►
From YouTube: 2020-07-02 meeting
Description
No description was provided for this meeting.
If this is YOUR meeting, an easy way to fix this is to add a description to your video, wherever mtngs.io found it (probably YouTube).
A
A
A
A
A
A
A
B
C
B
A
C
C
B
I
think
I'll
just
call
out
a
quick
I,
just
add
a
topic
there,
so
I
think
just
so.
I
should
just
look
like
I
think,
like
Layton
and
Alex
and
and
Chris,
but
just
for
the
larger
room.
I
am
I'm
switching
jobs,
so
so
I
don't
be
able
to
do
the
kids
much
time,
I
think
to
open
telemetry
as
before.
So
we
had
a
conversation,
I
think
the
results
of
that
is
all
priced
I'm,
gonna
step
down
as
a
maintainer,
but
probably
stick
around
as
an
approver.
B
B
B
C
C
So
you
know
it's
like
they
knew
that
you
were
leaving
or
something
yeah
if
I
also
have
another
topic
terrible,
so
the
views
API
PR
has
been
out
for
like
a
while.
Now
the
main
reason
why
is
kind
of
steered
away
from
it
is
because
we
discussed
in
the
metrics
sake
that
light
they
wouldn't
like
before
I
had
the
pasady
to
just
like
you
know,
experiment
and
like
try
to
push
out
a
prototype,
because
I
had
time,
but
now
I
don't
really
have
time
to
do
that
anymore
and
Chris.
C
Was
there
helping
me
with
a
lot
of
discussions,
but
I?
Don't
think
he
can
do
that
anymore.
So,
right
now
the
old
tip
is
kind
of
just
sitting
there
and
without
any
more
discussion
or
like
actual,
we
do
spend
a
lot
of
the
stuff
in
like
the
SDK
specs
too.
So
with
an
ever-changing
spec
and
like
the
views
otep
not
even
being
discussed,
it
doesn't
make
much
sense
to
keep.
C
C
I
have
stepped
into
it
quite
a
bit
and
like
I'm,
pretty
knowledgeable
about
it,
it's
just
that
my
branch
isn't
like
merged
with
the
latest
master.
So
if
anyone
else
wants
to
pick
this
up,
you
know
feel
free.
I
will
like
merging
the
latest
changes,
because
I
know
there's
a
lot
right
now.
We
you
can
either
like
pull
my
branch
or,
like
you
know
we
can.
We
could
just
talk
about
it,
one-on-one,
so
anyone's
interested
in
this
work,
then
it'd
be
pretty
pretty
okay
too.
So
let
me
know
I.
B
C
I
really
I
think
people
are
just
talking
about
like
like
semantics
and
like
naming
right
now.
I
haven't
contributed
much
because
I
don't
really
have
much
insight
or,
to
be
honest,
I,
don't
really
care
about
it.
That
much
like
in
terms
of
what
we're
gonna
be
naming
things
like
the
the
scenarios
for
Microsoft
have
already
been
lit
up,
so
it
is
what
it
is.
C
D
Yeah
I,
don't
think,
there's
much
to
talk
about
it.
Basically
I
mean
in
exam
players
pure
like
a
month
and
a
half
ago,
and
it
seems
to
have
been
received
reasonably
well
and
then
I'm,
just
I,
don't
know,
I've
had
enough
approvals
for
like
three
weeks
and
they're,
just
not
merging
it.
So
I
don't
really
know
what's
up
with
that,
but
it
seems
like
it's.
D
C
D
C
B
Think
I
think
Ted
might
be
here
and
when
I
discussed
sort
of
like
frustration
with
the
spec
moving
forward,
ted
has
always
offered
to
kind
of
be
a
person
so
to
try
to
help
some
with
that
I
mean
I,
don't
know,
I
think
it's
I
mean
people
can
already
read
between
the
lines.
Right.
I
think
this
is
certainly
a
a
blocking
point
for
a
lot
of
open,
telemetry
stuff,
and
so
it
needs
to
be
addressed.
B
I
think,
unfortunately,
that's
a
very
macro
problem
and
I
would
say:
Ted
seems
like
the
only
person
who's
active
enough
to
invest
time,
to
kind
of
try
to
resolve
these
issues,
so
I
guess
I
would
suggest
reaching
out
to
him
and
he's
offered.
You
know
he
said
like
if
something's
not
moving
forward.
Let
me
know
so.
C
Yeah
yeah,
so
that
is
unfortunate,
but
yeah
like
Connor.
If
you
were
of,
if
I
were
you
I
would
try
to
like
just
take
ownership
of
this.
It's
it's
one
thing
they
just
like.
Yeah
I
know
like
you
know,
open-source
community,
it's
like
this,
but
it's
just
try
to
find
the
right
people.
You
know
to
to
bother
and
eventually
you'll
get
this
soon,
so
yeah
I
guess.
B
I
wonder
too,
like
if
you're
I
mean
yes,
you
should
bug
people
so
that
this
gets
in
officially
becomes
approve,
doesn't
know
Ted,
but
I
have
also
previously
made
API
and
spec
change
or
API
changes.
No
making
assumption
that
things
that
are
obviously
highly
I
guess
approved
they're
going
to
make
their
way
in
eventually
so
I
would
consider
it,
and
what's
the
next
step
for
you
presuming
this
Oh
tip
gets
it
like.
Would
you
then
make
a
change
in
like
ot
Python
to
support
exemplars
yeah.
D
That's
basically,
the
concept
I
mean
so
exemplar
is
in
the
spec
or
blocked
by
the
SEK
specification
right.
So
because
I
need
to
build
off
of
that
to
do
exemplars
in
the
spec
sword,
work,
Anya,
the
Python
implementation
and
then
eventually,
yeah
I
mean
there's
something
before
the
end
of
my
internship
in
five
yeah.
C
Yeah
cuz,
normally
like
we
talked
during
the
maintainer,
is
meeting
like
the
normal
control
flow.
It's
like
you,
make
an
issue
in
the
specs
right
to
make
it
serious
and
people
like
actually
actually
talk
about.
Okay,
like
we
want
this
change
and
then
you're,
the
otep
proposed
an
otep,
that's
where
you
actually
have
the
discussion
and
that
old
temper
gets
merged,
and
then
you
make
a
PR
to
the
specs
which
makes
it
like
official.
C
You
know
like
the
the
source
of
truth
and
everything,
and
then
everyone
starts
implementing
across
language,
but
right
now
like
we're
missing
step
one
or
like
we're
missing
the
you
know
like
the
the
specs
part
of
it,
which
is
unfortunate,
so
I
think
what
you're
doing
is
the
right
play.
However,
I
can
foresee,
like.
C
C
D
A
D
B
B
I
mean
Connor
I'd,
really
just
altered
the
PR
on
the
Python
side
and
probably
paying
Ted
at
there
like
at
the
same
time
like
I,
don't
know
how,
like
maybe
you,
know
Lane.
This
is
probably
your
call,
but
I
personally
find
to
approve
things
that
you
know
represent
things
that
are
obviously
gonna
get
in,
because
it's
like
I
mean
I,
don't
know,
I,
guess
I'm,
trying
to
personally
balance
like
the
frustration
of
trying
to
get
this
giant.
B
You
know
trying
to
move
the
whole
community
one
step
at
a
time
with
these
attempts
and
also
enabling
individuals
to
have
impact
right
like
I
know,
Connor,
obviously
you're
an
intern.
So
for
you,
like
your
scope,
is
very
project
like
your
time
is
limited
anyway.
Delivering
a
project,
I
think
is
kind
of
the
goal.
B
D
So
there's
there's
one:
the
hesitation
I've
had
with
making
appear
is
a
lot
of
the
value
of
exemplars
where
it
comes
from
the
views
API
and
like
the
histogram
aggregator,
and
so
especially
for
Google.
It's
their
whole
implementation
of
stack
driver
is
exemplars
in
a
histogram
aggregator
show
up
in
like
in
their
product
and
without
the
the
views
API.
You
can't
really
do
that
right
and
so
I've
been
I
was
holding
off
like
probably
a
couple
months
ago
from
just
making
the
changes
in
the
Python
repo,
because
I
thought
about.
C
Yeah
I
think
having
given
the
timeline
yeah
like
having
something
tangible.
Even
if
it's
a
it's
dependent
on
something
else,
it's
not
complete
it's
better
than
nothing
right,
so
I.
B
D
D
B
Like
could
be
useful
even
today,
because,
for
example,
you
know
I
think
today
we
have
aggregators
that
exist
like
/
sort
of
like
specific
tag
like
combination
of
tags
or
label
sets
label
sets
grouping,
and
so
we
represent
of
spam
for
that
particular
label
set
and
the
metric
that
gets
generated
from
that
label
set
I
think
could
would
be
advantageous
thanks,
I'm
Curia,
yeah
I'll
go
look
at
the
hotel,
I'm
curious.
How
that
gets
handed
to
you
yeah.
D
I
mean
I'm,
definitely
I'm
sure
after
reading
all
the
stuff
that
it
is
useful
outside
of
just
that
small
Google
future.
But
it's
like
it's
definitely
a
future
concept
versus
something
that's
implemented
in
the
present
in
all
of
these
products.
Right
and
so
I,
don't
know
Josh,
but
is
some
way
that
it
was
moving
the
the
metrics
community
forward
to
some
extent
by
putting
this
into
the
metric
spec,
and
it's
telling
people
that
this
is
a
useful
feature
right,
yeah.
B
I
mean
I,
guess
I
from
one
well,
so
I
hear
the
fact
that
there's
less
value
without
views,
API
I,
guess
I'll
state
that
if
you
do
create
something
that
at
least
provides
what
you
can
around
the
SDK
for
this
exemplary,
like
these
employer
functionality,
I'll
approve
it
and
this
other
data
point.
The
other
thought
I
have
too.
B
Is
that,
like
you
know,
there's
like
the
agile
way
of
working
right
is
like
go
on
to
the
thing
that
blocked
you
and
go
help
them
block
that
thing
until
your
thing
is
unblocked,
so
you
know,
maybe
this
is
a
good
reason
for
you
to
kind
of.
Maybe
this
is
a
good
officer
opportunity
for
you
kind
of
look
at
views
and
see
what
you
can
do
to
help
move
that
forward
and
I'm,
not
sure
waiting.
If
that's
like
a
situation
where
a
single
person
can
help
move
the
needle
forward
or.
C
I
mean
like
any
any
more
eyes
and
like
hands
on.
It
would
be
helpful.
The
it
just
doesn't
address
the
underlying
problem
of
like,
similarly
to
how
the
SDK
specs
isn't
being
moved.
Well,
it's
being
moved
by
one
person
which
is
Josh
yeah
so,
and
this
is
like
a
big
player
no
offence.
Connor
like
really
pushes
this
then,
like
I,
will
have
the
same.
C
So
that's
just
like
move
the
issues
along
and
get
people
unblocked.
What
do
you
guys
think
sounds
good
me
yeah,
like
I'm,
pretty
sure
like
we
have
a
pretty
good
understanding
of
what
it's
supposed
to
do
and,
like
any
other
feature,
changes
would
be
probably
pretty
small,
so
I
don't
mind
just
like
merging
in
our
understanding
of
what
views
is
supposed
to
be.
B
Okay,
I
mean
I,
think
in
general,
I'm
on
board
and
I.
Think
the
other
compliment
here
is
that
there's
kind
of
some
flux
in
the
SDK
and
the
metrics
SDK
anyway,
so
you
know,
might
as
well
put
all
the
concepts
in
and
then
like
then
get
to
the
point
where
you
can
say:
okay
now
with
all
these
concepts
together,
here
is
the
challenges
that
we
run
into
rather
than
sort
of,
like
you
know,
get
some
high-level,
abstract
agreement
cut
the
code
in
and
then,
as
we
like,
add
more
functionality.
B
C
C
B
C
B
B
I
guess
the
only
I
guess.
Maybe
this
is
just
a
Diego
heads
up,
because
I'm
modifying
your
code
a
little
bit
I'm
just
moving
out
of
instrumentation
to
its
own
folder,
because
I
made
a
package
that
also
is
using
the
open
flow,
MIT
instrumentation
package
name.
So
I
had
an
issue
when
I
was
doing
was
the
word
Auto
instrumentation.
E
B
B
The
entry
points,
and
so
one
aspect
of
that
is
that
directory
has
to
be
pretty
empty
and
only
contain
a
site
customized
PI,
because
if
you
have
some
other
file,
so
Mike
situation,
I
had
starlet
as
instrumentation,
so
open
telemetry
that
instrumentation
that
startled
it.
Then,
when
you
try
to
import
starla,
it
looks
at
it
sees
the
starlet
folder
in
opens
lemon
tree
instrumentation,
because
that's
the
PI
s
method
to
the
Python
path
and
then
basically
has
a
recursive
import
error,
because
when
I
start,
instrumentation
is
also
trying
to
import
from
starlet.
B
And
so
this
is.
This
is
only
a
problem
now
because
I'm
trying
to
drive
a
change
of
renaming
our
extensions
to
instrumentations,
because
that's
sort
of
the
spec
defined
standard
now
and
so
to
ensure
that
we
don't
run
into
that
I'm
just
moving
the
directory
into
its
own
sub
module
site
customizes
in
the
autumn,
instrumentation
directory.
As
a
result,
there
will
be
no
conflicts
there.
All.
E
C
C
Cool
all
right,
second,
one
for
Conor
yeah
we
discussed
about
this-
is
the
you
know,
merging
the
exclude
lists
environment
variables
for
flask
us
for
our
Web
Apps.
You
covered
pyramid
as
well,
so
it's
pretty
good
I
took
a
look
at
it
if
we
could
get
another
approver
for
this
that'd
be
great
I,
don't
know
who
I
don't
know
who
this
guy
is
or
this
nice
girl.
C
B
I
started
reviewing
it
and
I
think
the
code
looks
good
to
me.
The
cut
the
conversation
I
was
gonna
bring
up.
Was
you
know
it
is
possible
to
get
pretty
identical
logic
for
like
excluding
URLs
or
really
whatever
by
attaching
like
well
I.
Guess
not
yet
so
everybody
I
was
gonna.
Argue
that
you
can't
add
a
sampler
to
do
this
for
you,
but
then
I
realized
that
the
sampler
and
provide
the
hooks
necessary
to
update
the
paths.
B
C
B
B
You
may
want,
for
example,
something
a
lot
more
complex
around
the
like
the
matching
of
the
URI
itself
and
so
I'm
an
argument
there
is
that
if
we
can
provide
a
more
flexible,
sampler,
API
and
basically
we'll
still
incur
the
cost
of
creating
these
spans,
but
what
will
happen
instead
is
we
won't
pass
them
along
to
be
exported
and
we
get
more
functionality,
and
then
we
also
don't
have
the
need
to
go.
Add
this
functionality
everywhere.
So
it's
every
instrumentation.
Rather
so
so
that's
my
two
cents.
C
B
Yeah
yeah
so
I
think
just
for
the
just
a
pad
a
point
of
conversation.
It's
clear
that
this
PR,
if
Merce,
would
basically
just
not
create
spans
at
all.
So
there's
you
know,
there's
an
advantage
that
you
minimize
some
of
the
overhead
span
creation.
Now,
if
we
did
this
the
sampler
way,
there
is
always
going
to
be
some
level
of
overhead
with
that
at
minimum
right.
B
In
a
practical
scenario,
there
will
also
be
a
cot
most
of
the
time
because
you
don't
have
access
to
the
HTTP
route.
Well,
I
mean
this
is
there's
an
interesting
nuance
here,
but,
aside
from
that,
the
HTTP
route,
you
will
have
access
to
the
HTTP
route
until
kind
of
like
halfway
through
your
processing
of
the
request
itself.
Most
likely,
you
will
always
create
a
real
span,
quote
unquote,
you
know,
there's
the
other
option
creating
a
default
spam
which
does
nothing
which
theoretically
has
lower
overhead,
but
practically
doesn't
really
right
now,
but
you
can
create
this.
B
You
always
have
to
create
a
real
span
and
then
you
will
always
have
to
have
that
span,
collect
attributes
up
until
the
path
and
then
it'll
stand,
and
at
that
point
you
can
basically
make
okay
I.
Don't
care
about
this
path
right
and
and
from
that
point
on,
you
know,
the
span
object
will
live
in
memory,
it'll
go
through
the
end
cycle,
but
what
will
happen
there?
Is
that
because
a
sampling
decision,
it's
like
the
sat
you
know,
sampled
flag-
is
false.
When
hits
the
processor
or
sorry
it
will
never
hit
the
processor.
B
C
I,
don't
know
for
my
understanding
of
how
sampling
was
implemented
is
incorrect
or
what
I
always
thought
it
was
just
like
the
span
was
created
it.
It
checks,
it
checks
like
the
trace,
ID
or
something
that
that
was
made
in
like
there's
that
whole
random
thing
and
compares
it
with
the
percentage
and
then
creates
a
no
op
or
like
a
default
span,
instead
of
a
real
span
to
indicate
whether
it's
I
thought
that
was
the
information
yeah.
B
B
This
is
the
way
the
current,
the
current
implementation,
or
so
whatever
day
that
you
have
in
the
creation
of
the
span,
whether
that's
the
span
name,
if
it's
correct,
send
span
attributes
as
well
as
like
eye
information
about
the
spanner
trace
idea,
is
passed
in
and
a
sampler
can
make
a
decision
based
off
of
those
variables.
The
probability
sampler
is
what
you're
referring
to
when
you
say
like
it,
looks
at
the
span
or
trace
ID,
and
it
makes
it
a
decision
of
whether
they're
properly
at
that.
B
C
B
I
mean
the
sampler
API
was
always
designed
to
allow
more
complex,
samplers
in
mind
right,
like
that's
the
reason
why
consumes
more
than
just
the
trace
ID
or
the
trace
like
context
yeah
this
Bank
context,
so
you
know,
there's
always
a
simplicity
idea
that
okay,
we
can
do
something
like
actually
look
at.
You
know
the
URL
itself
or
various
attributes
on
this
band
and
make
a
sampling
decision
based
off
of
that.
C
Right
but
the
design
that
you're,
referring
to
it
like
that,
would
so
like
I
guess
solve
this
issue
like
it's
not
there
yet
right,
like
our
sampler,
isn't
like
that,
yet
in
which
it's
it
does
intelligent
sampling
based
off
of
actual
properties
right.
Is
that
what
you
mean
by
like?
Oh,
we
can't
really
do
this
yet
like
this.
Is
that
what
you
said
that
yeah.
B
So
well
so
anyway,
I
think,
there's
sorry,
some
I'm
late
and
I'm
drawing
Billy
of
the
land.
But
there
is
some
nuances,
and
this
is
where
my
I'm
filing
an
attempt
to
try
to
kind
of
change
the
way
that
we're
doing
things
but
I'll
see
what
the
heck
okay,
interesting
sorry
I'm,
trying
to
send
a
question.
B
D
B
I
do
agree,
I,
don't
you
know,
I,
don't
have
a
crystal
ball
and
in
terms
of
how
this
stuff
goes,
but
based
on
what
I've
seen
like
oftentimes,
you
have
to
be
very
judicious
about
the
functionality
that
you
expose
and
like
often
like.
Basically,
what
you're
doing
here
is
prescribing
a
very
specific
use
case
that
any
like
very
specific
API
that
enables
a
specific
set
of
use
cases
right.
B
C
C
Like
me
personally,
like
I,
wouldn't
mind
just
merging
this
in
and
getting
a
B,
because
right
now,
like
Connor,
like
this
issue,
that
he
brought
up
like
the
way
that
we
do,
it
is
actually
like
it
doesn't
cover
a
lot
of
use.
Cases
like
this
is
an
important
change,
regardless
of
how
we
implement
it
so
like
for
now.
I'd
rather
have
this
into
actually
fix
a
problem,
rather
than
kind
of
like
stew,
and
like
doing
it
exactly
how
like
we're,
probably
going
to
be
doing
in
the
future,
maybe
not
at
all.
B
C
C
D
C
All
right
wait,
this
one
I,
don't
really
know
who
it's
spread
to
P
is.
Is
he
on
the
call
I,
don't
think
I,
don't
think
he
is
fixed
serializing
a
tuple
elder
than
my
core
into
a
string.
I
think
this
was
pretty
straightforward
yeah.
This
was
for
the
ATA
exporter,
so
yeah.
If
we
get
some
some
eyes
on.
This
looks
like
a
pretty
simple
change:
you'd
have
to
like
be
knowledgeable
but
eager
to
test
this
I
guess,
but
that's
pretty
much
it
I.
C
C
Yeah
great
yeah
on
a
side
note
if
we,
if
we
come
across
these
PRS
like
during
life
from
these
people
that
aren't
in
the
sig,
is
there
any
point
in
bringing
them
up
if
they're
like
like
this,
one
was
pretty
easy
because
it
was
straightforward,
but
like
next
time,
if,
like
someone,
has
like
a
pretty
big
PR,
and
it's
not
like
something
straight
for
like
an
instrumentation,
how
should
we
go
about
it
in
terms
of
like
to
sleep
meetings,
I
guess.
C
Cool
all
right
next
one,
since
it
sets
span
status
on
whiskey
errors,
all
right.
This
is
the
what
the
hell
okay,
this
is.
The
original
issue
looks
like
alex
is
already
left
some
comments,
so
I
guess
I'll
be
just
waiting
on
him
very
straightforward
change,
disliked
to
do
when,
like
a
span
errors
out
or
something
you
know,
we
didn't
set
the
status
properly,
so
it
looks
like
everywhere
else
is
setting
it
to
internal
I.
Think
that
meant
like
something
happened
that
would
like.
We
didn't
know
what
it
was
really
so
we
kind
of
encapsulating
this.
C
C
C
B
Was
raised
in
a
git
er
discussion
if
I,
recall
and
I
think
the
concern
was
this
exposes.
This
can
potentially
expose
PA
personal,
identifying
data,
API,
I
and
so
I
have
not
had
a
chance
to
look
at.
You
know
the
ramifications
of
that
choice
and
not
capture,
query
parameters,
but
you
know
that's:
that's
the
Y.
That's.
C
B
C
F
C
F
At
otech
gaming,
Sinha's,
very
I,
basically
just
formalizes
how
resource
destruction
should
be
done
so
hasn't
been
emerged
yet
I'm,
just
kind
of
making
a
prototype
based
off
of
the
changes.
F
Assuming
that
everything
nothing
in
here
will
change,
which
I
guess
is
not
too
great
an
assumption,
but
I
guess
it's
pretty
easy
to
change
in
the
future,
if
maybe
but
yeah
there's
two
big
parts
of
the
PR
one
is
just
resource.
Auto
detection
right
now,
I've
only
added
it
for
GCP
and
I'm.
Looking
for
people
to
add
it
for,
like
other
companies
like
Amazon
and
other
part
of
ads,
is
that
for
the
existing
exporters
it
actually
uses
that
information
which
newly
gets
populated.
F
So,
for
example,
like
a
lot
of
the,
if
you're
in,
for
example,
a
GCP
instance,
you
need
to
do
like
call
us
a
specific
URL
or
like
looking
at
specific
environment
keys.
It
basically
just
makes
that
process
all
automated,
so
the
the
user
just
calls
a
function
that
they
pass
on
a
bunch
of
detectors
and
then
it
calls
all
the
detectors,
sequentially
and
emerges
them
in
terms
of
priority.
And
then
you
get
a
big
resource
object
in
return
which
the
user
should
then
power
on
to
the
provider.
I
see.
F
C
B
B
F
Think
kind
of
the
the
thought
process
was
like
this
there's
not
a
lot
of
things
that
people
are
contesting
too
heavily
emotes
up,
I,
guess,
okay,
so
I
think
that
this
you
know
prototype
is
basically
very
close
to
what
a
real
implementation
should
look
like:
okay,
yeah,
yeah.
C
Yeah
seems
like
a
common
thing
that
we
got.
You
know
like
implementing
this
with
us
about
the
old
tips
going
in,
but
yeah
I
think
it's
safe.
Look.
You
have
a
pretty
good
understanding
of
what
this
is
supposed
to
do
and
any
changes
that
they
might
make
in
the
future.
You'd
probably
know
how
to
do
it,
regardless
cool
all
right,
makes
sense.
C
B
C
B
B
E
This
genius
can
can
be
merged,
I
mean
they
actually
fix
the
problem,
but
estimation
they
don't
deal
with
the
the
root
problem
is
that
is
that
some
configuration
like
the
setting
of
trace
the
trace
for
provider
or
spam
processors
can't
be
done
right
now,
the
the
configuration
object
job
that
is
done,
that
that
should
be
every
problem
that
we
have
everything
awesome.
C
C
In
not
like
what
you
guys
think
would
be
like
the
best
like
what
we
should
implement
for
Python
like
Microsoft
side.
We're
actually
have
ran
into
some
of
these
issues
too,
in
which
we're
only
exploring
it
sorry
exposing
an
exporter,
but
there's
so
much
like
you
know
stuff
that
we
allow
users
to
choose
like
to
enable
or,
like
you
know,
to
set
that
putting
in
the
exporter
doesn't
really
make
sense
like
something
that
we
own
yeah
yeah
I'm
interested
in
what
you
guys
think
I.
E
E
For
example,
for
example,
to
set
spam
processors
and
exporters
the
environment
variables,
because
you
are,
you-
have
to
mix
and
match
the
spam
processors
with
exporters
that
you
want
so
I
think
these
were
required,
a
more
powerful
approach,
the
it's
probably
a
situation
file.
It
will
be
nice
equivalence
between
what
can
be
done
with
a
federation
file
and
be
an
environment
of
our
rules
where
I'm,
not
sure,
that's
going
to
happen
and
I
also
think
that
these
will
affect
every
implementation
of
open
telemetry.
So
this
is
probably
something
that
escalate.
B
Right
I
mean
honestly,
there's
probably
a
simple,
a
really
simple
thing
we
could
do
here.
Oh
we
just
have
like
I,
don't
know
an
open,
telemetry
configurations
out
Pyatt
or
something
and
just
modify
the
other
instrumentation
to
see
if
that
directory
exists,
relatively
the
users
current
directory
and
then
just
load
that
before
you
load,
the
instrumentations
right
I
would
be
a
simple
hack
which
I
haven't
thought
through
the
whole
ramifications
of
I'm.
Doing
that,
but
I,
don't
know
how
it
kind
of
you
were
thinking
you
go
or.
E
B
Mean
well
so
I
guess:
there's
multiple
you
maybe
you're,
considering
like
a
markup
file,
but
the
place
where
you
would
inject
either
way
the
place
that
you
would
inject.
That
type
of
configuration
code
would
be
in
the
site
customized
out
pi
before
you
start
reading.
Through
all
the
entry
points
and
running
the
load
function.
A
E
E
C
C
C
C
C
Then
I
don't
have
any
discussions
there.
Yeah
sounds
good,
sounds
good.
All
right,
cool
I
think
that's
pretty
much.
It
anyway
thing
else.
Anyone
wanna
shout
out
before
we
terminate
all
right.
If
that's
it,
then
cool
I
think
that's
it.
For
today,
all
right,
cool
elves
see
you
guys
next
week,
peace
out.