunch
of
pr's
to
really
just
simply
rename
a
bunch
of
things
that
are
not
going
to
bring
us
close
to
our
end
goal,
but
are
going
to
be
there.
So
it's
more
a
matter
of
guiding
and
providing
the
the
information
when
we
are
going
to
be
doing
that.
C
And
there
is
another
thing,
so
exactly
what
paula
said,
plus
one
additional
thing,
michael,
you
need
to
consider
this
from
from.
Like
again,
microsoft
is
just
interested
to
take
something
that
is
nice
in
this
case
which
no
problem
but
consider
this
from
our
perspective
and
splunk's
perspective,
because
we
already
have
something
and
for
us
there
is
two
two
considerations
to
balance.
C
First,
we
want
to
drive
and
con
and
and
promote
and
contribute
to
the
open,
telemetry
project
and
community,
but
also
we
have
customers
who
need
to
be
satisfied
right
now.
So
we
have
a
limited
number
of
resources
and
consider
this
from
from
a
product
owner's
perspective
right,
we're
all
ingenious.
We
all
want
to
do
the
right
thing
and
we
all
wanna,
you
know,
do
the
right,
open
sourcing,
but
consider
this
from
a
product
owner's
perspective.
C
If,
when
you
choose
this
right
and
we
have,
we
decided
to
go
and
all
in
on
open,
telemetry
and
deal
with
this,
it
means
that
every
time
we
are
spending
time
on
open,
telemetry
we're
not
spending
time
on
our
customers
and
strategically
it
may
be
the
right
thing.
Tactically
though,
it's
where
or
it
just
means
we're
not
spending
time
on
our
customers
and
understand
that,
for
you,
it's
different,
you
you,
at
microsoft,
you're
very
far
from
this
customer
you're
you're
mainly
doing
thinking
much
more
strategically.
C
Now,
if
we
start
suddenly
er
at
the
in
the
very
early
phase,
if
we
take
hours
and
hours
on
a
code,
merge
that
could
take
minutes,
we
simply
start
doing
doing
less
code
mergers,
because
we
have
limited
amount
of
time
that
that
we
can
contribute
right
and
suddenly
we
have
now
forced
this
code
base
and
who
is
actually
making
it
better
right.
C
So
so
right
now
there
is
a
whole
sizable
team
who
is
keen
to
essentially
continue
merging
all
of
what
we
do
into
open
telemetry
for
a
indefinite
period
of
time
and
unless
something
majorly
changes
right.
But
what's
that,
like.
E
You're
talking
about
like
when
that
doctor
will
be
able
to
migrate
to
this
open
to
limit
attributes
so.
C
Yeah
then,
most
likely.
C
C
E
C
We
certainly,
we
certainly
couldn't
delay
shipping
because
of
this
right.
Yes,
but
I
think
talking
about
timelines
specific
time
between
right
now
and
shipping.
Is
I'm
not
sure
that
we're
already
at
that
point
to
talk
about
that?
I
think
we
definitely
would
like.
C
E
C
No,
no,
the
shipping
is
it's,
not
our
shipping
timeline.
The
shipping.
The
shipping
timeline
is
ours
here
to
to
decide
like.
We
are
actually
I'm
very
easy
when
we
decide
shipping,
if
we
decide
to
ship
it
right
now,
that
would
only
mean
that
we
have
to
do
this
branding
work
sooner
than
we
thought.
C
So
I
wanna,
I
wanna,
see
the
shift
as
soon
as
possibly
possible,
and
this
is
why
I'm
sort
of
cognizant
of
timelines.
I
just
wanted
to
say
that
talking
about
killing
the
time
right
now
is
too
early,
because
we
just
don't
know
when
we
plan
to
ship.
E
Okay,
let's
postpone
this
conversation,
I
still
like
two
weeks
from
now.
Next
week
we
will
be
discussing
new,
radical
data
data
and
but
I
got
your
perspective
on
data
documents
and
all
I
got
your
effects
for
fun
plants
and
yeah.
Let's
come
back
into
victories
and.
C
C
Hey
a
completely
different
matter,
michael
I
sent
you
a
bunch
of
emails
was
a
question
about
something
unrelated
related
to
azure
app
services.
It
just
I
didn't
get
an
answer,
so
I
just
was
just
checking
you
didn't
have
the
time
to
look
at
it
or
you
didn't
get
it.
E
C
I
I
I
have
another
meeting
after
this,
but
send
me
a
text
or
like
we
can
chat
later
on
the
phone.
It
should
be
a
five
minute
conversation
and
then
we
can.
C
A
Anything
else
guys
so
just
just
just
to
to
say
that
to
reinstate
that
thing,
if
you
have
the
things
ready
as
soon
as
you
have
read
put
on
the
guitar,
if
possible,
let's
try
to
have
a
meeting
before
so
we
can.
If,
if
the
discussion
takes
longer,
we
can
kind
of
be
sure
that
we
are
limiting
the
time
that
we
keep
discussing
this.
E
J
Here,
I'll,
just
I'll
just
drop
my
email
into
the
chat
here
and
you
can
reach
out
to
me
and
then
I'll
make
sure
to
include
the
right
individuals.