►
From YouTube: 2020-05-21 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
B
A
A
B
A
A
A
You
know
what,
let's
just
talk,
we
can
be
recorded.
Who
cares?
I
just
wanted
to
throw
a
couple
of
things
out
before
the
metrics
meeting,
because
I
know
that
it's
gonna
be
full
and
and
there's
a
big
agenda,
we're
still
talking
about
metric
naming
and
there's
a
new
proposal.
A
new,
a
checklist
filed
by
one
of
my
colleagues
today
that
that's
coming
thing
I
want
to
get
back
to
you
about.
Was
that
conversation
about
terminology
from
how.
A
A
A
A
The
reason
I
like
the
word
interval
is
that
if
you
look
up
in
this
sort
of
statistics,
literature
very
often
there's
a
like
a
chapter
or
part
a
chapter
on
like
types
of
measure,
types
of
measurement-
and
this
is
all
very
relevant
like
this
notion
of
distributional.
This
is
instantaneous,
perhaps
or
or
sorry
distributional
versus
active.
B
A
What
that
says
is
0
is
not
special,
but
you
can
always
subtract
these
things
to
get
a
meaningful
value
and
if
you
have
a
ratio
measure,
if
it's
something
where
0
is
meaningful
and
therefore
you
can
take
ratios
that
are
meaningful,
and
so
a
UNIX
timestamp
is
a
perfect
example
of
an
interval
and
it
also
kind
of
maps
on
the
concept
of
an
interval
that
we've
talked
about.
So
maybe
that's
actually
a
conflict
in
terms
of
terminology.
The
fact
we
already
use
the
word
interval,
but.
A
A
good
word
to
leak
at
least
consider
but
I
still
I'm
starting
to
come
to
this
position
or
belief
that
that
we
only
need
to
ever
joined
the
person
is
joining.
We
could
talk
about
go
again
just
to
finish
that
thought.
I
still
think,
but
I
haven't
convinced
myself
in
a
formally
written
down
way
that
we
only
need
two
terms
to
describe
an
aggregation.
I
was
going
to
say
instrument,
but
bogan
wants
me
to
say
to
think
about
the
aggregation
and
it
has
to
do
with
whether
the
interval
is
meaning
the
interval.
A
B
Sorry
tiger
who
just
joined
was
continuing
a
thought,
but
so
yeah,
just
kind
of
like
put
a
bow
on
that
or
just
follow
on
yeah
the
after
you
jumped
off
and
bogged
in
scolded
me
totally
called
for,
but
the
I
think
we
did.
We
talked
a
little
bit
more
and
he
was
I
think
we
both
came
to
the
same
conclusion
that
Delta
and.
B
A
A
How
can
you
disjoint,
if
you
would
like
to
speak
or
say
anything
otherwise,
but
you
haven't
really
there's
like
no
open,
PRS
or
there's
one
of
yeah,
the
only
open
the
release
and
we're
down
to
35
issues.
A
A
B
A
Don't
know
like
we're
being
a
little
bit
unilateral
I
I
want
to
vote
in
his
opinion
on
this
yeah,
but
in
the
spec
document
that
just
merged
it
says
they
all
use,
knew
that
and
I
put
that
there
explicitly
I
support
this.
What
everyone
approved
it
so
anyway,
so
I
could
say
the
spec
now
says
new
and
and
that's
enough
justification
and
I
could
do
it
in
ten
minutes
and
you
could
prove
it.
We
can
purge
it
and
get
released.
Yeah.
A
B
A
Right
exactly
cool!
Well,
why
don't?
We
call
this
meeting
short
and
came
back
to
work?
I
will
update
this
issue
number
seven.
Oh
eight
pad
model,
Prometheus
gauge
and
hotel
after
sometime
in
the
next
day
or
so
because
Lisa
released,
but
I
did
I.
Maybe
I
could
give
you
a
free
you,
a
half
a
conversation
since
I
expect
this
metric
sig
to
like
to
go
towards
the
standardizing
metrics
topic
and
stay
there.
There's
this
other
one
that
I
think
of
is
pretty
minor.
A
He
has
suggested
the
absolute
value
recorder
instrument
like
right
now.
We
have
six
instruments
you.
If
for
symmetry,
you
probably
won't
do
both
the
synchronous
and
asynchronous
don't
put
us
up
to
date,
I
kind
of
like
having
six.
It
seems
like
a
lot
already
and
my
philosophically
I
think
I,
don't
like
the
proposal,
because
it's
not
that
great
deal
for
a
reason
to
for
the
programmer
to
have
to
say
these
are
non-negative.
A
If
you're
gonna
define
a
view
or
change,
some
configuration
I
think
it's
okay
for
the
operator
to
assert
these
would
have
to
be
non-negative.
We
know
they're
not
negative,
therefore
we're
going
to
configure
view
and
the
rate
will
be
well
defined.
I
mean
like
you
could
just
have
that
have
a
configure
of
you.
A
The
one
he's
been
talking
about
and
have
it
be
an
error
that
tells
the
user
hey
you
computer
to
some
rate
view
and
got
a
negative
number
make
an
error
like
I'd
rather
have
these
I'd
rather
have
vendors,
invest
in
making
errors
like
observable,
rather
than
like?
Changing
the
API
to
to
ladies
errors.
So
that's
my
philosophical
objection.
I
have
a
practical
reason
to
support
the
proposal
and
it's
that
related
to
the
issue
that
I
just
mentioned
708
or
whatever
708
yeah,
and
that's
that
the
question
keeps
coming
up
and
doesn't
really
go
away.
A
How
can
I
have
a
gauge?
That's
what
I
want.
You
know
like
there's
lots
of
philosophical
arguments
why
gauges
are
kind
of
weird
where
I?
Maybe
maybe
it's
not
it's
like
not,
there's,
there's
something
funny
about
gage.
It's
not
I,
don't
even
know
what
I'm
trying
to
say:
it's
not
a
well-defined
aggregation
in
some
sense.
A
A
Last
value,
and-
and
so
they
call
get
a
gage
I-
would
prefer
to
call
it
a
last
value,
but
nevertheless
there's
no
instrument
that
we're
providing
that
gives
you
an
out
of
the
box
aggregation.
That's
going
to
be
exactly
that,
and
when
we
talk
about
conversion
into
Prometheus
or
dusty,
we
end
up
mapping
everyone.
Those
instruments
to
something:
it's
not
a
gauge.
A
So
my
practical
reason
to
support
bogus
raposo
is
that
if
we
had
two
different
instruments
that
were
for
reporting
distributional
type
data,
then
you
could
make
a
pretty
good
argument
that
one
of
them
should
default
to
gauge
and
one
of
them
should
default
to
histogram.
So
the
idea
would
be
the
absolute
value
recorder
becomes
a
histogram
and
the
value
recorder
becomes
a
gauge
I
think
that
was
my
thought
and
I.
A
B
A
A
So
if
somebody
has
something
that
is
synchronous
though,
and
they
want
to
use
a
gauge
like
that's
the
place
where
there's
no
there's
no
out
of-
and
so
you
can
tell
them,
we
think
you
should
set
a
callback
and
fire
gage,
but
that
that's
probably
not
that
what
they
want
and
so
I'm
wary
of
of
trying
to
tell
Prometheus
users.
You
don't
want
gauges
and
you
can't
have
them
like
I.
A
B
A
B
B
Off
on
that,
you
want
to
try
and
get
that
you
name
anything
in
or
I
mean
we
still
have
another
release
following
next
week.
That's
been
a
pretty
major,
because
I
was
trying
to
put
all
the
milestone
things.
There's
gonna
be
some
like
changes
for
the
SDK,
better,
probably
gonna
be
breaking.
So
we
could
just
wait
till
then.
But
it's
up
to
you
wait
an
hour.