►
From YouTube: Open Telemetry's Personal Meeting Room
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).
C
C
C
C
B
D
He
welcome
everybody
to
another
edition
of
the
Python
Sig
as
usual.
Please
add
your
names
to
the
attendees
list.
Have
any
topics
or
issues
you
would
like
to
discuss?
Please
add
them.
D
Okay,
apparently
leading
is
not
joining
for
the
first
Power
at
least,
and
it's
not
here
as
well.
D
If
not
Jeremy
I
think
I
just
went
through
one
of
our
issues.
B
Yeah,
so
this
issue
is
related
to
the
resource
detection
like
Contender
ID
in
the
auto
instrumentation
part.
So
there
is
a
way
to
do
that
is
just
we
Implement
our
own
resource
detector
class,
which
extend
the
actual
resource
detector
class,
and
in
that
we
basically
collect
the
container
ID
and
set
it
as
a
attribute
in
the
resource.
B
The
thing
is
that
I
mean
this
is
what
the
basic
feature
that
we
are
planning
to
implement,
so
that
it
can
be
helpful
to
implement
the
application
and
infrastructure
instrumentation
a
correlation
for
application
infrastructure
instrumentation.
So
the
thing
is
that
in
that,
while
I
was
just
you
know,
thinking
about
the
design,
so
there
is,
there
is
a
way
where
we
can.
B
You
know,
provide
the
resource
detector
like
I,
have
mentioned
here
in
that,
while
you
know
calling
the
set
Tracer
provider,
but
this
is
only
helpful
when
user
hasn't
provide.
The
you
know
hasn't
created
the
Tracer
provider
himself,
and
we
can.
We
can
create
this
test
provider
inside
the
auto
instrumentation
part,
but
in
case
user
has
created
the
pressure
provider
yourself.
So
in
that
case
there
is
no
SDK
or
you
know
any
other
way.
We
can
modify
that
Tracer
provider
with
our
new
resource
I
mean
which
contains
our
resource
detector.
D
B
Ahead,
yeah,
please
so
I
had
a
discussion
which
you
can't
cover
this,
so
he
said
that,
as
of
now
I
mean
that
it
is
not
supported
by
SDK,
so
he
has
created
an
issue
in
the
spec
specification.
For
you
know:
creation
of
a
new
environment,
variable
Hotel
resource
detector,.
B
So
in
the
hotel
resource
detector,
we
can
specify
the
detector
that
we
want
to
you
know,
load
and
that
will
basically
fetch
the
resource
with
the
key
I
mean
interested
key
value.
Pair
and
settings
are
current
Tracer
provided.
D
Okay:
okay,
couple
of
questions
is
this
resource
detectors,
environment
variables,
something
that
other
six
may
use
Java
or
go
or
whatever?
No.
B
It
basically
I
think
I'm
not
sure,
though,
but
I
guess
it
is
implemented
in
this
one:
I
guess:
Java,
dot,
JavaScript
and
Java.
This
feature
is
already
implemented,
but
they
are
not
using.
This
kind
of
you
know
enormous
variable.
D
Okay,
yeah.
What
I
mean
is
that
for
that
PR
to
be
successful
in
this
spec,
we,
it
probably
needs
to
be
a
feature
that
it's
of
interest
of
for
other
things
as
well.
B
D
What
I'm
trying
to
say
is
that
for
that
PR,
the
frequent
opened
in
the
specification
repo
for
that
PR
yeah
to
be
successful.
There
will
probably
need
to
be
a
feature
that
it's
also
important
for
other
six
right.
B
D
I,
don't
think
it
will
be
I
think
the
the
spec
will
add
a
feature
just
for
python.
D
E
D
It
will
be
hard
for
this
PR
to
work
right,
because
it
will
probably
require
the
the
other
six
to
refactor
themselves
so
that
they
will
use
the
the
environment
variable.
D
B
Other
other
I
mean
there
is
a
I
mean
Serene
who
has
created
that
issue,
so
he
has
provided
this
PR,
where
you
know
they
are
basically
creating
their
own
container
detector,
but
he
hasn't
mentioned
I
mean
how
he
hasn't
mentioned.
That
I
mean
if
they
are
using
any.
You
know
about
variable.
D
D
Yeah,
because
what
you
actually
need
is
a
way
to
modify
an
existing
resource
right.
B
D
So
here
in
in
JavaScript,
how
do
they
do
that?
How
do
they
modify
an
existing
ring
resource.
B
That
I
see
here
to
go
through
the
whole
code,
because
it's
a
little
bit
big
there
so
I'm,
okay,
but
idea
how
to
actually
use
that.
But
this
is
just
a
PR
where
the
basically
implement
this,
but
that
part
is
already
done
in
not
just
in
JavaScript
but
in
Java
and
Dot
net
as
well.
Yes,
it
is
mentioned
yeah.
It
is
Java
javascript.net
and
go
it's
already.
There
I
I'll
dig
into
this
and
find
out
okay,
how
exactly
they
have
managed
to.
You
know,
modify
the
resource.
D
Yeah
I
will
suggest
that,
because
I,
don't
I,
don't
think
adding
an
environment
variable
will
get
much
traction
in
this
pack.
If
dr6
already
do
this
without
using
that
environment
variable
right,
maybe
they
are
interested
in
in
having
their
environment
variable
as
well.
Maybe
I
mean
that
that
could
also
be
a
possibility
right,
but.
E
D
I'll
suggest
that
we
take
a
look
into
how
your
language
is
implemented,
so
maybe
there's
a
way
that
we
can
modify
the
resource
without
having
to
go
so.
B
Let's
say
if
we
don't
consider
the
economic
variable
right
now,
so
how
do
we
manage
to
do
that?
You
know
how
can
we
specify
the
resource
detector
inside
the
you
know,
Tracer
provider?
If
the
Tracer
provider
is
already,
you
know
been
created
by
user
and
pass
it
to
this
address
again,
yeah.
D
Good
question:
that's
that's
what
I
I
think
looking
into
the
other
six
might
yield
as
an
answer,
because
I
think
what
would
be
ideal.
Probably
for
you
will
be
that
we
have
a
way
or
we
create
a
way
to
do
that
right
to
modify
an
existing
resource
so
that
you
can
do
this.
D
B
But
according
to
the
specification
there
is
I
mean
the
resource
must
not
be
modified
after
the
created.
Okay.
B
I
got
also
confused
that
we
have
to,
we
have
to
add
the
resource.
Detector
I
mean
the
resource
which
we
got
from
our
resource
detector.
At
the
time
of
press
you
know,
while
we
are
setting
it
in
the
retraceable
provider,
only.
D
Okay,
okay,
so
yeah
so
I
guess
the
stricken
plant
with
the
environment
variable
is
to
Define
it
there
in
the
environment
where
in
the
environment
variable,
so
that
when
the
Tracer
providers
created
the
resource
takes
that
information
to
instantiate
the
the
architecture
they
want
right,
which
makes
sense
in
order
to
avoid
modifying
the
resource.
Well,
something
else
that
could
be
done
maybe
is
not
to
modify
the
existing
resource,
but
to
create
a
copy
of
the
resource.
F
B
F
Yeah
yeah
so
like
there
is
one
get
Tracer
function
when
like
it
is
passed,
may
I
share
the
screen
or
Samsung,
or
is
it
just
yeah.
F
So
you'll
be
able
to
see
my
screen
yep
yeah,
so
this
is
the
function
like
originally
yeah.
F
Originally,
this
is
the
function
that
is
getting
called
whenever,
like
we
get
a
tracer
okay,
so
it
calls
this
function.
Okay,
and
here
there
is
a
global
Tracer
provider-
okay,
unless
and
until
a
user
gives
it
okay
like.
If
a
user
sets
a
tracer
provider,
okay,
we
will
get
a
tracer
provider
class
here,
okay
and
currently
we
cannot
modify
the
resource
of
the
stressor
provider
class
or
the
instance
of
it.
F
Okay,
so
what
we
can
do
is
we
can
call
another
function:
okay
and
pass
the
Tracer
provider
class
and
the
default
resource
that
we
need
to
give
him
like
the
resource
generate
like
detectors,
okay
and
it
will
inherit
the
norm
and
we
will
create
a
new
class
named
new
Tracer
provider,
okay,
which
will
inherit
the
normal
Tracer
provider
class,
and
it
will
like
merge
the
resource
or
the
current
resource
of
that
pressure
provider
and
the
default
resource
that
we
are
providing
him
and
we
can
again
like
return.
F
D
I
think
yeah.
Somehow
we
will
need
to
patch
the
Tracer
provider
if
we
do
not
follow
the
environment.
Variable
idea
that
silicon
has
this
could
be
a
way.
Yes,
definitely
yeah,
I
see
now
why
the
the
environment
variable
approach
would
be
convenient
right
because
it
will
save
us
from
having
to
patch
we,
which
I
guess
will
be
also
convenient.
D
D
G
F
Okay,
do
you
need
me
to
share
the
screen
or
like.
B
F
B
I
think
Diego
says
that
I
should
look
at
the
other
six,
but
others
see
how
other
six
have
implemented.
This
I,
like
I,
have
a
look
at
them
and
get
the
idea
which
I
mean
how
we
can
Implement,
which
is
nearly
similar
to
them.
You
know
without
I
think
without
creating
a
new
enormous
variable,
or
you
know,
patching
to
API.
D
Yeah,
in
fact,
I
think
that
if
you
look
at
the
other
six,
it's
probably
I
mean
it's
definitely
probable
that
they
also
do
some
kind
of
patching.
D
D
All
right,
yeah
great
discussion
as
I
could,
please
you
know
when
you
have
a
chance
to
play
that
that
issue.
D
If
you
want
to
you,
can
tag
me
there
so
that
I
can
take
a
look
at
it
when
all
right
right
should
we
move
to
another
topic
or
are
there
any
more
comments?
I.
G
Reminds
me
a
little
bit
of
there
was
a
discussion
about
basically
like
how
patterns
for
injecting
configuration
like
that,
and
it
seemed
like
people
were
leaning
towards
Factory
functions
that
can
provide
whatever
piece
of
of
code
we're
looking
to
inject
there
like
specific
classes,
potentially
I,
don't
know
if
this.
This
feels
like
a
very
similar
thing
to
want
to
do
is
to
inject
a
specific
resource,
detector
that
could
be
used
and
I
don't
know
if
there
are
other
places
in
the
code
base
already
that
are
using
this.
D
D
D
And
and
this
class
will
have
the
it
will
create
a
Trader
provider
class
plus.
G
Well,
I
was
wondering
if
we
could,
if
there
was
a
place
where
we
use
the
resource
detector
and
instead,
what
we
really
want
to
allow
to
be
done
is
to
like
inject
a
custom
resource
detector
function,
but
that's
maybe
that's
not.
Actually
what
we
want
to
do
I
was
just
because
it
seems
like
what
we
without
actually
patching
needing
to
patch
the
Tracer
provider.
We
could
instead
say
here
now
the
Tracer
provider
is
just
going
to
call
this
function
to
get
the
resource
or
wherever
that
happens
exactly.
G
Enough
to
it
looks
like
we
do
some
we
do
some
merging
and
shaking
already
so
I'm,
not
sure
if
that's
actually
going
to
work,
but.
D
Yeah
that
that
also
sounds
like
like
a
possibility.
In
fact,
we
have
use
the
factory
pattern
before
some
other
places,
but
but
yeah
in.
D
Fact,
I
suspect
that
that
that'll
that's
also
a
good
candidate
to
finding
in
the
implementation
of
other
six
like
for
factory
function
or
something
like
that
one
yeah
definitely
possibility
now.
Let
me
mention
it.
C
D
Yeah
good
point:
okay!
Yes,
any
other
comments
lead
to
this
issue.
D
Okay,
Jeremy
I
was
looking
at
your
issues
and
I.
Think
I
found
an
issue
related
to
some
circular
dependency.
E
The
first
one
was
created,
basically
to
because
I
noticed
that
the
custom
sampler
injection
that
was
added
has
like
an
issue.
I
wanted
to
quickly
fixed,
and
so
this
issue
is
basically
to
change.
I
basically
have
a
PR
to
do
this.
E
It's
ultimately
just
to
change
the
way
it's
done
so
that
there's
not
really
a
chance
of
circular
dependency,
but
in
order
to
do
that,
it
meant
that
custom
samples
would
have
to
be
manually
passed
into
the
Tracer
provider
for
manual
instrumentation,
and
that
this
entry
point
and
environment
variable
level
injection
could
only
happen
for
auto
instrumentation
like
the
way
we
use
other
entry
points.
E
So
in
the
last
thing
people
mentioned
that
they'd
ideally
like
to
see
another
way
to
solve
this.
That
would
allow
us
to
actually
even
use
entry
point
injection
for
the
first
time
for
manual
instrumentation,
so
basically
keep
the
previous
functionality
where
you
could
do
for
auto
or
manual,
and
so
for
that
reason,
I
created
this
ticket
going
into
a
bit
more
depth.
C
E
Merged
the
idea
is
that
well,
I
would
close
this
issue,
which
is
the
first
issue.
That's
just
hey,
there's
some
circular
defenses
happening,
but
leave
this
feature
request
to
open,
which
is
look
into
how
to
solve
this
more
broadly
yeah.
D
Mean
if
it
says,
fixes
yeah,
he
yeah
I,
don't
know
if
GitHub
understands
that
syntax
I
think
you
just
need
fixes
and
then
pound
sign
and
then
I.
D
Correct:
okay,
great
just
question:
I,
remember
something
about
you
reporting
an
issue
with
the
circular
dependencies
that
was
related
to
typing
and
that
could
be
fixed
with
type
checking.
Is
that
the
same?
Is
this
the
same
thing
or
it's
a
different
yeah.
E
So,
like
I
think
two
sigs
ago
people
recommended
that
I
in
the
sampler
that
was
causing
this
issue
just.
E
Basically,
like
like,
have
the
Imports
in
code,
but
that
that
didn't
solve
the
problem
either.
So
I
just
ended
up
going
through
with
the
with
the
six
VR.
D
Okay,
okay,
all
right,
so
it's
somewhat
related,
but
type
checking
didn't
fix
everything,
yeah!
Okay,
all
right!
Yes,.
E
E
Sampler
example
and
explain
how
people
can
use
it
to
reproduce
this.
It's
kind
of
tricky
like
you
have
to
basically
give
people
an
outline
of
the
package
since
it
needs
an
entry
point,
so
I'm
I'm
still
just
like
working
on
that
yeah
great.
D
Okay,
all
right,
thank
you.
Yes,
this
makes
it
much
clearer
a
little
confused
but
perfect.
Thank
you
for
your
explanation.
Okay
senior
issues.
We
may
have
okay,
so
there's
something
else.
There's
another
issue:
I'm
just
gonna
share
this
quickly
and
just
opened
PR.
In
fact,
a
while
ago
I've
been
working
on
this
PR
for
quite
a
long
time
now,
can
you
see
my
screen
now.
D
Yes,
so,
besides
the
exponential
histogram
implementation,
a
few
weeks
ago,
we
immersed
the
first
part
of
this
project,
which
was
the
logarithm
and
explanation
mappings.
D
They
said
a
new
aggregation
for
metrics,
it's
a
it's
a
big
bear.
It's
definitely
a
big
PR,
unfortunately,
but
it's
everything
is
kind
of
similar
to
what
we
expected
I
mean
here.
We
have
an
aggregation
and
just
as
a
new
aggregation,
it
has
the
Aggregate
and
collect
methods.
D
So
the
algorithm
itself
can
be
quite
long,
but
it's
no
thing
special.
In
that
sense,
it's
just
another
aggregation.
D
So
please.
If
anyone
has
a
chance,
I'll
appreciate
any
kind
of
review.
I
understand
that
it's
a
little
bit
big
but
but
yeah.
It's
it's
a
big
topic,
so
it's
impossible
to
make
a
small
repair
anyways.
So
please.
If
anyone
has
questions,
please
let
me
know
I'll
be
more
than
happy
to
answer
any
questions
that
you
may
have
so
yeah.
That's
what
I
wanted
to
share
with
you.
D
All
right,
any
other
topics,
PR's
issues.
A
Okay,
so
yeah
so
I've
been
working
on
this
PR,
so
this
is
related
to
manual
instrumentation
in
Falcon,
so
we
had
a
auto
instrumentation
earlier
available
in
Falcon,
but
not
the
manual
one.
So
I
need
like
basic
task
was
to
you
know
we
are.
We
are
given
an
application,
an
object
of
Falcon
API
and
we
want
to
add
instrumentation
to
it.
So
I
came
up
with
an
approach,
so
basically
what
basically,
what
we
did
is.
A
Let
me
show
you
so
what
we
did
is
so
basically
we
wanted
to
add
some
attributes
to
the
existing
app
and
and
like
and
want
to
patch
the
call
function.
That
is
there
and
also
add
add
our
middleware
for
instrumentation.
So,
but
the
issue
with
Falcon
is
Falcon.
Use
slots
wait
a
minute.
A
Let
me
let's
go
to
the
discussion,
so
Falcon
use
slots
which
is
like
you
know,
python
slots
is
externally
what
it
uses
it
uses
to
reduce
the
memory
because
of
like
you
know
it.
It
not
assign
the
memory
to
Keys
only
the
pair
of
attribute
only
the
values
of
the
attributes.
So
this
way
it
somehow
reduces
the
time
of
like
object
creation,
but
the
thing
is
once
it
uses
the
slots
any
of
any
class
uses
slots.
It
cannot
further
add
any
attributes
to
it.
A
A
Instead
of
adding
new
attributes
to
then
already
created
Falcon
object,
what
I
did
is
I
I
created
a
new
object,
which
is
a
new
object
of
the
child
class
of
Falcon
and
added
copied
all
the
attributes
into
it
and
added
our
own
on
attributes,
and
instead
of
returning
the
original
object
I,
we
will
just
return
the
new
object
that
we
created.
So
this
way
we
are,
we
are
able
to.
You
know
like
instrument
the
Falcon
object,
but
the
thing
is
this
comes
with
a
lot
of
bugs
and
issues.
A
So
we,
if
we
like
iron,
she
can't
actually
give
her.
We
are
having
discussions
regarding
it,
so,
like
I
explained
why
I
use
why
I
use
this
approach,
as
I've
already
explained,
so
she
can't
came
like
you
know,
explain
like
she
can't
show
me
some
examples
where
this
was
actually
failing
and
although
I
fixed
those
cases
but
I
mean
I
mean
it
like
I.
Like
you
know,
there
are
other
possible
cases
also
where
it
could
fail.
So
she
can't
mention
something
this.
A
She
can't
mention
this
thing
that,
because
these,
the
so
like
in
his
words
I,
would
move
the
Tracer
meter
instrument
config
state
to
trace
middleware,
since
it's
the
one
that's
actually
using
them,
so
what
he
meant
that,
instead
of
like
you,
know,
storing
all
these
attributes
into
a
falcon,
API
or
Falcon
object,
we
can
move
these
this.
These
data,
these
attributes
to
the
trace,
middleware
that
we
are
adding.
But
the
issue
with
this
approach
is
because
call
function
of
the
Falcon.
A
Api
also
uses
these
these
values,
so
we
cannot
just
simply
move
them
to
trace
middleware
because
otherwise,
we'll
not
able
to
access
those
variables
that
we
need
into
the
call
function.
So
that's
what
I,
like
you
know,
wrote
to
him
and
then
he
like,
he
replied
that
I
think
instrumented
Falcon
API
is
a
simple
middleware
which
is
actually
true,
so
he
said
that
it's
not
necessary
that
they
maintain
separate
Trace
meter
instrument
on
its
own.
It
can
use
what
will
be
used
for
the
trace
middleware.
A
So
from
what
from
this,
what
I
understood
is
he
wants
me
to
use
the
attributes
from
the
trace
middleware,
but
I
do
not
understand.
How
can
we
actually
do
this
because
Trace
middleware
is
not
like
the
trace?
Middleware
is
not
available
in
the
Falcon
API,
so
we
cannot
directly
like
cannot
access
any
attributes
of
the
trace
middleware
there.
It's
not
like
you
know
and
when
I'm
not
sure
what
exactly
he
meant
so
I
thought
maybe
we'll
discuss
this
thing
here
itself.
D
Right
so
first
question:
is
it
or
would
it
be
convenient
for
you
to
First,
submit
a
working
example
that
does
not
use
the
slots
and
then
try
to
use
the
slots
as
an
improvement,
or
it's
better
for
you
to
try
to
use
it
right
away.
A
Yeah,
so
it's
not
us
that
are
using
slots,
it's
the
Falcon
object
that
is,
like
you
know,
used
slots
to
store.
You
know
its
attributes,
so
Falcon
Falcon
API
itself
is
implemented
using
slots
yeah.
It
already
has.
A
So
right
now,
what
like
you
know,
because,
although
I'm
not
sure,
what's
the
correct
approach
to
solve
this
problem,
but
one
possible
way
is
to
you
know:
change
the
whole
instrumentation
that
Falcon
currently
uses
that
is
it
patches
the
call
function
it
that
is
there
the
call
function
of
Falcon
API
and
also
uses
a
trace
middleware
for
to
you
know,
hook
the
process
request
with
resource
and
response.
A
So
what
I'm
thinking
is,
instead
of
using
the
call
function,
patching
the
call
function,
we
can
move
the
whole
instrumentation
logic
to
the
middleware,
and
this
way
we
can
solve
this
issue
so
that
now
we
do
not
need
any
any.
Those
like
you
know,
attributes
such
as
stress
like
those
attributes
are
basically
this
Tracer
provider
and
meter
provider.
So
we
will
not
use
the
will
not
need
them
now
and
like
in
this
way.
We
can,
you
know,
get
around
this
thing,
but
not
if
you
cannot
think
of
anything
with
the
current
implementation.
D
Yeah
right,
when
you
say
moving
everything
to
the
middleware,
what
do
you?
What
do
you
mean
so.
A
Moving
everything
to
the
middle
there,
but
by
that
I
means
that
all
those
the
logic
of
creating
spans-
and
you
know,
collecting
attributes
from
the
request,
object,
and
you
know,
and
also
you
know,
closing
exiting
the
spans
and
all
this
thing
can
be
done
in
the
middleware
and
not
in
the
call
function
that
currently
what
we
are
doing.
A
So
all
this
thing
is
in
the
call
function,
but
when
this
can
also
be
done
using
middleware
also
I
mean
we
use
this
approach
in
fast
API,
okay,.
D
Yeah
I
mean
that
that
could
be
yeah
definitely
make
sure
to
mention
that
to
srikanta
it
could
be
a
possibility
if
that
helps
you.
Okay,.
A
D
Yeah
I
don't
know,
did
you
mention
anything
pretty
sick
but
well?
Hopefully
he
we'll
have
second
again
soon
take
a
look
at
and
you're
reachable.
D
All
right.
I
am.
D
Work
so
I'm
going
through
all
my
notifications,
which
are
a
lot
but
I'm
going
through
every
one
of
them.
So
if
there's
something
that
you
need
me
to
take
a
look,
please
be
patient.
I'll
try
to
to
advance
as
much
as
I
can
today,
but
if
not
in
the
probably
in
the
next
few
days,
I'll
be
done
with
all
of
them,
but
if
you
need
any
help,
also,
please
tag
me
or
contact
me.
Business
luck
directly.
D
D
All
right
any
other
issues,
but
anyone
wants
to
discuss.
D
Okay,
in
that
case,
thank
you
all
for
joining
and
we'll
talk
again
next
week.
Okay,.