►
From YouTube: 2021-06-03 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
Oh,
so,
thank
you.
In
fact,
I
have
been.
C
E
A
Well,
I'll
guess
give
a
few
minutes:
hey,
there's
alex
for
later
on
and
full.
A
F
G
G
Yeah,
like
I
was,
I
took
a
road
trip
with
my
friends
and
stayed
overnight
at
my
pm's
cabin.
E
Let
me
bring
up
the
screen
here
and
see
how
we
do
all
right.
Everybody
see
my
browser
window.
E
Oh
yeah,
actually
I
realized
this.
E
That's
true
that
that
one,
that
one
was
surprisingly
uncontroversial,
yeah
yeah
all
right,
yeah,
please
add
yourself:
how
do
you
salty
agenda
just
want
to
call
out?
There
was
a
new
release
that
happened
on
tuesday,
thanks
sladen
for
going
through
that
just
wanted
to
talk
about.
If
there
was
any
issues
with
the
with
the
release,
I
know
I
noticed
it
was
a
fix
for
a
bug
that
oak
opened.
E
Maybe
I
think
that's
the
pr
that's
at
the
at
the
bottom
there,
but
yeah
anything.
You
want
to
call
up
for
the
release
laden
anything.
G
No
everything
was
it's
pretty
great,
like
thanks
for
making
that
the
the
the
release
script
much
more
robust,
especially
when
it
comes
to
the
multiple
versioning
and
then
like
now
we
can
specify
version
files
in
the
config
file
now,
so
it's
pretty
great,
and
this
will
definitely
be
extendable
in
the
future.
If
we
ever
decide
to
like,
perhaps
make
it
so
that
all
the
versions
aren't
updated
in
lockstep,
so
yeah
that's.
This
is
great.
It
was
really
good.
G
Just
had
a
bit
of
issues
because
I
run
it
on
windows
and
windows
had
issues
with
like
the
slash
characters,
but
other
than
that
it
was.
It
was
pretty
good,
makes
me
actually
look
forward
to
releasing
now.
E
G
E
Yeah,
that's
that's,
definitely
a
different,
a
different
message
than
from
previous
releases,
where
it
was
like
two
hours
at
minimum
to
do
a
release
so.
D
Yes,
I
think
this
brings
up
another
issue
so
with
release
scripts
breaking
on
windows.
It's
I
think,
not
that
big
of
a
deal
it's
still
annoying,
but
but
but
this,
but
there
could
be
other
issues
like
there
could
be
bugs
and
actual
instrumentations
of
sdk
on
windows
and
we
never
test
it.
So
maybe
we
should
add
a
ci
job
to
just
run
one.
Maybe
the
latest
version
of
python
on
windows.
E
Yeah
there's
a
there's,
an
issue
opened
for
this:
let's
see
if
I
can
find
it.
D
G
E
Cool
yeah!
No!
That's
that's
awesome
thanks
for
taking
that
one
on.
Hopefully
it
should
be
pretty
straightforward.
I
think
github
actions
makes
this
pretty
pretty
easy,
but
cool.
So
I
guess
the
next
release
will
be
in
the
first
week
of
july
unless
there's
any
hot
fixes,
which
we
should
probably
talk
about,
but.
E
Actually,
yeah,
let
let's
just
quickly
jump
to
this,
because
I
I
wanted
to
have
a
longer
discussion
about
the
metrics
and
the
logging
prototypes
just
to
get
everybody
on
the
same
page
as
to
what
what
is
happening
and
the
prs
themselves.
But
let's
just
take
a
quick
look
through
this
pr
here,
oh
wait:
do
you
want
to
talk
about
it.
D
Yeah
this
was
a.
This
was
a
bug
that
I
didn't
detect.
Well,
while
I
was
developing
the
feature,
it's
got
something
to
do
with
package
resources
and
how
that
library,
parses
versions
and
returns
the
versions
I
suspect
locally,
when
you're
doing
like
python
setup.
by
develop
it
sort
of
treats
that
probably
differently
and
that's
why
I
didn't
detect
it.
But
yesterday
I
was
trying
it,
and
so,
if
you
install,
let's
say
flask
instrumentation
without
flask
being
installed.
D
Instead
of
the
script,
raising
distribution
not
found
error,
it's
somewhere,
it
crashes
with
some
like
it.
It
gets
an
unexpected
format
of
the
requirement
and
that
comes
down
to
the
marker
field
on
dependencies.
A
marker
is
something
like
it.
So
if
you
have
a
dependency,
let's
say:
floss
open
admitted
flask
instrumentation,
and
it's
marked
as
as
an
extra
requirement.
D
So
what
I'm
doing
here
is
I'm
just
setting
marker
to
none,
because
we
really
don't
need
it
and
and
then
this
works,
so
it's
kind
of
a
little
hacky,
but
but
I
think
it
should
be
stable,
because
this
the
function,
it's
calling
is
not
using
marker
in
any
way.
So,
okay.
So
so
I
don't
think
this
should
cause
any
issues.
G
Also
speaking
of
explanations,
yo
yo,
oh
wait.
I
love
what
you
did
with
the
video
with
your
other
pr.
That
was
really.
C
D
E
Yeah
yeah
it
actually
yeah.
I
I
have
to
be
honest
when
I
first
saw
the
feature
that
github
supported
videos-
I
I
hated
it,
but
now
that
I've
actually
seen
it
in
in
action,
it's
pretty
it's
pretty
awesome,
but
just
coming
back
to
this
video
for
a
quick
second,
do
you
think?
Do
you
think
this
is
something
that
we
should
like
release,
a
hot
fix
for
for
the
instrumentation.
D
So
yeah
this
is
a
big
enough
problem,
but
the
upside
is
that
we
still,
we
didn't
update
any
documentation
to
recommend
installing
all
the
instrumentations
and,
if
you
don't
use
it
and
bootstrap
still
works
like
it
works
like
it
used
to.
So
it
detects
what
packages
you
have
installed
and
then
installs
instrumentations
for
those.
So
all
people
who
run
into
this
are
only
people
who
will
like
manually
go
out
of
their
way
to
install
an
instrumentation
for
a
library
that
they're
not
using.
So
it's
pretty
unlikely
that
someone
will
run
into
this
but
yeah.
D
E
I
actually
have
a
question
about
the
speaking
of
patrick's:
do
we
have
the
process
documented
somewhere
laden
around
how
we,
how
we
did
the
hot
fix
for
the
other
thing
we
released?
Was
it
last
week
or.
G
Yeah
good
question:
we
don't
have
it
documented
anywhere.
I
can
add
something
to
the
contributing.
I
guess
I
think
yeah
yeah.
Oh
that's
something
that
contribute
this
is
the
tornado
fix
right,
you're,
referring
to.
G
E
E
D
E
Okay,
so
so
yeah,
so
the
there
are
two
issues
I
was
chatting
with
laden
earlier
and
yeah.
I
was
just
hoping
to
get
so
you
know
we
have
these
prototypes
for
metrics
and
we
have
the
prototype
for
logging
support.
E
I
just
wanted
to
maybe
get
diego
and
sherkanth
if
you're,
if
you
guys,
are
okay
with
it,
not
to
put
you
on
the
spot,
so
you
know
you
can
feel
free
to
to
decline,
but
also
maybe
you
could
walk
us
through
what
what
is
being
implemented
and
yeah
just
kind
of
giving
everybody
a
bit
more
context
on
it,
so
that
we
can
kind
of
move
forward
with
the
spheres,
because
they're
pretty
they're
they're
more
complex
than
the
regular
pr.
E
So
I
go
first
yeah
whatever.
If
you,
if
you're
ready
for
it
yeah.
A
A
Okay,
yeah
metrics.
Well,
I
guess
everybody
remembers
metrics
right
something
that
we
always
wanted
to
have
so
yeah
big
pr
here
pretty
much.
I
have
been
trying
to
follow
what
specified
the
specification,
but.
E
Sorry,
do
you
wanna,
do
you
wanna
share
because
I
figured
you
like
you
might
wanna
if
you
wanna
walk
us
through
pr,
I
don't
know
if
you
wanna
walk
through
the
pr
or
not,
but
if
it's
all.
A
Right,
yeah
I'll
I'll
share.
Let
me
see
how
can
I
share
here
where's,
my
pr
okay.
So
let's
move
this
here
and
screen.
E
Can
you
see
something
you
can?
We
can
see
open,
telemetry
python,
okay,
because
he
might
be
alright.
Here,
that's
good!
Oh
okay!
We
could
and
now
it
disappeared.
Now.
I.
A
A
Okay,
this
pr
fill
it
has
a
different
implementation
that
we
had
before,
because
this
pr
pretty
much
defines
a
hierarchy
of
classes
that
describe
the
different
instruments.
So
we
use
multiple
inheritance
to.
A
Make
the
instruments
be
defined
by
their
parents?
So,
for
example,
we
have
a
base
class
that
is
instrument
everybody
will
inherit
from
it,
but
we
also
have
synchronous
asynchronous,
adding
grouping
monotonic
and
non-monotonic.
So
every
instrument
he
inherits
from
these
classes
so
that
we
can
tell
if
an
means,
if
a
certain
instrument
is
monotonic
by
checking.
A
A
Now
this
concepts
adding
and
grouping,
especially,
I
think
they
are
not
that
well
defined
in
the
spec
now
I
think
they
used
to
be
because
I
think
this
is
an
an
an
idea
that
josh
mcdonald
who's,
a
big
metrics
expert
explained
to
me
before,
and
I
think
it
was
pretty
useful.
So
I
kept
it,
I
don't
I'm
not
sure
if
the
spec
is
going
to
move
in
this
way,
but
I
feel
like
it
is
useful.
A
Let
me
show
you
something
that
makes
it
a
bit
easier
to
understand
why
I
chose
this.
A
A
Yep,
okay,
so
yeah.
So
here
are
the
old
instruments
that
we
had,
and
I
think
this
table
makes
it
easy
to
understand
how
they
are
defined
right,
how
they
are
grouped
in
by
their
different
characteristics.
A
So
this
is
this
table
is
old,
because
I
think
we
don't
have
now
the
this
concept
of
being
bound
or
unbound,
but
I
pretty
much
used
the
same
approach
to
this
implementation,
all
right.
So
that's
the
main
thing
that
you'll
notice,
when
you
get
to
cspr
a
bunch
of
classes
and
as
you
can
see,
every
instrument
inherits
from
their
parents
so
that
they
can
be
defined
with
them
nice.
So
far,
I
have
this
implemented
the.
I
think
this
will
be
the
main
difference.
A
The
rest,
I
think,
it'll
be
kind
of
similar
to
what
we
had.
We
will
have
a
meter
provider
a
meter,
all
that
kind
of
stuff
right,
so
yeah
now
the
testing
for
this.
Well,
there
are
some
regular
expressions
define
how
the
names
have
to
be.
Where
distance
is
here
nothing
out
of
the
ordinary
here
we
are
testing
that,
for
example,
a
counter
is
synchronous
and
monotonic.
A
A
Okay,
that's
it
for
metrics.
I
still
have
something
else
that
has
nothing
to
do
with
metrics.
That
is
inside
of
this
pr
that
I'll
probably
move
it
out
of
this
pr.
But
I
don't
want
to
take
time
explaining
what's
the
very
controversial
stuff
that
follows,
so
I'm
not
scrolling
down
anymore.
Anybody
has
any
questions.
G
Diego
when,
when
this
is,
is
this
going
to
be
in
draft
for
a
bit
or
or
because
this
is
just
going
to?
This
is
just
leading
like
the
sdk
effort
right
like
a
prototype,
so
are
we
just
going
to
keep
it
in
draft
for
a
while,
or
what
do
you
think.
A
A
That
depends,
I'm
not
sure,
but
right
now,
the
the
specification
document
in
the
specification
repo
that
specifies
metrics
has
a
big
experimental
tag
there.
So
I'm
not
sure
how
the
community
wants
every
sick
to
handle
this.
If
we
are
expected
to
keep
it
draft,
I
think
it
will
maybe
not
make
much
sense
to
have
this
marriage
into
maine.
G
Right
so
what
the
stakes
have
decided
or
the
the
metric
sig
has
decided
is
that
the
api
is
marked
as
experimental
and
they're
trying
to
release
it
soon,
probably
it
was
planned
at
the
end
of
may,
but
I
think
it
got
pushed
back
so
it'd
be
it
would
make
sense
to
when
that
happens,
or
like
even
now
to
merge
the
api
into
the
our
experimental
branch.
G
However,
for
the
sdk
effort,
the
specs
haven't
started
yet,
and
they
want
different
languages
to
like
prototype
out
the
scenarios
like
similar
to
what
you're
doing
right
now
to
see
if
it's,
if
it
works,
and
what
what
the
sdk
surface
is
supposed
to
cover.
So
that
could
be
what
drives
you
know
the
sdk
specs
to
be
going
so
for
example.net.
Right
now
has
several
prs
that
implement
what
they
expect
sdk
to
be
so.
G
A
Yeah,
okay,
actually
I
have
been
keeping
this
as
draft,
because
I
mean
tests
are
obviously
not
passing
here
and
I
have
still
much
code
to
yeah
right.
So
I
didn't
want
to
waste
anybody
anybody's
time
in
reviewing
this
pr,
when
I
still
have
a
bunch
of
code
to
add.
So
let's
do
this.
I
can
keep
working
on
this
draft
and
I'll
make
it
ready
to
review
when
it
at
least
passes
this
test,
so
that
you
have
something
better
to
review,
because
I
don't
want
to
have
you
guys
reviewing
something.
That's
not
is.
G
Yeah
something
like
that
that
way
like
we
could
actually
merge
things
in
quickly
without
having
to
like,
because,
like
my
issue
with
drafts,
is
that,
like
I
look
at
them
and
then
I
forget
about
them,
and
then
I
don't
know
what
they
were
about
and
they
get
outdated
quickly.
So.
A
A
But
I
can
do
that
yeah.
Definitely
so
yeah.
E
I
wonder
I
wonder
if,
like
I
wonder,
if
the
real
value
that
we're
trying
to
get
to
here
is
not
not
even
so
much
getting
the
api
merge
as
much
as
just
getting
a
sense
for
whether
like
getting
that
feedback
about
what
we
expect
the
sdk
to
work
or
whether
the
api
is
good
enough.
Right
like
I
like.
I
wonder
if
that
really
should
be
the
focus
of
this
right
of
this
prototype,
as
opposed
to
trying
to
get
something
something
actually
mergeable.
G
I
think
I
think
that
is
true
like,
but
if
that
has
to
be
the
clear
message
and
goal,
it
seems
like
I
thought
originally
just
from
diego's
explanation:
we're
actually
trying
to
implement
like
an
sdk
or
something.
G
If
that's
the
case,
then
like
that's,
not
really
what
we
were
trying
to
do,
but
if
it's
the
former
like
what
you
said
alex
then
like,
how
do
we
execute
that?
Like
do
we
just
leave
it
in
draft
kind
of
thing
like
how
or
how
is
this
going
to
drive
like
the
metrics
compensation,
then.
E
Yeah
I
mean
I,
I
kind
of
imagined
that
it
would
have
to
stay
in
draft,
and
you
know
it.
This
would
be
the
kind
of
thing
where
I
guess
they
go
as
you're
as
you're
working
through
the
scenarios
and
in
that
otap
I
think
it's
like
156,
maybe
146..
Maybe
you
can
kind
of
feed
that
back
into
the
into
the
metric,
sig
or
whatever,
like
maybe
use
this
as
the
pr
that
can
be
at
146.
E
A
E
So
if
you
look
at,
if
you
look
at
that,
otap
there's
there's
a
couple
of
scenarios
that
are
being
discussed
in
there.
E
If
you
look
at
the
actual
the
file
here,
but
essentially
it
says:
hey,
like
here's
two
scenarios
where
we
want
someone
to
be
able
to
use
the
metrics,
sdk
and
metrics
api
to
be
able
to
calculate
to
measure
things
for
different
scenarios,
and
one
of
them
is
like
a
like
a
store,
that's
adding
requests
and
orders
and
a
bunch
of
other
stuff.
Then
the
other
one
is
like
a
machine
machine
host,
metrics
type
scenario.
E
E
E
E
Sorry,
as
long
as
we're
we're
all
understanding
that
this
code
might
just
be
thrown
away
or
whatever
like
like,
it,
won't
emerge
as
is-
and
that's
that's
okay.
This
is
strictly
to
prove
that
these
scenarios
will
work.
E
A
E
Yep,
okay,
yeah
and
I
think
the
from
my
understanding
like
since
there
is
no
sdk
this.
This
is
an
opportunity
to
drive
what
the
sdk
should
look
like
because,
like
we,
we
take
these
scenarios
and
we
think
you
know
you'll
have
to
think
through
them
and
essentially
say
like
here's.
What
here's
the
best
way
that
I
would
imagine
this
would
work
and
and
propose
that
to
the
to
the
rest
of
the
sig
right.
A
E
G
E
I
would
almost
imagine
it's
the
other
way
around
a
little
bit
like.
I
would
start
with
these
two
examples
and
then
try
and
imagine
like
how.
How
would
you
go
about
like
configuring,
the
sdk
exporter?
How
would
you
go
about
actually
adding
these
metrics?
You
know
considering.
There's
no
sdk
right
now
like
how
would
you
then
build
the
sdk
for
it.
E
A
Sure,
okay,
good
plan
I'll
do
that.
G
I
don't
know
the
process
in
which
how
they're
trying
to
do
it,
it's
still
relatively
early
like
not
even
net,
has
a
prototype.
Well,
they
do,
but
I
don't
know
if
they're
talking
about
it,
so
I
think
the
best
way
would
be
to
attend
the
metric
stick
and
then
bring
up
that
like
hey
pipe
python
has
like
a
working
example
and
stuff
like
that.
So
yeah.
A
G
G
But
it's
like
like,
if,
if
you
don't
have
the
sdk
prototype
ready,
like
just
put
it
on
the
radar
that
you're
working
on
it
and
they'll,
be
very
very
grateful
that
we're
where
you're
involved
in
this
so
yeah.
E
I
think
I
think
just
getting
a
sense
for
where,
like
where
that
work
is
happening.
Right
now
would
be
really
good,
because
it's
it's
not
super
clear
to
me.
A
And
let
alex
put
somebody
else
on
the
spot
yeah.
Thank
you.
E
Thanks
thanks
diego
for
for
jumping
in
did
you
want
to?
Did
you
want
to
go
over
the
the
the
logging
prototype
that
you
have.
F
B
I'll
have
to
quit
and
reopen
to
give
the
permission.
Screenshot
information
I'll,
join
again.
E
We
we
lost,
there
was
one
new
person
that
we
didn't
get
a
chance.
Yeah.
A
E
We
need
to
up
our
open,
telemetry
python
swag
game
to
get
more
people
involved.
G
B
B
Yes,
so
yeah
before
before
I
go
into
this
code
right,
so
the
logs
specification
this
this
new
way
that
will
be
defined.
The
goal
is
to
make
it
work
with
the
existing
logging
libraries.
I
know,
for
example,
java
has
log4j
and
python
standard
library,
standalone
library
in
in
google.
B
B
We
can
do
that
with
so
there's
a
way
which
we
can
do
that
in
standalone
library.
That
handlers
so
except
accept
that
part.
Almost
all
of
this
logging,
you
know
sdk
mostly
it
mimics
the
what
we
have
in
the
tracing,
so
the
log
processor,
our
log
exporter
pipeline,
all
of
it
that
is
very
similar
to
what
we
have
in
the
tracing.
B
So
I'll
go
through
this
code,
one.
So
this
this
is
analogous
to
stand
like
we
have
what
we
have.
B
So
so
one
thing
one
thing
I
also
did
a
research
is
so
like
mostly
people
use
a
standard
library
login,
but
there
are
also
some
third-party
libraries
one
one
example
is
a
struct
log,
so
it
also
has
a
mechanism
where
you
can
have
it
like
plug
in
a
hook
every
even
every
log
that
is
generated
with
that
library
will
go
through
that.
B
So,
if
we
want
to,
if,
if
say,
if
we
want
to
support
that
third
party
login
library
as
well,
we
can
we
can
do
that
in
this
like
in
this
pr.
We
are
only
targeting
standard
library
login.
So
that's
that's.
The
log
record
part.
B
So
remaining
remaining
like
whole
log
processor,
multi-log
processor,
like
what
we
have
in
the
spa
like
the
tracing,
so
stand:
processor,
multi,
multi-span,
processor,
synchronous,
concurrent.
This
is
like
pretty
much
similar
to
that
on
this
one
as
well.
B
So
here
we
have
a
lot
like
similar
to
what
we
have
in
dresser
provider.
We
have
login
data
provider.
They
can
get
the
login
using
like
this
definition.
B
Using
yeah
so
yeah
this
one,
so
this
this
is
the
like
the
important
part.
So
we
we
are
subclassing
this
handler
from
the
standard
library
and
every
time
like
every
time
we
there's
a
log
event,
so
it
gets
processed
through
this
log
emitter.
B
So
every
time
and
there's
a
log
emitted
from
the
standard
library,
we
we
transform
it
to
the
otlp
format.
No,
we
have
we
have
and
then
and
then
and
then
inject
that
to
the
pipeline.
So
it
goes
through
the
log
processor
and
then
export
or
it
gets
you
know
either
either
either
to
a
file
or
over
the
network.
B
It
can
emit
the
data
and
that
that
is
this
and
so
yeah.
Here
we
have
again
similar
to
spanx
span
exporter.
We
have
log
exporter
like
simple
spam
process,
like
log
processor,
batch
batch
log
processor,
pretty
much
same
what
we
have
in
the
tracing
sdk.
B
Error,
critical,
the
naming
naming
in
standard
library
and
the
naming
in
otlp
is
different,
but
we
so
there
is
a
specification
for
like
how
how
to
map
it
from
the
source
to
the
odlp
format
so
and
like
the
step
step
from
the
one
level
to
and
the
level
is
also
different
here,
it's
10
to
20,
but
it's
in
what
dlc
is
like
10
to
15
and
then
15
to
20,
so
there's
a
specification
that
is
defined
to
like
how
do
we?
How
do
we
translate
it
from
the
source
to
the
the
odlp
format?
B
Yeah
and
this
these
are
these
are
tests,
so
so
the
major
major
parts
are
one
of
the
this
severity
concept:
race
correlation
that
we
do
when
we
are
generating
the
log
record
and
and
the
subclass
that
we
are
using.
So
this
log
emitter
that
we
are
subtracting
from
the
standard
handler
class
and
log
record
so
remaining
remaining
classes
and
definitions
are
very
similar
to
what
we
have
interesting
sdk.
B
A
Yeah
I
have
looking
into
this
that
you
a
few
comments
there.
I
think
you
were
asking
about
what
to
do
with
interviewing,
so
I
can
take
this
opportunity
to
explain
myself
a
little
bit
better.
I
think
in
one
of
your
comments.
A
Second,
can
you
go
to
the
comment
where
I
suggest
using
entry
points?
It's
I
think,
yeah.
B
A
One
okay!
Yes,
thank
you,
so
I
think
that
what
is
being
attempted
here
is
to
add
a
new
method
that
will
create
one
of
these
open,
telemetry
log
records
from
other
log
records
from
other
libraries
right.
That's
the
idea
behind
this
method.
A
Yeah:
okay,
yes,
so
this
pretty
much
like.
Can
I
translate
a
record
from
another
library
into
an
open,
telemetry,
lock
record?
What
I
mean
with
using
entry
points
here,
is
that
let's
say
that.
A
Its
own
login
records
right.
So
what
I
mean
is
that
I
don't
feel
like
we
should
have
something
named
from
django
to
open
telemetry.
A
Hardcoded
in
our
sdk,
but
we
should
handle
that
with
enterprise
so
that
a
third-party
library
can
register
or
third-party
module
can
register
itself
so
that
it
incorporates
here
in
just
in
the
same
way
that
we
do
so
for
instamentor,
so
that
we
don't
hardcode
third-party
library
stuff
in
our
sdk.
That's
what
I.
B
B
Like
so
I
I
I'm
not
sure
I
I
looked
at
the
other
popular
libraries
so
like
based
on
their
uses,
so
they
have.
They
have
very
similar
the
logging
library
card
and,
like.
D
Maybe
sorry
to
hijack
diego,
I
I
think
what
digger
strength
says
that,
like
sdk,
cannot
have
knowledge
about
third
party
libraries
only
about
the
standard
library
correct.
So
so,
if
you
want
to
add
support
for
any
third
party
library,
it
would
have
to
be
like
an
instrumentation
or
a
plug-in.
So
if
you're,
using
like
a
third-party
library,
I
don't
know
like
big
one,
you
would
install
a
plug-in
for
that
for
open
telemetry.
Then
that
would
hook
hook
that
libraries
logs
to
the
hotel
logging
sdk.
D
B
Yeah,
I
understood
that
so,
but
but
the
like
specification
like
like
is
like,
like
python
like
in
python,
that,
like
we
are
using
like
there's
a
standard
library
but
like
in
a
bigger
picture
log
specifications.
B
So
they
they
motivate.
You
know
only
like
make
it
work
for
the
third
party
libraries.
It
happens
to
be
that
in
python,
like
we
have
it
in
in
standard
library
that
so
that
even
everyone
can
can
use
it.
But
if
yeah
so,
like
the
like,
the
bigger
motive
of
the
specification,
like
the
log
specification
itself,
you
start
getting
the
third
party
library
so.
D
But
we
can
still
achieve
that
right
with
with
something
like
plugins
or.
B
D
Different
classes
for
different
libraries
right
yeah:
do
you
see
any
issues
with
like
if
someone
installs
that
third
body
use
a
third-party
login
library
and
then
has
to
install
one
additional
package,
do
you
see
any
issues
with
that.
B
No,
no,
no
so
yeah!
I
I
I
I
I
I
think
I
get
the
idea
now
I
don't
see
like
I.
I
don't
see
any
issue,
I'm
not
opposed
to
it,
but
I
wanted
to
understand
it
clearly
like.
B
F
A
I
think
we
I
feel,
like
you,
now
understand
what
I
was
trying
to
say.
This
is
what
I
wanted.
G
G
An
issue
or
discussion
for
this
pr,
it
seems
like
it's
a
pretty
big
kind
of
architectural
edition.
We're
adding
and
it'd
be
good
for,
like
a
consolidated
place
for
us
to.
B
Okay,
so
I'll
create
an
issue
that
there's
no
issue
in
in
this
in
this
repo,
so
I
I
was,
I
was
joining
the
logic
also
I
was
announcing
like
like
how
do
we
want
to
go
about
it?
Yeah
I'll,
create
one
I'll
show
you
in
this
paper.
A
All
right,
yes,
you
can.
Can
you
scroll
down
a
little
bit?
Okay,
there
they're,
excellent,
okay,
so
yeah
we
were
having
this
conversation
regarding
the
quality
con
comparison
operator
here.
A
To
be
honest
with
you,
I
am
not
totally
sure
that
the
approach
that
I
am
suggesting
is
the
right
one.
That's
because
I
have
been
reading
the
recommendation
of
this
equality
operator.
Can
you
please
open
the
first
link
that
I
added
just
right:
yeah,
yeah!
Thank
you,
please.
Okay,
so.
B
A
It
says
there
is
that
a
rich
comparison
method,
which
is
equal
method
right,
may
return
the
singleton
not
implemented
if
it
does
not
implement
the
operation.
For
a
given
pair
of
arguments,
my
convention
falls
into
a
return
for
a
successful
comparison.
So
I'm
not
sure
that
means
that
if,
because
they
only
say
successful
comparison-
and
I
don't
know
if
comparing
two
things-
a
log
record
with
something
else
is
considered
to
be
non-successful
or
successful.
A
So
that's
why
I'm
not
completely
sure,
but
if
we
go
back
to
the
the
comment
that
we
had
there
yeah.
Thank
you.
A
What
I
have
seen
is
that
for
many
basic
python
types
it
is
possible
to
compare
between,
for
example,
an
integer
and
a
string,
a
boolean,
a
non
a
float
and
a
boolean,
and
in
all
those
cases,
what
the
comparison
returns
is
false,
which
can
also
make
sense,
because,
basically,
a
comparison
is
saying
it's
an
apple
equal
to
an
orange.
No,
they
are
not
equal.
That's
that's
false
right,
so
it
also
makes
sense.
That's
that's
why
I
was
suggesting
this,
but
I
think
I'll
I'll
leave
it
up
to
you
or
somebody
else
to
chime.
A
In
because,
according
to
that
documentation,
I'm
still
not
completely
sure
what
we
should
do
here,
so
I
pretty
much
just
wanted
right
now.
The
only
thing
that
I
want
is
to
bring
to
your
attention
that
this
kind
of
comparison
with
the
equal
operator
happens
and
it
is
defined
for
python
basic
types
that
are
not
the
same
thing
so
yeah.
I
wanted
to
give
you
this
context.
B
So
yeah
my
my
my
question
like
similarly,
so
these
these
types
are
like
not
not
what
we
are
dealing
with
right,
so
the
log
like
in
this
case
it's
the
log
record.
So
if
we
were
to
not
like
like,
when
does
it?
B
A
Yes
sure,
I'm
not
sure
again
if
we
should
return
not
implemented
or
false,
but
I
just
I
think,
I'll
leave
it
up
to
you
or
to
somebody
else
to
design
the
final
implementation,
because
I'm
also
not
that
totally
sure
about
this.
B
A
E
A
Just
just
to
let
you
know
extremely
controversial
topic
being
discussed
here,
so
don't
miss
this
opportunity
to
get
involved.
A
What
was
the,
what
was
the
drive.
B
Yeah,
so
I
I
was
attending
this
econ.
They
were
also
so
those
those
six
were
basically
like
they
were,
they
didn't
have
anything
to
talk
about,
so
no
no
there.
No,
but
no!
No
one
was
you
know,
implementing
anything.
D
G
Okay,
awesome
yeah,
I
have,
I
haven't,
really
been
attending
the
log
six.
So
thanks
for
doing
that,
do
you
know
kind
of
like
the
general
like,
if
there's
a
timeline
or
like,
if
it's
just
too
early
right
now,
they're
just
looking
for
volunteers.
B
I
guess
yeah
so
they're,
looking
for
the
warranties,
the
log
data
model,
they're
making
it
as
a
beta.
So
I
don't.
I
don't
really
see
many
major
changes
going
forward
as
well.
Unless
you
know
people
people
start
sharing
their
feedback
on
them,
many
many
they
want
to
introduce
many
things
later
onwards,
but.
G
It's
it's
different
from
the
metrics
prototype
in
which
the
metrics
prototype
is
actually
driving
the
sdk
design,
for
this
is
they're
just
curious
at
to
the
functionality
that
and
how
each
language
would
be
implementing
it
right.
It's
not
like
they're
going
to
have
any
like
hey
you're,
doing
this
the
wrong
way
or
something.
E
Yeah,
okay,
cool
sounds
good,
it's
awesome
and
it
looks
like
based
on
the
this
one
issue
and
the
spec
that
the
the
sig
for
metrics
is
planning
or
sorry
for
logging
is
planning
on
marking
the
logs
as
beta
from
experimental.
So
that's
that's
pretty
exciting.
B
G
No,
I
mean,
like
I've
only
ever
seen,
experimental
too
stable
right,
yeah.
G
E
G
A
Yeah,
if
you
can,
I
have
another
comment:
if
you
scroll
down
with
this.
A
I
just
read
the
last
comment
that
you,
you
posted
and
I
think
it
makes
sense.
B
A
Understand
the
logic
that
you
followed
to
map
the
these
severity
values
so
that
the
highest
ones
get
mapped
to
the
highest
severity
value
in
open
telemetry,
that
that
seems
logical.
What
I
mean
is
that
these
specifications
says
that
if
the
and
I'm
reading
it
actually
here,
if
the
source
format
has
only
a
single
severity
matches
the
meaning
of
the
range,
then
it
is
recommended
yeah
right
there
in
the
scroll
down
scroll
down
scroll
down
in
the
mapping
observation
number
section
yeah,
the
third
paragraph.
I
think
it
is
okay.
A
No
sorry
if
the
source
format
has
only
a
single
severity
that
matches
the
meaning
of
the
range
it's
recommended
since
verity
to
smallest
value
of
the
range
you
are
doing
that
because
you're
mapping
debug
to
the
first
ebook
level,
which
is
fine,
I
mean
I
think
your
implementation
is
according
to
the
spec
for
the
the
values
that
have
a
direct
mapping
between
one
system
and
and
the
other
one.
A
What
I
think
is
happening
here
is
that
your
function
that
makes
this
conversion
is
accepting
values
that
are
not
defined
in
the
in
the
standard
python
login
levels,
the
standard,
python
login
levels
only
have
as
many
values
as
severity
levels
are
right,
so
it's
10,
20,
30
40,
you
know.
So
what
I'm
suggesting
here
is
that
maybe
the
this
function
that's
doing.
The
mapping
should
not
accept
these
intermediate
numeric
values
like
11,
12,
13
and
14.
B
So
yeah,
so
the
the
the
reason
I
wanted,
the
intermediate
values
is
that
you
can.
So
this
is
what
to
get
out
of
the
standard
library,
but
you
can,
however,
add
your
own.
Let's
say
you
know
you
want
to
add
some
level
between
the
debug
and
info
right,
let's
say
some
imaginary
level:
okay
x,
so
you
can
define
that
and
then
and
then
assign
it
to
the
numeric
value
15.
It
still
works.
So
the
severity
like
these
are
what
we
have
in
the
standard
library
but
like
it.
A
Sure,
then
I
I
guess
I
guess
it's
fine,
I
think
yeah.
This
is
probably
much
more
of
an
implementation
detail,
but
it's
not
even
well
defined
in
the
specs,
so
so
yeah.
I
think
considering
that
I
think
the
implementation
right
now.
It's.