►
From YouTube: OpenTracing Spec Council Monthly Call - Sep 1 2017
A
But
I'm
just
going
to
go
down
the
agenda
here
feel
free
to
add
things
at
the
bottom,
as
you
think
of
it,
starting
up
at
the
top.
Do
you
have
any
conference
or
other
marketing
updates?
I
didn't
have
anything
to
put
in
here,
but
I'm
wondering
if
anyone
else
has
any
meetups
or
conferences
were
they
going
to
talk
about
open
tracing
they'd
like
to
mention
I.
B
C
C
C
D
E
D
To
mention
that
for
December
in
in
Austin
for
called
native
conference
the
so
the
de
luxe
I
mean
asking
for
some
open
tracing
related
or
tracing
related
talks,
but
also
their
suggestion
to
have
a
track
sort
of
like
a
salon
or
work
shop.
Anyone
is
planning
to
to
be
there.
You
know,
you've
been
okay,
so
I
guess
we
should
be
I
would
probably
just
like
send
them
a
proposal
of.
Do
we
want
worship
or
just
like
just
open
opens
along
with
discussions.
B
B
The
salon
concept,
which
I
believe
we
made
up
that
word
and
they've,
now
stole
it
and
used
it
for
all
of
their
Frederick's
but
I
like
the
idea
of
it
not
being
like
a
talk,
but
you
know
something
where
people
can
come
in
and
discuss
things
and
so
on.
I
like
I
enjoy
I've
always
enjoyed
those,
but
and
it's
not
that
hard
to
organize
which
is
sort
of
a
nice
property.
But
if
someone
wants
to
like
take
the
lead
on
on,
you
know
publicizing
a
CFP
and
reviewing
proposals,
and
so
on.
B
D
B
Yeah,
it's
speaking
of
khoob
khun.
I
think
that
we'll
end
up
doing
some
kind
of
I
think
there's
gonna
be
some
kind
of
he
noticed
thing
talking
about
open
tracing
and
service
mesh
stuff,
it's
a
little
bit
vague,
but
I.
That
seems
to
be
happening
as
well.
There
should
be,
which
would
be
a
nice
opportunity
to
tell
an
interesting
story
and
everything.
A
Moving
on
specification
updates
is
a
pretty
big
backlog
of
open
issues
and
specification.
I
apologize
I
have
not
been
triaging
that
spring
cleaning
will
come
to
that
next
I've
been
focused
on
getting
all
these
api's
at
the
door,
which
we
can
talk
about
after
this,
but
the
one
I
will
just
throw
over
the
wall
later
today.
It's
just
a
simple
update
to
the
language
of
or
PR
against,
the
spec
to
add
a
concept
of
incubating,
supported
and
deprecated
as
a
lifecycle
for
tags.
A
C
A
Yeah,
that
is
an
interesting
question.
I,
don't
know
if
there's
any
like
what
the
broad
guidelines
should
be
on
that,
because
I
think
in
generally
speaking,
tags
get
created
for
a
specific
purpose,
especially
if
they
have
some
kind
of
functionality
associated
with
them,
and
that
was
the
reason
why
we
needed
to
standardize
on
them
in
the
first
place.
A
I
think
there's
some
case
where
extending
a
tag
that
already
exists
is
good,
because
it's
extending
the
existing
functionality
and
improving
on
it
I
think
there
are
other
cases
where
we
go
to
add
a
concept
to
open
tracing
and
just
kind
of
look
around
and
we're
like.
Oh
this
fans
kind
thing
looks
kind
of
similar,
let's
just
attack
another
kind
into
that,
but
spandy
kind
was
designed
specifically
for
a
kind
of
Zipkin,
like
use
case,
of
combining
server
and
clients
bands
together.
A
That's
the
main
thing
not
talking
about
a
specific
issue
you
raised,
but
that's
the
main
thing
I
think
to
keep
in
mind
when
we're
trying
to
evaluate
if
something
should
be
a
new
additional
tag
or
just
a
more
functionality
on
the
same
tag.
Okay,
I
have
any
skin
in
the
game
on
the
sampling
bit
tag
myself.
A
Okay,
no,
but-
and
maybe
we
can
write
up
some
of
that
I
do
think
we
should
come
up
with
some
guidelines
we
put
somewhere
about
the
process,
for
you
want
to
add
a
tag.
What
you
know
should
you
have
this
part
of
your
proposal
when
you
go
to
do
that?
What
should
you
have
already
evaluated
and
what
should
other
people
evaluate
when
we
kind
of
triage
it?
So
that
is
definitely
interesting
discussion
to
have.
We
should
have
them
get
her.
A
Anything
else
I'll
be
going
through
these.
These
specification
issues
next
week
by
the
way
and
things
that
seem
like
they
need
highlighting
your
discussion.
I'll
be
I'll,
be
pointing
them
out
and
get
er
apologize
for
not
doing
it
already.
I
feel
like
the
API
is
in
particularly
the
Java
stuff.
I've
been
just
mentally
focused
on
trying
to
get
all
of
that
over
the
wall,
because
it
seems
like
that's
the
area
where
we
need
motion,
because
from
blocking
some
other
things,
if.
A
A
My
big
question
was
sort
of
around
like
how
do
we
vote
on
some
of
these
things
and
talking
to
IETF
people
I,
don't
think
we
want
to
adopt
their
full
process,
but
I
did
like
the
idea
of
something
gets
moved
forward
by
in
number
of
you
know
implementers
implementing
it
and
that's
just
how
something
moved
from
one
step
to
the
other,
rather
than
trying
to
like
take
a
vote
or
something
like
that.
There's
probably
a
place
in
here
where
the
the
OTS
see
perhaps
takes
a
vote,
but
more
just
voting
with
your
feet.
A
If
enough
people
implemented
the
new
java
thing,
we're
running
it
in
public
somewhere,
instrumentation
has
been
ported
over
people.
Aren't
complaining,
then
great.
Let's
pull
the
trigger
on
it,
so
that's
the
aspect
of
it.
I
would
like
to
start
following
next
with
Java,
so
I
have
asked
a
couple
of
different
groups
so
far,
but
I
want
to
start
asking.
You
know
more
broadly
people
on
this
call
who
is
willing
at
this
time
to
bind
to
one
of
these
new
api's
a
Ben
siegelman
has
a
PR
out.
A
A
A
A
A
So
you
can
look
at
the
examples
there
and
try
to
extend
them.
That's
a
great
way
to
kind
of
give
this
API
from
an
instrumentation
standpoint
and
see
whether
it's
missing
something
or
not.
Well,
that's
that
I
would
say
is
the
first
step
and
so
I
think
it
would
be
great
if
people
check
that
out
and
and
pushed
on
that
and
gave
feedback
in
the
feedback
in
those
PR.
A
Those
long-winded,
PR
threads
by
you
know,
adding
something
to
Java
examples
and
then
coming
back
in
there
rather
than
just
posting
snippets
at
this
point,
so
that
would
be
the
first
task
and
then
the
next
ass
is
we
want
to.
We
want
to
create
a
release
candidate
out
of
one
of
these
things,
and
so
I
just
want
a
commitment
that
people
will
fork
their
their
tracer
and
actually
modify
it
to
work
with
this
API
just
whatever
bridge.
A
B
B
A
Yeah
and
that's
more
I,
where
I
I
expect
us
to
hit
the
rubber
meeting.
The
road
is
less
binding
to
this
new
API
I
presume
that
that
is
easier
than
the
old
one,
as
there's
less
stuff,
as
Ben
said,
but
rather
when
you
go
to
instrument
something
more
complex
and
real
or
you're
going
to
like
discover
that
it's
too
simple,
so
that's
if
we
hit
any
bumps
I
think
it
would
be
on
that
end
that
it's
too
simple
if
it's
too
complicated
but
I
also
have
a
lot
of
faith
in
in
this
new
one.
A
It's
I
think
more,
and
this
is
also
I
want
to
reach
out
to,
hopefully
in
Stannah
and
other
tracers,
here's
some
of
the
grid
we're
claiming
they
had
we're
not
using
thread
locals,
but
we're
using
some
other
mechanism
under
the
hood
and
for
them
the
having
the
ref
counting
baked
in
ways
and
other
things
like
that
were
problematic
for
them,
because
it's
sort
of
fighting
you're
doing
internally
I
think
grading
goes
a
lot
internally,
for
example,
so
I'm
those
are
the
tracers
in
particular.
You
know.
A
D
A
All
right
there,
any
other
Java,
related
discussions,
I,
don't
think
I
want
to
hash
out
the
continuations
here
on
the
call.
But
again
just
one
final
shout
out:
please
check
out
that
Java
examples,
repo
and
think
about
this
use
case
around
continuations.
That's
that's
really
where
we
need
thought
right
now.
I
wanted.
D
Back
I
heard
about
Java
examples:
is
it
has
no
documentation,
so
you
kind
of
have
to
figure
out
from
the
name
of
the
directories.
So
I
was
thinking
if
it
would
make
sense
to
have
like
a
readme
at
the
top
and
maybe
maybe
like
some
I,
don't
know
either
time
sequence
or
some
sort
of
like
visual
diagram,
saying
this
is
the
use
case
that
you
can
kind
of
understand
quickly.
What's
going
on,
I
don't
know
easy.
It
is
because
the
I
haven't
gone
through
all
day
use
cases
there
yeah.
A
Hopefully
that
all
the
test
cases
are
fairly
readable
but
yeah.
Definitely
one
helpful
form
of
PRS
that
repo
is
documentation
so
and
also
some
documentation
about
how
to
to
fork
it.
You
know
if
you
want
to
try
a
different
API
or
try
something
new.
You
know.
What's
the
approach
there,
it's
simple,
but
then
right.
Yes,
some
hope
there
would
be
good.
I
was.
D
Actually
surprised
by
the
number
of
examples
that
there
was
too
high
to
to
me
to
even
go
through
it
once
yeah
victim
of
our
own
success.
Another
thing
is:
does
this
repo
needs
to
be
separate
from
from
the
main
open
trace
in
Java,
because
people
doing
like
API
changes?
It
would
be
nice
to
have
the
examples
in
the
same
project.
A
C
A
A
Yeah
and
I'm
excited
about
these
examples
getting
cleaned
up
enough.
We
can
turn
them
into
like
a
cookbook,
that's
more
of
a
a
blog
post
kind
of
a
thing
that
you
read
pointing
out
how
to
do
some
of
these
things,
like
you,
know,
walkthroughs
and
stuff
of
that
nature.
So
a
lot
of
rich,
rich
material
I'm
out
of
this
work
and
I
really
like
the
idea
that
all
of
that
work
would
have
something
testable
sitting
down
at
the
bottom
of
it.
Runnable.
A
B
The
main
question
is
actually
about
brave
and
Zipkin,
and
I
think
that
they
were
frustrated
with
the
previous
significant
change
to
ot
java
and,
I
think,
would
be
advisable.
It's
like
directly
engage
with
well,
potentially
Adrian
were
you
know,
I
think
Jake
has
done
a
bunch
of
work
on
that
and
I
actually
don't
think
I
mean
I
think
that
the
either
of
the
PRS
that
are
outstanding
right
now
with
simplified
brave
considerably
so
I
imagine
they
would
be
pleased
with
them.
But
I
would
like
that
to
be
a
smoother
process
than
the
last
one.
A
A
Great
yeah,
and
also
those
there's
akka
cats,
Ivan
I,
forget
his
last
name,
but
they
also
had
some
interest
and
in
Stannah
I,
think
was
another
tracer.
That
was
perhaps
funnily
shaped
under
the
hood
and
had
interest
who
didn't
speak
up
a
lot
last
time
but
seemed
like
it
would
might
be
good
to
sort
of
engage
with
them
just
to
let
them
know
about
this,
since
there
were
a
little
quiet
last
time,
but
to
do
bind
to
this
API
and
use
it
all
right.
A
A
A
These
sidecars
and
sort
of
front-end
proxies
so
envoy,
ninja
necks
are
the
two
we're
actively
working
on
over
here
at
light
step
and
we'll
try
to
engage
people
on,
but
if
other
people
want
to
check
out
check
out
that
API
and
start
using
for
something
that
would
be
great
and
that
sort
of
dovetails,
maybe
with
the
sidecar
sidecar
proposal
stuff
that
you
assumed
this
is
you
been
who
put
on
the
calendar?
Do
you
want
to
talk
about
that
for
a
minute
yeah.
B
For
a
while
and
I
have
a
an
actual
document,
that's
like
a
proposal
that
I
sent
to
them,
both
since
without
their
consent,
this
is
just
totally
not
going
to
happen,
but
they,
thankfully
both
seem
pretty
happy
with
the
general
requirements
and
direction
that
I
described
so
I'm
going
to
hopefully,
either
this
weekend
or
early
next
week,
I'll
I'll
send
it
out
to
get
her
and
put
it
somewhere
else,
but
it's
basically
a
proposal
to
to
where
it's
not
really
a
proposal.
B
It's
a
set
of
requirements
for
through
getting
like
vendor-neutral
tracing
into
cars
and
primarily
service
mesh,
although
it
would
apply
to
h8
proxy
and
nginx,
potentially
as
well.
I
think
the
problem
right
now
is
that
those
things
are
like
it's:
it's
not
reasonable
to
expect
people
who
use
service
mesh
technology
to
recompile
their
service
mesh
every
time
they
want
to
deploy
something,
and
yet
that's
the
only
way
to
actually
switch
tracers
right
now,
unless
everyone
like
sends
PRS
and
gets
their
own
trace
or
baked
into
those
those
service
mesh
things.
B
So
it's
it
does
present
some
interesting
design
challenges,
but
I
I
think
it's
like
a
very
like
doable
thing.
If
we're
willing
to
like
take
a
opinionated
stance
on
a
standard
that
we
don't
necessarily
have
to
write
ourselves
like
we
could
adopt
one
of
the
other
ones,
but
there
does
need
to
be
some
like
on
tracing
opinion
about
like
the
right
way
to
propagate
trace,
span
context
between
processes,
as
well
as
the
right
way
to
represent
on
span
data
like
actual
span
records
as
they're
flushed
out
of
something
like
a
service
mesh.
B
So
anyway,
it
I'm
not
sure
what
we
should
do
in
terms
of
the
the
right
six,
but
the
requirements,
I
think
are
pretty
clear
and
I
do
think
it
would
be
really
beneficial
in
terms
of
making
tracing
like
a
easier
thing
to
integrate
if
we
can
get
into
service
mesh,
so
I'll
send
that
out
this,
you
know
this
weekend
or
next
week.
I
think
it's
definitely
lower
priority
like
the
Java
stuff
and
everything
and
it's
not
gonna,
be
like
a
week-long
effort
like
it's.
B
G
Ben,
this
is
a
Steve
day.
Can
you
hear
me
so
this
is
kind
of
a
huge
problem
for
us.
As
far
as
integrating
open
tracing
into
projects,
I
took
a
crack
at
it
with
container
D
the
other
day,
and
the
data
was
great
right.
It
was
really
cool
because
we
have.
We
have
a.
We
have
about
three
kind
of
two
levels:
deep
of
G
RPC
services
and
it's.
We
have
a
client
in
a
gr
PC
service
that
dispatches
to
another,
for
the
traces
are
really
interesting.
G
But
integrating
open
tracing
is
problematic
because
we'd
have
two
vendor
all
of
the
different
libraries
and
they
so
like,
like
Jaeger,
for
instance,
had
it
had
a
massive
amount
of
dependencies,
so
just
just
having
one
of
those
in
there
or
like
one
implementation
in
there
would
increase
the
number
of
dependencies.
So
if
we
could
find
a
way
to
make
it
simpler
to
integrate
it,
we
can
have
more.
G
B
A
Go
I
think
does
have
cleaner
ways
of
sort
of
binding
in
to
pre-compiled
pieces,
so
I
think
in
C++.
This
is
more
onerous,
there's
potentially
some
ways
forward
and
go
that
are
sort
of
a
middle
ground.
Since
you
mentioned
container
D
where
we
might
be
able
to
sort
of
sidestep
some
of
these
things.
But
it's
still
going
to
be
some
extra
thing
that
you
have
to
install
or
package
up,
which
is
not
great.
F
G
Yeah
we
experimented
with
go
plugins
a
little
bit
and
the
distribution
model
is
onerous
to
state
police.
You
basically
have
to
compile
the
plug-in
for
every
single
for
every
single
version
of
go
that
you
target
and
every
single,
oh
wait.
No
every
single
version
that
you
distribute
you
have
to
compile.
The
plug-in
was
to
target
them
directly.
So
it's
not
that
much
better
than
having
to
compile
them
in
statically.
A
Yeah,
this
is
the
worst
and
yes,
Chris
alluded
to.
Like
my
experience
with
Merson
talking
to
people,
no
one
wants
to
recompile
these
things
ever
like
that's,
not
a
thing.
People
really
want
to
do.
Even
if
your
thing
is
shiny,
they
don't
want
to
do
it,
so
something
standard
would
be
better.
Yeah
cross
recompiling
is
scary,
yeah.
It's
also
related
to.
If
people
have
some
kind
of
like
pipeline
of
you
know,
tested
stamp
or
vetted.
A
You
know
binaries
that
are
coming
from
somewhere
else,
where
you
know
they're
they're,
getting
security
patches
and
things
you
know
pushed
downstream
to
them.
Having
then
trying
to
insert
it
into
that,
oh
and
then
compile
it
into
some
arbitrary
version.
That's
like
a
thing
like
larger
enterprise
customers.
Don't
don't
like
that
kind
of
thing.
C
F
A
So
yeah
standard
standard
protocol
might
be
coming
our
way,
whether
you
want
it
or
not.
A
There's
also
to
be
clear
that
this
thing,
even
if
you
can
forget
away
to
install
these
as
plugins
it
falls
apart
as
soon
as
you
go
multi-tenant
right,
so
people
aren't
really
doing
multi-tenant
kubernetes
yet
so
much
but
like
that's
coming
I
mean
I
worked
on
the
Cloud
Foundry
runtime
previously,
so
I
know
what
multi-tenant
container
platform
looks
like
and
I
assume,
kubernetes
and
other
things
gonna
get
there
as
soon
as
that
thing
becomes
multi-tenant,
then
what's
your
service
mesh
gonna
install?
So
then
you
really
need
some
kind
of
standard
protocol,
so
it's
somewhat
inevitable.
B
Mean,
like
I,
think
the
timeline
for
sending
out
of
proposals
like
imminent
and
then
the
Bulls
right
for
sending
out
requirements
imminent
I
think
it
will
probably
be
a
little
bit
controversial.
It
touches
on
a
lot
of
things
that
make
people
angry
or
like
that's
it's
just
like
there'll.
It's
like
three
bike
sheds
and
one
kind
of
thing
on.
B
If
we
just
like
wanted
to
do
something,
it's
not
very
difficult,
like
the
tactical
work
is
not
hard
at
all,
but
the
decisions
will
probably
be
it'll
take
as
much
time
as
we
give
it
put
it
that
way
and
I
and
I
think
we
should
wait
until
there's
like
it's,
like
you
know,
microwave
popcorn
or
something
like
wait
until
the
threads
are
sort
of
like
one
one
note
every
couple
of
days
kind
of
thing,
but
I
imagine
that'll
be
a
bunch
of
discussion
about
it.
I
would
like
to
see.
I
would
like
to
see
something.
B
F
A
I
was
actually
talking
to
them
briefly
because
they're
using
the
IETF
RFC
process
for
managing
that
I've
become
a
processing
governance
wonk,
trying
to
figure
out
what
what
flavors
of
this
open
tracing
needs,
but
I
haven't
actually
talked
to
them
about
the
open,
metrics
API
itself,
I
mean
that
looks
interesting,
but
they
seem
to
like
that
process.
I
think
their
thing
is
a
little
simpler
than
ours.
That's
really
the
problem
with
our
process
is
just
like
everything's
like
have
you
tried,
like
versioning
the
API
I'm,
like
which
API,
which
thing?
What
are
you
gonna?
A
A
Okay,
moving
on
I
can
get
my
computer
to
move
on
so
PHP
whoo
that
got
released.
That's
great
I'm,
not
sure
who
who
is
binding
to
it
and
and
trying
to
use
it
and
who's
out
there
instrumenting
all
the
things
in
PHP
now
that
there's
an
official
API,
but
you
should
ideally
move
on
that
sooner
rather
than
later.
Just
in
case
there
is
something
about
the
API
we
want
to
change.
We
don't
want
to
come
people
using
it
extensively
and
then
the
rest
of
us
kind
of
paying
attention
to
it
later
and
being
like.
A
Oh
can
you
change
this
thing,
so
this
is
just
another
shout
out
please,
if
you're
thinking
about
PHP
and
have
a
PHP
tracer,
try
binding
to
this
and
consider
adding
instrumentation
we're
not
super
focused
on
PHP
at
the
moment
and
light
step,
but
we
would
like
to
turn
some
attention
there
we're
just
a
little
short-staffed
in
terms
of
like
PHP.
You
know
sort
of
in-house
know-how.
E
So
we
are
trying
to
instrument
PHP
in
or
the
yes
mano.
Little
tight
for
problem
is
that
now
we
we
started
with
the
with
dated
August
the
Bakken,
but
seems
to
be
pretty
expensive,
so
we're
probably
moving
those
two
SIPC
in
there
is
no
official
SIPC
in
instrumentation
for
PHP.
So
I
am
working
that
now,
once
we
have
these
two
pieces,
I
can
build
the
bridge
and
then
start
instrumentation.
However,
I
saw
some
people
opening
issues
in
the
repository
like
commenting,
and
there
is
a
couple
of
forks.
So
maybe
people
started
doing
this.
A
A
A
The
word
presses
in
the
world
need
you
last
and
certainly
not
least,
Python
Python
API.
We
also
want
to
upgrade.
It's
just
been
a
little
hesitant
to
try
to
charge
forward
with
that,
while,
while
Java
is
still
on
the
flight
deck
just
because
the
api's
are
so
similar,
it
would
be
great
if
they
all
use
the
same
nomenclature
and
everything
else
I
know
Carlos
has
been
and
mano
had
been
working
on
some
Python
examples,
I
believe
in
the
background,
getting
something
together
similar
to
the
Java
examples.
A
Stuff
I'm,
not
sure
where
that's
living
right
now,
so
part
of
the
hope
is
we'll.
Take
that
also
the
process
from
from
Java
here
and
be
able
to
apply
it
to
Python
and
do
that
kind
of
efficiently.
So
you
should
see
Python
examples
and
then
a
new
API
that
you
can
bind
to
you
and
see
if
we
can
have
sort
of
smoother
rollout
of
Python
kind
of
just
stepping
in
the
same
footsteps.
The
Java
took
I,
don't
have
anything
to
report
on
that
front.
A
B
Sounds
like
that's
it
well.
I
have
one
thing:
I
was
waiting
for
someone
else
say
something
but
I,
since
no
one
else
did
I
was
I
was
curious.
I'm
around
the
the
Python
stuff
I,
don't
want
to
well
I'm,
going
to
but
I'm
sorry
I'd,
to
put
people
on
the
spot,
but
I
know
that
data
dog
had
done
a
bunch
of
work
and
that
we're
sort
of
waiting
for
the
Java
API
stuff
to
settle
down
everything.
There
was
also
a
bunch
of
work
that
had
been
done.
B
It
I
think
I
dated
out,
and
maybe
a
few
others
and
adding
instrumentation
to
various.
Like
you
know,
popular
Python
things
like
tornado
and
so
on,
and
so
forth
and
I
was
just
curious
to
like
have
some
sense
of
like
does
anyone
know
how
far
the
instrumentation
is
sort
of
semantically
from
like
the
kind
of
current
open
tracing,
Python
proposals
and
stuff,
like
that
I'm
just
curious
to
understand?
If
there's
much
of
a
gap
there.
C
So
so
I'm
I'm
I
think
I'm
the
only
data
dog
representative
here
right
now,
but
I,
don't
really
I've
been
totally
heads
down
focused
on
the
Java
side
of
things.
I,
don't
know
anything
about
how
the
Python
implementation
works
or
how
that
what
the
plan
is
to
integrate
there.
But
it
sounds
like
you're.
Also
kind
of
asking:
can
the
existing
data
dog
Python
instrumentation
be
kind
of
back
ported
or
made
available
to
the
open
tracing
community?
Is
that
kind
of
what
you're
asking
not.
B
Precisely,
although
it's
a
similar
thing,
I
snore
to
that
I
mean
that
would
be
I,
think
that'd
be
great
and
it
doesn't
even
have
to
be
dated
about
doing
that
work.
But
it's
it's
more
that
you
know
pair
the
conversation
earlier
about
the
value
of
that
examples.
Folder
for
Java,
like
in
some
ways
that
instrumentation
of
things
like
tornado,
that's
about
as
complicated
as
things
are
probably
going
to
get.
So
it
would
be
it.
B
It's
like
a
great
test
bed
to
confirm
that
the
API
is
in
Python
or
reasonable,
even
if
it
was
just
like
a
spike
that
demonstrates
like
yes,
that
the
open
tracing
API,
that's
kind
of
under
consideration,
Python
satisfies
those
requirements.
I
think
that
would
actually
be
a
really
valuable
thing
to
know
and
I'm.
B
Just
what
I'm
doing
in
my
head
is
like,
rather
than
have
all
the
Python
stuff,
totally
blocked
on
the
Java
stuff,
like
again,
not
with
the
intention
of
even
submitting
the
code,
but
just
being
you
know
confirming
that
the
api's
that
are
under
discussion
in
Python
would
satisfy
the
requirements
that
instrumentation
would
be
a
really
valuable,
valuable
extras.
I
have
a
feeling
that
it
has
been
checked
out
so
I
I,
but
I'm
not
positive
at
that's.
What
I'm
asking
them
yeah.
C
A
And
just
just
all
so
to
answer
your
question:
Ben
yeah,
that's!
Actually
what
they've
been
focusing
on
as
far
as
Python
examples
go,
it's
not
just
the
basic
things,
but
actually
most
the
focus
is
bringing
in
all
of
those
various
flavors
of
asynchronous
runtimes
and
and
instrumenting
them.
So
that's
that's
the
bulk
of
what's
getting
getting
tested
over
there.
That's.
B
A
Know
so
that's
great
so
yeah,
it's
an
example
for
each
flavor
and
then
you
know
the
sort
of
final
final
thing.
All
the
way
the
edge
is
like
a
callback
based
tornado
is
the
sort
of
lowest,
as
you
can
go
on
that
totem
pole
of
pain
or
highest,
whatever
it's
the
worst.
So
if
that
works,
then
everything
works.