►
From YouTube: 2020-09-17 Python 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
All
right,
everyone,
if,
if
you
haven't
already
please
feel
free
to
add
yourselves
to
the
list
of
attendees,
looks
like
we
have
a
few
a
few
new
folks
here,
if
you're
attending
for
the
first
time,
it
would
be
great
if
you'd
like
to
go
around
and
just
kind.
C
C
E
B
Okay,
all
right,
all
right:
hey
do
you
want
to
drive
the
meeting
for
a
bit
I'll,
try
and
figure
out
my
audio
issues,
all.
E
E
E
E
E
So
it
looks
like
we
just
had
a
release.
Yesterday,
a
bunch
of
stuff
went
in.
I
think
that
was
a
first
topic
yep
and
there
were
some
pr's
that
didn't
go
in,
but
there
was
a
lot
of
changes
in
the
last
release.
E
E
E
A
amos
law,
that's
how
you
pronounce
it
right,
close
enough,
I'm
amos,
and
I
I
I'm
in
new
intern
at
aws-
and
I
guess
is
my
first
time
at
the
sick
meeting
nice
man
welcome
dude,
welcome.
F
Hey
hi,
my
name
is
nathaniel,
I'm
actually
an
intern.
This
fall
with
the
aws
x-ray
team
and
awesome.
I'm
gonna
nice,
I'm
gonna
be
working
on
just
getting
our
exporter
on
open,
telemetry
working
with
x-rays,
so
that
and
the
other
way
around
too.
So
we
can
export
our
x-ray
into
open
telemetry
format
so
exciting.
E
All
right,
finally,
getting
amazon
to
on
this
on
board
on
time.
E
In
yeah,
just
just
waiting
until
everything
is
has
been
established
already
and
then
they
sneak
in
with
their
cloud
stuff.
Yeah,
cool
cool
cool.
G
Cool,
so
then
I
guess.
Lastly,
we
have
david
yeah
hey,
so
my
name
is
david
ashpole.
I've
been
hanging
around
the
kubernetes
community
for
a
couple
years
at
google
and
I'll
mostly
be
a
fly
on
wall.
So
I'm
just
trying
to
learn
what
you
guys
are
up
to
and
maybe
see
if
there
are
ways
for
me
to
contribute.
A
G
E
E
I
don't
know
I
just
assume
everyone
at
google
knows
each
other,
so
nice,
nice,
awesome
cool,
so
yeah.
This
is
the
open,
telemetry
python's
sake.
My
name
is
leighton.
I
work
at
microsoft.
I
and
alex
are
the
maintainers
of
this.
This
group
and
that's
that's
pretty
much
it
man,
we're
just
trying
to
you,
know,
drag
along
push
push
our
our
language
to
ga
and
hopefully
that
we're
spec
compliant
by
the
end
of
by
the
beginning
of
november.
That's
like
our
our
plan.
E
If
you
guys
want
to
know
more
details
about,
like
the
you
know,
release
like
timeline
and
milestones,
so
you
guys
can
check
out
the
open,
telemetry,
specs
repo,
as
well
as
the
community
and
yeah.
Like
any
other
questions,
you
guys
might
have
just
join
this
gator
group
open
telemetry
python.
Hopefully
you
guys
are
in
it
already
and
like
we're
pretty
active
on
there.
So
any
questions
you
guys
might
have
feel
free
to
go
there
as
well,
there's
a
bunch
of
issues
and
pull
requests.
E
Oh
sorry,
issues
on
our
github
and
we've
marked
a
lot
of
them
with
a
good
first
issue.
You
know
if
you
guys
want
to
try
to
get
your
feet
wet.
We
like
kind
of
triage
these
and
think
that
these
are
pretty.
You
know,
good
problems
to
try
to
tackle
to.
You
know,
try
to
familiarize
your
familiarize
yourselves
with
the
code
base
a
little
disclaimer
right
now,
since
we're
kind
of
low
on
approvers
and
resources
and
time.
E
A
lot
of
our
focus
is
going
to
be
placed
on
these
issues,
with
the
tags
released
required
for
ga,
mostly
because
are
the
main
goal
by
you
know.
By
the
time
the
specs
go,
ga
which
is
was
supposed
to
be.
E
I
think
the
middle
of
the
september,
which
is
like
right
now
is
yeah
or
the
goal
was
to
be
spec
compliant
and,
and
python
is
a
little
bit
behind
on
this,
as
since
we
do
lack
a
couple
members,
but
yeah,
don't
don't
like
be
offended
or
something
if
you
work
on
something
that's
not
required
for
ga,
you
don't
get
that
much
traction
just
like
try
to
ping
it
on
the
the
getter
channel
or
bring
it
up
as
an
issue
during
like
the
sig
meetings
that
we
have
like
today,
cool.
E
All
right,
awesome,
awesome,
and
it
looks
like
the
next
topic,
with
an
incorrect
meeting
link
to
the
calendar,
looks
like
everyone.
E
Found
it
today
pretty
well,
but
I
I
was
sitting
in
the
wrong
meeting
room
with
hector
actually
a
few
minutes
ago,
so
alex
do,
you
think,
is
that
something
that
needs
to
is
that
a
recurring
problem
or
something
like
are
people
complaining
about
this.
B
E
Awesome
cool
all
right,
we'll
we
could
take
a
look
at
that
after
cool,
so
I
guess,
if
you
guys
so
so,
usually
the
formats
of
our
sigs
is
like.
We
usually
go
through
hot
topics
that
people
want
to
talk
about
like
that's
important
and
then
after
that,
if
there's
nothing
that
is
like,
we
need
to
really
address
right
away.
We
go
through
the
open,
pr's
and
issues
that
we
have
in
that
order.
E
So
does
anybody
else
have
any
other
topics
that
they
want
to
bring
up
before
we
go
into
the.
E
Prs
all
right
cool
sure,
so
I
guess
we
can
just
go
through
the
open.
Pr's
then
also
like,
if
you
want
to
like
kind
of
get.
You
know,
visibility
on
your
pr
and
get
more
reviews
on
it.
Now
is
the
time
to
like
add
it
to
this
this
dock
right
here
and
I
actually
should
have
added
some
of
mine
beforehand,
but
anyways.
E
We
have
a
pr
right
here.
It's
pretty
straightforward,
take
a
look
at
it.
Basically,
wait
is
daniel
here
in
this
call.
D
E
E
D
D
Probably
that's
correct,
I
didn't.
I
didn't
really
look
at
the
release.
E
All
right,
awesome,
okay,
cool,
oh
so
yeah,
pretty
straightforward!
I
guess
aaron
could
do
it.
Oh
nice
awesome
looks
like
we
have
another
call
out
here.
Let
me
just
open
this
one:
real,
quick.
C
And
you
want
to
talk
about
this
one
real,
quick
sure
this
one
was
in
here
last
week
too.
I
think,
basically,
it
just
automatically
pulls
the
resource
attributes
from
that
hotel
resource
attributes,
environment
variable
in
the
providers
or
rather
in
the
in
all
resources
for
the
spec.
So
I
think
it's
got
one
approver
from
one
yeah.
B
Can
you
can
you
assign
it
to
me?
I
started
taking
a
look
at
it
already:
cool
whoa.
E
Cool
all
right.
Next
up,
we
have
amos
yeah,
this
one's
pretty
simple.
It
just
adds
some
type.
Instead,
otlp
exporter,
all
right,
you
already
look
and
approve.
This
just
needs
one
more
poover
yeah.
I
could
do
this
one.
E
Well,
I
think
you
already:
oh
this
yeah,
never
mind.
Okay,.
G
E
Cool
nice,
so
we
don't
have
any
active
callouts
right
now.
E
Alex's
chore
I'll.
Take
a
look
at
that
after
we
have
a
zipkin
exporter
issue
by
will
guo
is
this
is,
is
this
person
on
the
call
right
now.
E
Also
intern
at
aws,
so.
B
I
I
don't
think
the
person's
on
the
call,
but
I
already
took
a
look
at
it
and
I
just
I
mean
the
change
looks
pretty
good.
I
just
wanted
to
make
sure
that
the
the
status
was
being
tested
for
because,
right
now
it's
just.
B
Setting
the
default
so
yeah.
E
Okay,
so
once
it's
pretty
straightforward
I'll,
take
a
look
at
this.
E
E
Do
you
want
to
talk
about
this?
One.
H
Making
possible
to
define
propagators
via
environment
variables,
the
I
think
the
most
important
thing
in
this
pr
is
the
fact
that
we
are
using
the
string
from
the
environment
variable
to
locate
an
entry
point
and
that
entry
point
then
points
to
the
class.
H
I
don't
think
the
the
approach
of
using
a
string
and
then
try
to
use
it
as
a
path
to
import
a
module
or
an
object
in
a
module
with
a
right
approach,
because
the
specification
says
the
default
value
should
be,
for
example,
baggage
or
http
propaganda,
or
something
like
that,
which,
of
course,
doesn't
match
with
a
audio
path.
H
E
Nice
yeah,
this
looks
this
looks
pretty
straightforward.
Cool
thanks,
I
think,
did
someone
review
this
already.
Oh
okay
looks
like
alex.
Did
nice.
H
E
Nice,
okay,
we're
just
going
along
yeah,
okay,
so
this
is
one
of
mine,
pretty
straightforward.
It's
still
part
of
the
you
know,
just
using
the
is
recording
flag
for
all
instrumentations
this
one's
for
whiskey,
very
straightforward.
E
I'm
just
checking
this
this
flag
and
just
not
running
the
set
attributes
as
statuses,
if
you
know,
doesn't
exist.
So
if
someone
can
take
a
look
at
this
it'll
be
great.
I
have
like
another
pr
coming
that
like
modifies
three
of
them
at
the
same
time,
because
this
is
too
slow.
So
you
want
to
sign
me
late
and
since
I
looked
at
the
other
one.
C
E
C
E
E
Nice:
okay:
we've
got
some
ad
support
for
telemetry,
I'm
pretty
sure
this
is
batch
span.
Processor,
nice,
it's
another
invariable
one,
pretty
straightforward!.
E
I
guess
diego:
do
you
want
to
talk
about
it?
A
little
bit.
H
Yeah,
I
don't
keep
track
here,
scroll
down
scroll
down,
try
to
remember
what
I
was
doing
here:
yeah,
okay,
yeah.
The
specification
says
that
we
got
a
support
configuration
for
the
batch
span.
H
Processor,
so
that's
pretty
much
just
adding
support
for
that.
Using
the
standard
configuration
nice.
E
B
You
can
you
can
just
assign
that
one
to
me
I'll
take
a
look
at
it.
B
B
A
E
E
Cool
that
sounds
good
yeah,
so
we'll
just
await
the
response
for
that
add
and
variables.
So
this
is
a
work
in
progress.
One.
B
He
is
not
in
the
call,
I
think,
there's
a
there's
a
question
from
the
contributor
around
whether
or
not
there's
some
changes
that
we
wanted
to
have
in
the
spec
to
clarify
some
of
the
implementation
details
here
actually,
but
I
think
always
already
got
some
comments
on
the
on
pr.
So.
E
I
see
okay.
This
is
like
four
days
ago.
We
could
like
see
if
anything
comes
out
of
that,
like
whether
he's
doing
some
digging,
so
I
think
we're
fine
for
that.
It
is
a
work
in
progress,
one
too
so
cool
next
up
all
right.
So
this
was
aaron's
man.
This
global
error
handler
thing
man.
I
think
I
gotta.
I
gotta
review
this.
H
H
H
Okay,
yes,
so
this
pr
introduces
a
new
mechanism
for
our
users
and
when
I
say
users
I
mean
it
in
the
broadest
sense.
I
mean
application
code
developers,
instrumentation
developers
or
even
core
open
telemetry
api
or
sdk
developers
right.
H
So
this
provides
a
context
manager
that
you
can
instantiate.
You
will
do
right
with
global
error
handler
and
whatever.
F
H
So,
for
example,
if
a
serial
division
error
is
raised
in
that
context-
and
there
is
a
an
error
handler
that
can
handle
serial
division
errors.
Well,
that
handler
will
execute
error
code
right.
If
not,
if
there
is
no
error
handler
register
for
the
exception
being
erased,
then
the
default
error
handler
is
called
and
that
the
folder
handler
just
logs
the
exception
to
standard
output.
H
A
Normally
awesome
cool
is
there.
E
H
Yeah,
you
know
the
this.
E
F
C
H
No
worries
yeah;
okay!
Yes,
if
you
open
that
file,
error
handling
and
be
in
the.
H
H
That's
daniel
yeah,
okay,
there
yeah.
Thank
you
for
the
comments
daniel.
I
don't
know
if
you
concerns
are
already
addressed
or
not.
D
Yes,
I
mean
my
comments
were
basically
it
started
with
what
I
mentioned
at
last
week's
sig
meeting,
but
in
the
end,
it's
really
just
that.
I
found
the
api
a
bit
strange
with
the
context
manager
in
case
as
well.
We
have
to
raise.
A
D
Yeah
exception
directly
so,
but
I
think
our
conversation
regarding
that
is
pretty
much
resolved.
I
think.
E
All
right
cool
yeah,
so
if,
if
everything
is
resolved
there,
can
you
like
comment
on
this
pr,
so
other
people
will
know.
E
Okay,
cool
we're
about
halfway
through
got
this
support
for
hotel,
looks
like
a
work
in
progress.
It's
by
daniel.
E
Okay
and
that's
pretty
much
it
for
the
ga
tasks,
I'm
seeing
alex,
there's
still
a
bunch
of
like
these
old
pr's
that
we
got
was
the
decision
just
to
like
leave
them
until
after
ga
and
then
we
deal
with
them
afterwards.
B
Yeah,
I
think
so,
there's
one
there's
one.
If
you
go
back
to
the
previous
page,
there's
one
that's
been
approved
for
a
long
time
and
it's
kind
of
just
sitting
there,
this
min
max
some
count.
Aggregator
is
this
one
still
worth
doing
with
all
the
changes
that
are
still
needed
in
the
metrics
sdk
or.
C
I
think
the
I
think
the
aggregator
is
still
in
question
for
what
they're
gonna
use
yeah
for
for
the
which
one
is
a
value
recorder,
so
I
think
they're
trying
to
change
the
dd
sketch.
C
B
E
The
the
prometheus
exporter
before
didn't
have
like
a
type
check
for
the
certain
metric
types,
so
they
didn't
support
the
min
maximum
account
aggregator
they're.
Just
adding
that
you
know
that,
if
statement
now,
I
think
before
it
was
just.
E
C
Yep
yeah,
if
I
remember
right
the
reason
that
josh
wanted
to
change
the
default
like
a
few,
maybe
like
a
month
or
two
ago
to
min
max
sum
count
last
was
because
they
wanted
to
send.
I
think
it
was
prometheus
gage
with
just
the
last
value
yeah
and
yeah.
So
I
mean.
E
E
I
actually
agree
with
that
too,
like
there's
a
lot
of
use
cases,
especially
in
azure,
where
we're
using
the
valley
recorder,
but
we
can't
get
like
like
a
gauge
type
thing
yeah
and
if
it's
cumulative
like
or
like,
if
it's
stateful
like
we
lose
the
we
lose
that
information.
So,
but
did
you
say
like
they
were
trying
to
make
it
a
dd
sketch
instead.
E
Yeah,
okay,
I
guess
we'll
just
have
to
wait,
see
what
happens.
C
Yeah
one
one,
quick
question,
though:
do
we
already
change
it
to
min
max
some
count
last
or
for
some
min
max
last,
some
count
whatever
it
was
in
the
in
our
sdk.
E
Oh,
we
did
not,
I
don't
think
so:
okay,
okay
yeah
cool,
but
that
that
pr
was
already
merged
in
the
specs
to
make
that
as
the
default
aggregation.
I
don't
think
so.
I
think
it
was.
E
B
I
think
we
just
need
someone
to
pick
up
the
work,
because
I
think
there
was
some.
There
was
a
question
from
aaron
on
what
was
actually
being
tested
here
right.
E
Yeah
that
was
a
while
ago.
I
think
we
answered
that
aaron.
Do
you
still
have
confusion
about
what
this?
What
we
need
for
this.
C
C
E
B
I
don't
know
if
we
need
it,
but
I
mean
it
would
be
nice
to
have
the
added
confidence
of
having
these
tests
running
and
proving
that
things
are
working.
The
way
it's
expected
to.
E
Yeah,
it
just
says
like
we
just
need
the
concurrency
for
all
of
the
top
level
api,
like
methods.
C
Okay,
but
I
think
that
mostly
is
referring
to
like
the
parts
in
the
spec
that
say
this
must
be
non-blocking
or
like
the
concurrency
requirements
in
the
spec,
more
so
than
testing
right.
E
Yes,
I
guess
testing
would
better
enforce
that.
I
guess
you
know,
instead
of
just
us
saying,
hey,
look
we
have
locks
and
stuff
so
and
batch
band
processor
would
be
part
of
tracer
provider.
I
feel
so.
E
But
yeah,
I
think,
you're
right,
like
they
they're
more
referring
to,
like
you
know
like
get
tracer
or
stuff
like
that,
like
the
api
methods,
but
I
think
this
is
also
important.
We
just
had,
like
you
know
some
issues
before
with
the
the
bsp
in
terms
of
concurrency
before
that
daniel
solved
so
yeah
yeah.
C
So
then,
I
guess,
like
the
action
item,
would
be
figuring
out
what
we
want
to
test
and
then
actually
writing
writing
a
test
for
that.
But
I
think
I
think
we
agreed
it
would
it
would
either
be
flaky
or
sometimes
past
when
it
should
be
failing.
Yes,
exactly
et
cetera,
so.
E
So
my
understanding
is
like
there.
This
is
supposed
to
test
like
the
batchman
process
has
like
the
the
the
flush
and
shutdown
methods
which
can
be
invoked
by
the
user,
and
we
want
to
see
whether
or
not
the
behavior
remains
consistent.
If
the
processor
is
like
currently
exporting
or
something
like
that
or
currently
flushing
and
someone
calls
flush
or
shutdown,
I
think
that's
pretty
much
the
gist
of
it.
E
That's
just
my
understanding,
but
it's
difficult
to
test,
because
we
don't
know
how
to
set
up
that
specific
situation
yeah.
So
I
guess
that's
that's
pretty
much.
E
It
all
right
cool.
Does
anybody
have
any
other
topics
before
we
move
on
to
issues,
so
we
have
20
minutes
left.
B
No,
I
think
most
of
the
issues
that
I've
been
creating
are
all
around
the
all
around
the
required
requirements
for
ga
or,
at
the
very
least,
some
of
the
items
in
the
compliance
matrix.
So
maybe
you
want
to
bring
up
the
compliance
matrix
that
everybody
kind
of
gets
an
idea
for
what
we're
talking
about
when
we
say
that.
E
E
So
if
you
guys
haven't
seen,
this
is
the
spec
compliance
matrix
for
all
the
languages,
and
this
is
supposed
to
be
a
comprehensive
list
of
features
or
behavior
that
we're
supposed
to
have
I'm
not
actually
too
certain
whether
or
not
this
is
what
we
need
for
ga
like
this
is
the
ga
requirements,
but
I
guess
it
could
be
safe
to
assume
that
it's
for
ga
so
basically
like
alex-
and
I
have
been
like
just
going
through
our
code
and
seeing
like
whether
or
not
we
have
like
fulfilling
all
of
these
most
of
them.
E
We
got
so
this.
This
concurrent
calls
thing
that
we
mentioned
before
is
like
for
here,
tracer
provider,
tracer
and
all
the
ones
that
we
are
missing.
We
alex
created
issues
for
those
so
yeah,
so
this
is
pretty
much
what
we're
going
to
be
focusing
on
for
the
next
month
or
so
a
bunch
of
them
are
pretty
straightforward,
so
feel
free
to
you
know,
pick
them
up.
E
E
E
Okay,
cool.
I
have
a
pr
out
that
I
wanted
to
talk
about.
That's
not
part
of
ga,
so
I
guess
like
we
could
talk
about
that
now.
This
is
the
adding
metric
collections
as
part
of
instrumentations
pr.
E
This
is
related
to
like,
what's
it
called
so
there's
certain
there's
certain
metrics
that
we
want
to
collect
based
off
of
the
instrumentations,
for
example,
duration
of
the
request
is
one
of
them,
as
well
as
the
count
which
will
be
encapsulated
in
this
value
recorder.
So
there's
no
mechanism
right
now
currently
to
be
able
to
do
this,
and
I've
explored
several
different
methods
to
try
to
abstract
this
without
having
to
change
every
single
instrumentation.
E
However,
none
of
them
are
that,
like
viable.
The
main
reason
is
because
we
don't
really
want
to
couple
this
with
the
tracing
architecture,
so
we
don't
want
to
depend
on
you
know,
span
or
sampler
or
span
processor
or
anything.
So
that
means
we
must
like
change
every
single
instrumentation
that
we
create.
So
this
first
one
is
just
the
just
a
proposal
for
like
the
request,
instrumentation.
E
The
basic
idea
is,
since
all
the
requests,
all
the
instrumentations
depend
on
the
base:
instrumenter
the
pack,
the
instrumentation
package.
You
know,
I
simply
just
wrote
a
like
a
metrics
collector
within
the
instrumentation
package.
Oh
sorry,
metrics
like
mix
in
so
if
any
any
instrumentation
wants
to
like
enable
this,
they
would
just
like
add
this
mixing
in
and
they
will
be
able
to,
like.
You
know,
automatically
collect
these
kind
of,
like
oh
request,
counts
or
durations.
E
Every
time
a
request
comes
in
the
design
is
pretty
straightforward.
You
know
I'm
just
open
to
like
behavior
kind
of
feedback
and
implementation
feedback
as
well.
C
There's
hey
leighton
yep
on
that
line
to
256
there,
where
you
have
spanking
dot.
Client,
yes,
so
so
is
what
is
that
isn't
that.
C
E
Oh
no,
so
this
this
just
specifies
like
what
kind
of
metrics
we're
collecting.
So
it's
like
this
is
the
requests
instrumentation
like
it
only
sends
like
client
type
requests.
You
know
like
outgoing
requests,
so
this
is
just
to
like
populate
the
like
the
description
and
the
name
of
the
metric.
C
E
Yeah,
so
it's
not
exactly
coupling
like
it,
doesn't
use
anything
related
to
span.
It's
just
like
seeing
what
type
of
you
know
like
do
you
want,
because,
like
the
labels
are
populated
based
off
of
like
whether
it's
a
client
type
or
a
server
type
for
the
metric,
so
it's
just
an
indicator
of
like
what
kind
of
type
it
is.
That
does
bring
up
a
good
question,
though,
like
it's
like
what,
if
like
an
instrumentation,
provides
both
right.
E
Like
I'm
sure
there
are
some
like
servers
that
also
provide
some
api
to
like
send
outgoing
requests,
or
something
like
that,
which
I
haven't
really
addressed.
Yet
this
is
just
like
a
first
kind
of
pass-through.
We
can
just
change
the
api
as
we
see
fit
so
sure.
C
Wasn't
there
an
issue
for
several
issues
about,
like
specifically,
mean
I
think,
the
way
that
it
was
phrased
and
I've
seen
this
request.
A
lot
is,
how
can
I
collect
metrics
from
spends,
so
what
people
actually
seem
to
want?
Is
they
want
them
coupled
like,
rather
than
a
design
design
decision?
That's
like
how
people
seem
to
be
reasoning
about
it.
I
think
that
I
think
there
was
some
discussion
in
the
actual
spec
as
well,
and
maybe
people
like
agreed.
C
We
don't
want
to
have
spans
and
metrics
coupled
together,
but
was
this
like
a
spec
decision
not
to
do
that
not
to
implement
it?
That
way.
E
Yeah,
so
I
haven't
looked
into
that
too
much.
I
don't
know
if
there
was
any
like
recent
development
for
that,
but
I
have
like
from
what
I
read
before:
it's
like
people
didn't
want
them
to
be
coupled
and
also
like.
We
as
like,
like
we
as
azure.
We
we
need
this
feature,
which
is
why
I
made
this
and
we
can't
have
it
coupled
between
tracing
and
metrics.
So
this
is
less
of
a
like
spec
compliance
thing
and
more
of
a
like
we
as
a
customer.
E
We
need
this
so
yeah,
okay,
when
you
say
you
can't
have
them
coupled
what
do
you
mean
exactly
like
we
want
to
so
from
a
customer's
standpoint
like
there
are
customers
who,
like
we
expect
them
to
like
they
want
to
have
these
kind
of
standard
metrics
collections,
but
if
they're,
if
they're,
not
you
know
if
it's
not
encapsulated
within
like
a
start
span
or
something
like
that
or
if
they're
not
using
the
tracing
mechanisms,
they
still
want
to
be
able
to
get
these
instrumentations
metrics
related
to
instrumentations.
E
Yeah,
it's
a
good
question,
so
the
main
reason
why
we
don't
want
to
do.
That
is
because
it's
also
related
to
like
the
is
recording
changes.
I've
been
making
the
the
performance
is
a
very
like
big
deal
for
our
customers.
So
if
we
leave
it
down
to
the
span
processor
like
we
want
to
be
able
to
collect
these
decisions
like
before,
like
sampling
is
done
and
everything
okay
regardless,
if
like
sampling,
is
done
so
and
if
it
goes
down
the
span
processor
we
won't
be.
E
We
won't
be
able
to
see
that,
like
all
the
attributes
related
to
the
the
method
and
stuff
like
that,
so.
C
E
Yes,
so
if
you
could
suggest
a
different
way,
which
I
probably
haven't
found
yet
yeah
I'm
all
years,
because
I
really
am
not
looking
forward
to
making
this
change
for
every
instrumentation.
So.
C
Yeah,
my
other
concern
is
just
because
I've
seen
like
a
lot
of
questions
either
in
beginner
or
issues
that
is
like
very
specifically
people
saying.
How
can
I
make
metrics
from
his
fans
because
that's
what
they
want
you
know.
So
I
guess,
if
it's
settled
in
the
spec,
then
that's
that
so.
E
Yeah,
I
haven't
seen
these
specifically
in
the
getter,
but
I
I
don't
know
why
like?
Why
do
they
do?
They
actually
mean
they
want
to
see
it
from
a
span
or
they
just
want
to
see
that
metrics
in
general
and
they're,
just
throwing
in
the
word
span
there,
because
they
associate
that
with
like
request
tracking
you
know
like:
do
they
really
need
it
to
be
coupled
with
a
span,
I'm
not
really
sure.
C
E
E
Yeah
yeah
right
exactly
so.
Why
don't
we
just
use
that
yeah,
but
anyways
yeah
so
feel
free
to
take
a
look
at
this
pr?
It's
not
really
quite
for
ga,
but
I
kind
of
need
this
so
yeah
awesome.
Do
we
have
ten
minutes
left?
Does
anybody
else
have
any
other
topics?
They
want
to
talk
about
any
open
issues
we
kind
of
like
zoom
past
today's
meeting,
so.
D
No,
no,
I
I
mean
I
assume
we
we
want
to
have
the
documentation
in
somewhat
of
a
decent
state
for
ga
and
I
think,
on
the
website
sick.
There
was
an
issue
regarding
cross
language,
examples
for
getting
started
and
stuff
like
that.
D
So
I
was
just
wondering
if
doing
anything
for
documentation
was
on
the
radar
or,
if
the
state
of
the
documentation,
basically
if
we
have
kept
it
up
to
date
until
now
and
in
a
user-friendly
state
or
if
we
want
to
do
anything
to
some
kind
of
deadline
like
j.
E
Yeah,
that's
a
good
question,
so
I
think
like
as
like
the
code
changes
are
iterating.
We
are
making
the
the
doc
changes
accordingly.
However,
you
do
bring
up
a
good
point.
I
think
we
do
need
like
one
more
pass-through
of
like
to
see
whether
or
not
the
docs
are
in
good
shape
for
ga,
so
we
should
definitely
probably
put
that
on
our
radar,
even
though
we've
been
focusing
on
the
feature
matrix
and
it's
not
explicitly
outlined
there,
but
yeah-
that's
a
good
point.
D
No,
that
was
on
their
websites,
sick.
They
were
talking
about
about
including
a
getting
started,
guide
and
setting
up
exporters
and
stuff
for
each
language
like
they
don't
have
many
issues,
so
it
shouldn't
be
difficult
to
find,
but.
E
Okay,
awesome
should
I
maybe.
E
Yeah,
if
you
can
open
issue
and
find
the
like
the
link
for
that
that'd
be
great.
E
Oh
okay,
so
fast,
yeah,
okay,
so
cool
yeah.
We
will
do
a
pass-through
for
that
and
if
we
find
any
like
discrepancies,
we'll
open
up
issues
for
those.
D
C
C
E
C
Right,
I'm
just
curious,
like
I've
noticed,
we
have
a
lot
of
examples
which
is
great
and
then
there's
not
so
much.
We
basically
just
have
the
getting
started
guide.
E
E
Yeah,
like
I
feel
like
we
have
a
lot
of
like,
I
guess,
features
in
our
code
base
that
are
kind
of
like
hidden
away
and
not
exposed.
A
E
Each
of
these
sorry
sorry
alex
each
of
the
examples
are
like
very
straightforward
and
basic,
and
I
guess
we
leave
it
up
to
the
user
to
try
to
derive
what
to
do
for
their
use
cases.
B
B
I
mean
things
could
always
be
better,
so
hopefully
the
rework
and
the
open
telemetry.io
can
kind
of
lead
us
in
a
path
that
we
can
all
kind
of
follow
along.
Yes,
things,
yeah.
C
Yeah,
I
was
wondering
from
your
guys
perspectives
like
microsoft,
light
stuff.
Whatever
are
you,
do
you
guys
have
like
your
own
documentation
for
sort
of
your
users
or
or
were
you
planning
on,
like
I
don't
know,
having
tech
writers
help
with
the
open
source
docs
just
curious.
E
So,
on
the
microsoft
side,
when
we
ga
our
azure
exporter,
we
will
have
microsoft,
specific
documentations
on
our
production
website.
So
it's
it's!
It's
purely
for
using
the
azure
exporter
with
the
open,
telemetry
apis
and
like
the
most
common
like
instrumentations,
we
would
expect
users
to
use
like
requests.
Flask,
django
stuff,
like
that
yeah.
C
E
B
We
actually
started
this
opentelemetry.lightstep.com
website
to
allow
us
to
kind
of
iterate
quickly
through
our
documentation
to
help
users
of
open
telemetry,
but
also
tie
into
the
lightsaber
product.
So
I
mean
you
can
go
and
check
it
out.
There
there's
a
bunch
of
concepts
that
are
explained
there,
which
I
I
think
we've
had
like
technical
writers
actually
involved
in
the
process,
which
has
been
really
good.
E
Is
there
a
reason
you're
asking
that
like
is
there
some
discrepancy
that
you
saw.
C
Oh,
no,
no,
nothing
in
particular.
It's
just.
I
know
that
sometimes
like
when
I
go
to
write
documentation
like,
like
you
said
like
from
a
user
perspective,
it
doesn't
come
out
that.
E
Yeah,
I
get
you.
Okay,
cool
makes
sense
all
right.
We
got
one
minute
left.
I
think
I
think
we
pretty
much
covered
everything
we
need
today
yeah,
so
I
new
people
welcome
again
yeah.
If
you,
if
you
have
any
trouble,
you
know,
feel
free
to
reach
out
to
one
of
the
veterans
and
the
getter
channel
is
probably
the
best
way
to
do
that
so
yeah
cool.