►
From YouTube: 2021-02-25 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
How
are
you
well?
Thank
you
all
for
joining.
I
think
I'll.
Let
some
ladies
will
be
joining
shortly
as
usual.
Just
please
add
your
names
to
the
attendees.
A
A
A
Also
here,
if
you
have
any
topics,
poor
requests
or
issues
that
you
want
to
discuss,
please
ask
them
to
be
document.
Everybody.
C
A
Hey
later,
I
was
just
doing
folks.
D
Cool,
I
guess
we
could
just
get
started
alex,
would
be
here
in
a
bit
so.
D
D
For
those
of
you
who
were
here
last
week,
I
think
carlos
dropped
by
and
we're
going
to
be
trying
to
do
the
what's
it
called
the
you
know:
ga
release,
tc
review
process,
we're
trying
it
out
with
python,
and
carlos
has
been
working
hard
last
week
to
you
know,
come
up
with
a
bunch
of
issues.
You
know
just
compare
our
api
sdk
surface
to
see
if
it
complies
exactly
with
the
specs.
D
You
create
a
bunch
of
issues,
so
we're
going
to
be
trying
to
prioritize
this
for
our
rc2
release
we
already
kind
of
added
it
to
the
project
board.
So
please
consider
like
picking
some
of
these
up,
because
we
are
going
to
be
prioritizing
these
over
other
other
things.
D
All
right,
cool,
sorry
did
someone
say
anything
aaron.
You
said
no
question:
okay,
cool
cool,
yeah,
nice.
So
I
guess
someone
someone
else
added
intros.
So
I.
E
E
Yeah,
I
think
I
think
we
we
missed
doing
intros
the
last
couple
times.
New
people
showed
up,
so
I
I
thought.
Maybe
we
could
just
like
go
around
and
say
a
quick
hello
for
everybody,
I'm
happy
to
start
so:
hey,
I'm
alex
bowden.
I've
been
on
the
open,
telemetry
python
project
for
a
little
bit
over
a
year
and
a
half
almost
a
year
and
a
half.
Now
I'm
I
work
with
lightstop
and
yeah,
I'm
just
I'm
one
of
the
maintainers
excited
to
see
every
everybody
here.
D
Can
go
next,
hi
guys,
I'm
leighton.
I
work
at
microsoft
and
I've
been
work
working
with
folks
at
open
telemetries
for
almost
two
years
I
would
say-
and
I'm
also
one
of
the
maintainers,
so
you
know
feel
free
to
reach
out.
If
you
guys
have
any
questions,
you
know
concerns
about
the
process
and
stuff,
so
yeah
welcome.
A
Okay,
well
I'll
go
next.
My
name
is
diego.
I
work
at
lightstop
and
yes,
I've
been
working
this
project
for
about
a
year
and
a
half.
E
B
I
guess
I'll
go
hi
everyone,
I'm
aaron.
I
work
at
google.
I've
been
working
on
opencountry
for
a
little
less
than
a
year
like
10
months
or
so
I
am
an
approver
stop.
B
Oh
this
happens.
Okay,
sorry.
B
Yeah
yeah,
I
thought
I
said
you
know
the
words
yeah
those
words
so
yeah,
I'm
a
approver
on
python
and
I've
done
a
little
work
in
the
metric
sig
and
I
maintain,
like
our
google
cloud
exporters
for
javascript
and
python
separately
from
these
repos
and
yeah.
Also
me
and
leighton
are
maintaining
some
open
census
stuff
as
well.
If
you
guys
have
any
questions
about
that,
so
yeah
feel
free
to
reach
out
and.
D
G
Nice
thanks,
hey
everyone,
I'm
os
from
splunk
I've
been
around
open
telemetry
and
before
that
open
census
for
a
while
and
contributing
different
projects,
mostly
active
in
python
and
collector
groups.
Right
now,.
H
F
How
about
I
go?
My
name
is
michael.
I
work
at
jw
player,
we're
not
involved
in
metrics
as
such,
but
I'm
a
user
of
this
doing
a
lot
of
work
on
grpc
internally,
and
so
you
know
when
I
started
looking
into
what's
available,
I
ended
up
doing
a
lot
of
bug,
fixing
and
stuff
on
the
contrib
repository
kind
of
really
interested
in
getting
the
stuff
in
production
for
us.
F
Yeah
alerted
snake
nice.
Thank
thanks.
D
I've
seen
a
lot
of
your
viewers,
yeah
yeah.
C
E
D
A
Yeah,
oh
by
the
way
I
have
to
leave
in
15
minutes
sure.
A
Yeah
that
issue
and
and
one
beyond,
okay.
Yes,
so
I
was
running
some
tests
this
morning,
I
realized
when
you
run
stuff
with
windows.
5
you
get.
This
message
that
says
will
has
reached
the
end
of
his
life
september.
13,
2020
and
also
pip
21,
will
drop
support
for
305
in
january
2021.
So
both
those
events
have
already
happened
and
thrill
5
is
that
so
I
think
we
did.
We
went
through
this
process
before
when
we
decided
to
drop
3.4.
A
D
I
think
this
could
be
starting
the
conversation
and
like
just
discussed
offline,
and
you
have
discussions
somewhere.
A
Yeah,
that's
not
yeah,
spend
that
much
time
cool
all
right,
I'll,
open
up
the
the
thread
there
and
everybody.
Please
leave
your
comments
and
opinions
well.
Next,
I
have
this
pr.
I
just
wanted
to
this
refactor
a
few
kills.
I
think
this
is
the
last
one
there
yeah.
A
I
just
wanted
to
clarify
something.
This
pr
is
okay,
there's
another
pr
that
depends
on
this
one,
but
this
pr
does
not
depend
on
any
other
pr.
Okay,
so
this
one
can
be
reviewed
as
it
is.
Please
do
the
I
have
a
mark,
the
pr
that
depended
on
this
one
as
draft
until
this
one
gets
married
or
anything
else,
because
I
I
understand
that
this
causes
some
trouble
from
the
people
who's
reviewing.
B
A
A
bunch
of
commits
that
some
other
related
that
we
are
sure
of
them
so
just
wanted
to
take
a
few
moments
to
explain
that.
A
And
yeah,
please
ask
you
to
keep
reviewing.
I
think
later
on
that
you
already
did.
I
answered
your
comment
there.
Also,
this
vr
has
an
unassociated
country
reboot,
pr
that
also
it's
reviewing,
and
this
could
help
get
us
done
with
the
utils
in
space.
D
Yeah
deek,
I
just
I
really
want
to
get
this
in
I'm.
I
was
just
kind
of
confused
that's
to
like,
because
we
did
discuss
this
on
gitter
and,
like
I
thought,
the
decision
was
to
like
not
have
a
util
folder
or
like
have
this
name
space,
because
we
won't
have
anything
in
it
right,
because
we
don't
provide
anything
as
a
public
method
for
people
to
use
in
the
util
namespace
anyways
right.
A
Yes,
there's
only
one
catch
the
types:
dot,
pi
module,
that's
right
there
in
new
hills,
I
mean
I'm
okay
with
the
having
not
having
youtubes,
I
mean
the
less
things
we
have
in
api,
the
better
yeah
but
yeah.
That
question
needs
to
be
answered
right.
Where
do
we
put
types
that
by
because
is
types
being
used
somewhere.
A
Yeah,
it
is
being
used
to
that's
like
I
think,
for
example,.
D
Yeah,
but
we
don't
want
to
expose
this
as
like
a
utility
function
or
anything
right
to
actual
users
we
just
want
to
like.
I
guess
I
guess
in
a
way
like
if,
if
this
is
even
needed
anymore,
like
we
could
just
put
this
in
where
it's
used.
If
it's
only
used
in
one
place,
oh
okay,
I.
D
Right
now,
but
the
only
like
honestly,
like
whatever
solution
you
come
with,
like
whatever
is
put
under
open,
telemetry,
util
namespace,
just
has
to
make
sense
as
something
we
want
to
expose
to
users
as
a
whether
it
be
a
function
or
be
a
type
right.
A
Just
to
answer
your
question:
yes,
types
is
being
used
in
the
sdk,
I
mean
in
the
trace,
init
module.
We
do
from
types,
import,
mapping,
proxy
type
and
facebook
type
and
check
it.
If
we
do
the
same
in
complete.
D
Oh,
that's,
that's
fine!
It's
like
if,
if
we
just
have
types
like
it
doesn't
have
to
exist
in
the
util
folder
right.
I
really
want
to
reserve
this
as
like
something
that
people
use
as
like.
Okay,
we
support
this
as,
like
some
convenience
function
or
some
some
utility
function.
You
know
like
that's
what
the
that
should
that
namespace
should
have
right
sure.
A
Just
I
think.
D
Opentelemetry.Type.Types.Pi
folder,
you
know,
okay,
that's
what
you
mean
all
right.
Yes,.
A
Okay,
yeah:
the
types
module
is
also
used
in
the
sk-11
instrumentation.
So
it
is
it's
something
that
so
far
needs
to
be
in
the
in
the
api.
If
you
want
to,
I
can
I
can
refactor
that
and
sort
of
remove
bluetooth
and
just
add
types
if
you
prefer.
D
Yeah,
I
don't
mind
it
being
in
the
api.
I
guess
just
underneath
this
namespace
it
just
because
I
think
we
have
a
util
folder
for
another.
D
I
think
the
contrib
repo
right
and
that's
purely
for,
like
utility
stuff,
so
I
think
it
would
make
sense
for
all
of
the
stuff
under
this
namespace
to
be
a
utility
function.
I
guess.
A
D
What's
going
on
with
metrics
plus
re
yank
anyways,
does
anybody
else
have
any
more
comments
on
that
pr
before
we
move
on.
D
D
G
So
I
I
did
that
I
I
made
an
assumption
that
it
doesn't
use
packages
from
wi-fi.
I
don't
know,
if
that's
the
case
or
not,
but
last
week
we
had
a
couple
of
instances
where
country
packages
were
published
to
pi-fi
and
they
were
broken.
G
I
assume
the
ci
job
for
publishing
packages
tested
with
core
main
branch
and
it
had
deviated
from
what
was
yeah
published
in
background.
So.
G
If
we,
if
you
don't
use
the
main
branch
from
core,
but
instead
just
use
the
dependencies
mentioned
in
the
in
the
packages,
I
guess
that
should
slow.
This
issue.
D
Yeah,
so
oh
wait,
you're
correct
in
assuming
that
the
the
release
job
does
depend
on
the
core
main
branch,
or
at
least
the
release
branch
of
the
core
right,
and
I
think
we
designed
it
that
way
because
we
were
like
whenever
we
release
core,
we
would
do
a
subsequent
release
in
contrib
and
the
only
reason
why
it
didn't
work
out
last
week
was
because
I
merged
in
something
last
minute
without
running
the
build
again,
and
so
that
was
like
a
personal
problem.
D
I
think
the
ci
job
does
exactly
what
we
need.
If
we
do
the
process
properly,
albeit
it
does
leave
a
lot
of
room
for
manual
error.
I'm
thinking
like
the
the
the
problem
with
contrib
the
build
the
release
ci
job,
relying
on
the
core
package
is
that,
like
I,
would
have
to
have
already
released
core
for
this
to
work
right,
and
I
can't
do
that
if
the
the
core
build
doesn't
pass
on
stuff.
That
is
relying
on
the
contrib
to
be
passing
right.
D
Yeah,
I
would
always
have
to
like
create
two
pr's
two
release
pr's
at
the
same
time,
because
we
have
this
whole
ecosystem
of
like
depending
on
each
other
and
trust
me
it
was
it
was.
It
became
way
more
evident
that
it's
annoying
when
we
tried
to
release
rc1
because
in
the
core
package,
just
rc1
components
and
there's
0.18
components,
and
that
was
super
annoying.
But
it's
something
that
we
signed
up
for.
D
It's
it'll
be
relatively
easier
now,
as
we
don't
have
as
many
merges
into
1.0
as
we
go
into
a
more
stable
state,
but
for
now
yeah
there's.
There
is
a
lot
of
room
for
error
and
I
guess
like
as
maintainers
alex-
and
I
are
just
trying
to
be
careful,
so
yeah
yeah.
G
D
Nice,
nice
yeah,
I
I
wish
there
was
like
some
more
defense,
I
guess,
but
by
the
time
I
think
we
implement
this
like
it
would
be
like
diminishing
returns.
You
know
like,
as
we
become
more
stable,
so
yeah
that
makes
sense.
D
All
right
cool,
so
next
up
aaron,
you
got
some
discussion.
What's
going
on.
B
D
Oh
awesome,
yeah
yeah,
we'll
be
we'll
be
active
on
this.
Don't
worry,
aaron
slack.
B
Yeah,
so
I
think
we
brought
this
up
in
getter
and
then
diego
filed
this
issue
as
well,
so
we
can
discuss.
The
second
one
is
from
is
the
actual
open,
telemetry
community
issue,
and
it
looks
like
the
idea
is
that
we're
moving
off
of
gitter
sort
of
regardless.
B
Yeah
I
want
that,
but
yeah
so
I
mean,
like
I
think,
oa
brought
up
the
point.
That
slack
is
not
indexed
very
well,
so
it's
hard
to
search
for
issues
and
find
that
find
the
information
I'm
just
I'm
not
sure.
If
there's
like
you
know,
if
we
have.
D
I
never
used
it
like
that,
but
yeah.
I
guess
that
is
a
good
point.
B
Yeah,
sometimes,
if
you
just
google,
like
an
error
message
or
something
you
might
find
a
getter
discussion
relating
to
it,
which
is
right,
which
is
definitely
nice,
but
I'm
not
sure
if
we
really
have
a
choice.
I
guess
what
I'm
trying
to
say.
I
don't
know,
there's
no
timeline
exactly
for
one
to
switch
it,
but.
A
Yeah,
on
the
other
hand,
I
will
say
that
having
technical
discussions
in
greater.
A
It's
not
the
best
practice
because
finding
stuff
in
guitar
it's
not
that
convenient
right,
so
I
will
prefer
if
we
track
any
technical
discussion,
whether
in
discussions
in
github
or
by
opening
it
up
and
up
bringing
up
issues
yeah
instead
of
discussing
them
in
in
gear.
So
even
when
this
could
be
seen
as
a
lost
feature
right,
because
we
can't
it's
not
indexed,
it
may
move
us
to
better
practices.
B
B
D
D
Yes,
here
that,
like
slack,
should
be
for
like
random
dms
and
entry
points
and
then
we'll
redirect
them
to
a
github
discussion.
If
it's
like
a
lengthy
topic,
I
think
that's
the
idea.
A
Well,
so
the
other
thing
that
can
be
considered
is
the
I
mean
if
you
want
just
to
have
slack
for
the
ends
is
not
to
open
channeling
slack
and
just
let
people
know
if
you
need,
I
don't
know
some
help
with
some
pr
issue
or
something
dm
us
directly
via
slack.
A
Since
I
think
I
don't
know,
I
might
be
completely
here,
but
I
am
assuming
that
most
of
us
already
use
like
so
we
don't.
We
won't
need
to
keep
checking
dealer
also,
at
the
same
time,.
B
A
Right
yeah,
as
long
as
we
are
disciplined
with
ourselves
and
we
move
lengthy
technical
conversations
to
github
discussions,
I
think
we'll
be
all
fine.
B
Not
that
I
see,
but
a
lot
of
I
don't
know
if
you
guys
have
been
on
the
cnn
cncf
slack
yet,
but
there's
the
ones
I've
seen
so
far
like
channel
wise,
there's
like
open,
telemetry
hotel,
collector
hotel
go
it's
ljs
specification.net
logs,
like
seems
like
a
lot
of
people,
moved
over
cp
people's
plus
java,
prometheus
ruby,
etc.
So
I
was
kind
of
surprised
we
didn't
have
a
channel
yet.
D
Cool
yeah-
I
guess
we
should
get
on
this-
is
there
anything
that
we
need
to
do
in
terms
of
migration?
D
Like,
I
think
sorry,
sorry,
I
think
we
already
replaced
a
lot
of
the
getter
related
stuff
on
our
docks.
Is
there
anything
else
that
we
need
to
do?
D
A
A
A
Oh
folks,
I
have
to
leave
good
to
see
everybody
see
you
see.
D
Okay,
cool
anyone
else
have
any
opinions
on
that
topic.
D
Awesome
all
right
next
aaron,
another
one
nice.
What's
going
on
with
metrics
plus
riley.
B
Yeah,
so
I've
been
in
the
metric
state
just
listening
in
and
it
sounds
like
riley
did
a
prototype.
I
don't
think
he's
open
to
pr,
but
it's
basically
like
a
sample
app
to
sort
of
highlight
like
the
use
cases
for
metrics,
and
I
guess
what
I
heard
was
python
and
I
think
it
was
java
are
going
to
be
used
to
sort
of
prototype
metrics
going
forward.
B
I
don't
I
see
riley's
not
here.
I
know
he's
been
pretty
involved
in
python
before
and
obviously
works
at
microsoft.
So
yeah
I
was
wondering
like
I
didn't
know,
because
I
missed
the
last
two
sick
meetings.
If
there
was
some
update
on
that.
D
Yeah,
so
basically,
so
I've
been
working
closely
with
riley
on
his
prototype.
He
he
is
spearheading
the
python
example
or
the
prototype
and
victor
from
microsoft
is
doing
the
dotnet
one,
and
I
think
I
don't
know
who
was
doing
the
java
one,
but
it's
like
in
progress.
Yeah,
it's
joshua,
google
yeah.
I
think
josh
is
doing
it.
So
the
progress
for
that
as
of
last
week,
we're,
I
think,
we're
still
making
like
they're
still
in
prs
java.
Isn't
a
pr
done?
D
Isn't
a
pr,
I
think
like
riley's
is
already
finished,
but
I
don't
think
he's
put
up
a
pr
yet
so
in
terms
of
timeline,
I
don't
think
there
has
been
much
development.
I
think.
Last
week
we
just
talked
about
other
api
related
issues,
but
hopefully
definitely
I
foresee
an
update
in
today's
metrics
meeting
so
and
I
don't
know
if
you've
seen
the
roadmap
or
the
timeline,
but
it's
pretty
aggressive
yeah.
I
think
we
expect
something
like
an
api
prototype
or
something
by
march
end
of
march
and
then
for
may.
D
We
expect
like
an
experimental
sdk,
so
we'll
see
if
that
is,
can
be
done.
But
I
appreciate
riley
doing
this
because,
like
honestly
like
before,
metrics
was
kind
of
in
a
limbo,
yeah
very,
very
slow,
very
high,
very
specific,
deep
discussions
about
like
prometheus
and
everything,
and
we
were
very
not
headed
in
the
right
direction.
So
hopefully
this
will
make.
B
Good
strides
agreed,
so
I
guess
I
was
wondering
like
in
terms
of
prototype
you're,
mentioning
api
prototype
by
the
end
of
march,
like
we
already
have
a
api
prototype
in
theory
like
it's
more
or
less
following
the
spec
that
we
have-
and
I
know,
there's
been
some
discussion
about.
B
D
Yeah,
so
for
right
now,
yeah,
I
don't
want
to
say
it,
but
like
we
kind
of
are
kind
of
starting
over,
obviously
there's
going
to
be
a
lot
of
components
that
are
still
existing,
but
I
think
the
purpose
of
this
prototype
is
like
start
with
the
customer.
Examples
like
the
most
common
examples,
come
with
a
super
set
of
things
that
we
want
to
be
able
to
do
and
then
see
if
our
current
apis
can
do
that
right.
That's
why
they're
prototyping
with
our
current
apis.
D
Obviously
that
brings
up
a
lot
of
issues
because,
like
as
there's
no
sdk
specs
right
now,
a
lot
of
people
implemented
across
languages,
different
metrics
implementations,
and
that
is
definitely
a
concern.
But
I
think
this
is
just
to
start
the
discussion.
So
hopefully
there's
not
going
to
be
a
lot
of
changes
to
our.
D
You
know
api,
but
it's
possible
right
because
nothing
is
marked
as
stable.
Yet
in
terms
of
like
what
we
can
do
to
contribute
definitely
be
active
and
join
the
metrics
meetings,
because
it's
not
like
it's
not
like
gonna,
be
like
the
metrics
meetings
that
we
had
before
anymore,
where
you
kind
of
just
join
and
tune
out,
because
it's
just
bogdan
and
josh
arguing
it's
we're.
Gonna
be
moving
pretty
fast
now
so.
D
Yeah,
I
think
they
changed
the
schedule
like
four
times
now,
but
I
think
it's
finally
like
kind
of
settled
so
yeah
yeah
other
than
that.
It's
whenever
there's
like
a
you
know
like
a
pr
or
something
or
a
specs
that
is
being
raised,
be
active
in
it
right
now.
I
think
that's
the
best
way
to
push
progress
along
in
terms
of.
D
In
our
python
repo,
we
don't
have
to
do
anything
just
yet
just
see
the
results
from
the
current
prototypes.
So.
B
So
I
mean
correct
me
from
wrong:
it
sounded
like
it
sounded
like
the
metrics
were
going
forward.
The
idea
was
to
always
have
a
prototype,
and
then
we
were
going
to
submit
oteps,
but
but
it
sounds
like
in
that
case,
for
any
like
proposed
feature,
the
idea
is
to
have
a
prototype,
so
it's
more
concrete.
We
can
see
some
code
right.
D
Yeah
similar
to
what
we
did
for
views
right
but
yeah,
now
that
we
have
the
right
direction,
I
feel
like
that
will
actually
happen
instead
of
us
prototyping
views,
but
nobody
cares
about
it.
Well,
people
care
about
it,
but
nobody
prioritized
it
because
it
wasn't
the
right
time.
I
guess.
D
Cool
any
more
questions
on
metrics.
G
This
yeah,
so
the
implementation
is
very
straightforward.
So
this
basically
instruments
the
standard
library,
logging
module
and
injects
tracing
context
into
the
into
log
log
records
and
then,
if
the
user
wants,
they
can
set
their
log
format
in
a
way
that
makes
use
of
those
injected
variables.
So,
as
an
end
result,
users
can
have
span
id
trace
id
and
service
name
in
their
logs
and
then,
when
they
export
logs
and
traces,
they
can
have
login
trace
correlation.
G
But
but
this
is
opt
in
because
it
by
opt-in.
It
sets
the
logging
format
for
the
user
if
they
want.
So
this
is
helpful
when
people
use
open
telemetry
instrument
one
and
they
don't
want
to
add
any
custom
code.
So
with
the
so
this
this
just
chooses
a
logging
format,
arbitrary
login
command,
but
it
also
exposes
an
environment
variable
to
let
users
specify
their
own
login
format
or
by
default.
It
doesn't
set
the
format,
then,
in
that
case,
users
can
call
logging.basicconfig
and
set
a
format
themselves.
G
So
I
think
that's
that's
pretty
much
it
like
the
overview.
I
have
a
lot
more
detail
in
the
docs
here.
G
One
thing
probably
I
need
to
call
out
is
so
the
dogs
mentioned
the
logging
format
in
a
few
places
and
in
the
pull
request
it
was
called
out
that
it
might
be
hard
to
keep
it
in
sync
if
the
format
changes
so
instead
of
doing
doc,
strings
like
we
normally
do,
I'm
assigning
the
strings
to
underscores
or
dark
underscore
score
variable
and
then
doing
variable
substitution
in
the
string.
So
that
way,
it'll
always
be
in
sync.
So
it's
it's
a
bit
unusual.
G
G
Yeah,
so
if
we
change
the
default,
it's
default
log.
D
D
Cool
yeah,
this
looks
this
looks
awesome.
I
understand
like
what
this
does.
I
was
just
wondering
like
is
this
the
kind
of
path
that
we
want
to
tell
users
to
do
in
terms
of
like
this
is?
Is
this
is
this
like?
Will
this
cover
the
logging
pillar
for
open
telemetry?
D
G
What
this
address
is,
but
no
I
don't
know,
I
don't
think
so
so
this
is,
it
suggests
to
like
inject
your
tracing
contacts,
your
span
and
race
id
into
enter
your
logs
automatically
yeah.
G
G
Is
that
logging
involves,
like
writing
a
logging
exporter
that
would
export
logs
to
hotel
collector
using
odlp
format
whenever
otl
supports
logs
native.
G
D
D
G
D
And
I
thought
that
was
part
of
it
right
like
we
have
the
trace
exporter
to.
We
have
the
otlp
trace
exporter,
but
we
also
have
tracing
api
and
sdk
that
supports
all
that
right,
like
you're
able
to
create
traces
and
stuff.
Is
there
anything
like
that?
Similarly,
for
logs
and
not
really
because
we're
using
the
python
standard
log
library
but
log
correlation
is,
would
be
part
of
that
api?
I
guess
right.
G
Yeah,
I
guess
there
will
be
some
overlap.
I
I
guess
when
we
do
logging
people
might
expect
this
feature,
but.
C
B
I
have
two
quick
questions
on
this.
The
first
one
is,
let's
say
this
is
I
took
a
brief
look,
it
looks
like
it's
using
monkey
patching
I
was
wondering,
does
doesn't
python
logging
expose
some
sort
of
like
handler
or
yeah.
G
G
B
Okay,
cool
cool
awesome
and
then
my
other
question
was
specifically.
I
don't
know
about
other
vendors.
I
imagine
it's
the
same,
but
specifically
for
google.
We
have
a
log
format
that
you
can
use
that
our
cloud
logging
product
will
understand
as
a
cloud
trace
id.
I
know
users
could
set
an
environment
variable
or
set
it
in
their
code,
but
is
there
any
way
that
we
can
make
like
a
nice
wrapper
over
this
like?
Is
it
extensible
in
that
way?.
G
Yeah,
so
so
this
expects
this
accepts
a
couple
of
keyword
arguments.
One
of
them
is
log
format.
So
when
you
call
logging
instrumented
on
instrument,
you
can
pass
log
format
to
that.
What
do
you
think
that
would
work
for
you.
D
All
right,
nice,
any
other
comments
on
this
pr.
D
Cool
aaron
is
right.
If
I
assign
you
this
to
you.
G
Sup
dude
yeah
and
this
one
is
quite
simple.
So
zipkin
exporter
supports
a
max
tag,
value
length
argument
and
it
basically
truncates
the
tag
values
to
to
that
specified
line.
So
this
adds
the
same
feature
to
jaeger
exporter,
but
by
default
it
it
doesn't
do
anything
to
keep
back
cool
yeah.
D
Oh
nice,
awesome
cool.
Oh,
I
think
this
is.
I
think
this
is
diego's.
We
already
talked
about
this
yeah
okay,
moving
on
oh
right,
so
we
were
going
through
some
old
issues
that
were
marked
rc2
and
we
were
thinking
like
whether
or
not
we
still
needed
to
be
rc2
or
we
can
just
easily
kind
of
just
close
them.
It
looks
like
we
assigned
this
to
air
and
aaron.
Was
there
any
development
on
this
or
progress.
B
No,
I
haven't
had
a
chance
to
work
on
it.
I
think
you
remember
what
it
was,
but
basically
there
was
this
car
and
we
kind
of
just
wanted
to
get
it
merged
because
it
was
open
for
a
really
long
time.
Yeah.
It
looks
like.
B
Us
were
assignees
jesus
yeah,
so
I
think
we
went
ahead
and
merged
it.
It's
just
gonna
need
it
could
just
use
some
cleaning
up,
because
we
were
having
trouble
getting
the
author
to
make
the
changes
we
wanted.
I
guess
so
right
that
said
it's
for
an
instrumentation
for
the
otp
exporter.
Oh
it's
sorry!
It's
for
otlp,
not
a
specific
again,
so
maybe
this
one
does
need
to
be
rc2.
Yeah,
okay,.
B
Yeah
I
can,
I
can
make
sure
I
work
on
this
one
this
this
one
this
week.
Sorry
about
that
nice.
D
E
I
did
yeah.
I
just
noticed
that
there
was
a
an
update
from
chicanth
and
I
guess
oh
wait
notice
it
as
well.
I
just
wanted
to
make
sure
that
we
didn't
lose
the
conversation,
so
we
can
get
this
addressed.
D
I
think
there
was
a
issue
that
carlos
brought
up
related
to
this.
D
So
what
was
that
so
sorry
aaron?
Could
you
say
it
again.
D
Oh,
there
was
just
development
from
chicanth,
but
we
didn't
see
it.
I
think,
and
then
oa
just
commented
an
hour
ago
or
so
what
can
can
oh
wait?
Can
you
just
summarize
what
was
concluded
here.
G
Yeah,
so
the
original
issue
was
that
docs
recommended
using
instantiating
a
resource
by
calling
the
class
column,
but
that
didn't
fill
the
resources
that
didn't
detect
the
resource
from
the
environment
variables
and
that
only
works.
When
you
call
resource
or
create.
C
G
The
original
ask
was
to
clarify
this
in
docs
and
recommend
what
is
what
and
then
there
was
confusion
between.
Why
do
we
need
the
constructor
if
we
have
the
static
method
to
create
an
instance
and
second
just
clarified
that
it
might
be
because
the
constructor
is
used
to
create
an
empty
resource
yeah
and
then
the
factory
method
used
to
detect
the
resource.
So
in
that
case,
maybe
I
was
thinking
the
fact
method
would
be
called
detect
instead
of
create
that
way,
the
api
would
kind
of
be
self-documenting
and
very
obvious.
D
What
so,
what
does
the,
what
does
detect
actually
do
so
just.
B
B
So
it
also,
it
also
has
to
do.
There's
like
I
think
this
was
recently
changed
in
the
spec,
but
there's
like
a
service
dot,
something
that's
supposed
to
always
be
there,
and
if
it's
not
passed
we're
supposed
to
generate
a
unique
id,
do
you
guys
know
what
I'm
talking
about.
B
If
the
service
has
no
inherent
unique
id
that
can
be
used
as
a
value,
this
tribute
is
recommended
to
generate
a
random
version.
One
uid
sorry
version
one
version:
four
euid.
This
is
recommended.
B
I
I
don't
know
if
we
have
this
but
yeah
there's
just
like
a
that
and
it's
supposed
to
have
service
dot
name,
that's
right!
Yeah!
If
you
look
at
the
number
three
footnote.
B
This
is
not
required,
but
then
at
the
bottom
it
says
if
the
service
has
no
inherent
unique
id
is
recommended
to
generate
one.
D
Oh
okay,
is
this
related.
B
B
Yeah,
so
the
detect
would
also
have
to
get
well.
Sorry
I
may
have
been.
I
may
have
misunderstood
because
it
doesn't
say
must,
but
I
thought
that
the
detect
would
have
to
add
that
by
generating
a
uid,
if
it's
missing.
D
D
But
if
this
is
a
must,
then
we
should
be
doing
it.
I
haven't
seen
carlos
at
an
issue
for
this,
so
I
don't
know
if
this
is
like,
because
it's
not
required.
I.
B
D
B
Sorry
yeah,
I
guess
I
didn't
mean
to
derail
if
that
I
thought
it
wasn't
for
the
stores,
but
anyway,
back
back
to
this,
with,
like
the
create
and
detect,
if
we
do
have
ones
that
are
required
like
should
we
not
have
create,
do
the
same
thing
as
detect
like,
should
we
in
the
way
to
create
an
empty
resource
only
for
like
internal,
open,
telemetry,
stuff.
D
Well,
looking
at
what's
it
called
don't,
we
already
have
a
way
to
create
an
empty
resource
like
judging
by
carlos's
issue
right
here.
Okay,.
C
D
D
H
D
D
Are
we
gonna
be
doing
the
rename?
I
don't
think
it.
I
don't
know
I'm
kind
of
indifferent
about
the
detect
to
create
thing.
G
D
Yeah
cool
all
right,
nice
continuing
on.
E
You
and
I
did,
but
I
added
a
note
at
the
bottom
that
just
says,
like
our
documents,
are
just
missing:
docs
right
yeah.
So
I
think
that's.
D
D
But
kind
of
lengthy
issue
so.
E
I
don't
remember,
I
think
I
think
our
what
we
said
is
that
well,
at
least
what
I
said
was
that
the
I
mean
we
if
we
have
these
environment
variables-
and
we
have
this
page
for
it,
we
they're
just
not
showing
up
like
it's
a
it's
a
fairly
easy
fix.
Hopefully
we
can
just
get
get
it
fixed
before
rc2.
So
that
way,
people
know
where
to
find
it.
D
D
Okay,
cool
awesome
this
was
assigned
to
diego
before
so
I
wanted
to
kind
of
just
ask
him.
I
can
pick
him
and
ask
him
if
he's
still
gonna
work
on
it.
Okay,
cool
yeah
sounds
good
nice.
I
think
that
was
all
of
our
issues
that
we
had.
We
have
five
minutes
left.
Does
anybody
else
have
any
other
pressing
matters
that
we
want
to
talk
about.
D
Okay,
well,
besides
from
that,
please
take
a
look
at
all
of
these
rc2
issues
that
were
created.
D
We're
really
going
to
be
prioritizing
this
so
yeah,
I
think
alex,
and
I
were
also
discussing
that,
like
we,
I
think
other
languages
are
having
similar
issues
where,
like
we
just
don't
have
enough
bodies,
you
know,
are
working
on
our
respective
languages,
so
we
really
have
to
either
ask
for
more
resources
or
really
like
prioritize
prioritize
things,
which
is
why,
like
I'm
sure,
you
guys
noticed
these
days,
like
our
pull,
requests
have
been
pretty
low,
which
is
which
is
good,
but
I
don't
know
if
it's,
because
we
don't
have
a
lot
of
contributors
or
because
we're
just
good
at
reviewing.
D
But
we
have
a
lot
of
issues
open
and
the
reality
is
we
just
can't
get
to
them
in
time,
especially
if
they
don't
relate
to
getting
our
ga
out.
I
understand
that,
like
a
lot
of
people
prioritize
other
things,
but
you
know
as
like
maintainers
as
a
community.
We
really
need
to
push
for
progress
in
the
release.
Sense,
so
hope
you
guys
understand.
B
Yeah
that
makes
sense
thanks
for
organizing
everything
too
and
putting
it
on
that
board.
I
think
if
we
know
what
to
like
work
on
first,
it's
really
helpful
yeah.
D
Definitely
take
a
look
at
the
rc2
projects:
oh
notice,
there's
no
rc1.
We
already
finished
that.
So
it's
pretty
awesome
cool
yeah.
Our
goal
is
to
try
to
get
these
assigned
and
worked
on.
I
think
I
think
the
timeline
oh
forgot
to
mention
this.
The
timeline
for
finishing
this
in
terms
of
the
process,
pr
that
bogdan
put
out
is
two
weeks
alex
correct
me
if
I'm
wrong,
but
I
think
they're
trying
to
kind
of
put
like
like
light
out
of
fire
under
people.