►
From YouTube: 2021-04-09 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
That
is,
basically,
you
multiply
all
the
combinations
and
you
define
each
individual
api
to
cover
a
very
specific
combination,
which
means
you
might
have
eight
or
16
apis,
and
that
seems
to
be
the
case
of
where
micro
return
permissions
was
well
established.
Things
are
coming
from
now
we
look
at
the
prometheus
api.
It
seems
they
were
never
trying
to
solve
that
from
the
mathematical
perspective,
like
the
api
is
orthogonal.
A
If
you
look
at
counter
and
gauge
impermissions,
they
don't
try
to
make
it
like
super
beautiful,
like
this.
One
is
like
totally
orthogonal
to
the
other.
It
looks
they
were
coming
from.
The
usage
scenario
like
the
counter
is
the
number
one
widely
used
like
80
percent
of
developers
could
use
that,
and
there
are
some
minor
cases
might
be
like
0.01
of
the
developer
will
use
a
particular
combination
and
they
don't
even
have
a
dedicated
api
for
that.
A
You
just
report
use
the
the
least
common
api
to
report
the
number,
and
then
you
define
some
aggregation
in
the
backend.
So
this
is
where
I
try
to
push
folks
to
think
from
the
scenario
and
only
cover
the
most
common
cases
in
the
api
and
leave
the
room
for
people
to
use
the
the
flexible
like
to
use
whenever,
like
the
I
see,
final
recorder
but
right.
This
is
the
protective
and
I
think,
like
folks,
understand
that.
But
they
were
from
like
when
we
talk
about
the
balance.
B
So
so
what
you're
trying
to
push
people
for
is
having
like,
basically
a
low-level
generic
api
like
a
single
api
that
could
do
everything
and
then
for
the
super
common
cases.
You
have
an
additional
api
that
gives
you
the
semantics
for
that
case,
yeah
yeah.
That
makes
sense
to
me
I
mean
especially
if
you're
saying
that,
could
then
are
you
saying,
like
those
special
cases,
you're
getting
the
semantics,
but
it
actually
is
equivalent
to
something
you
could
do
with
the
more
generic
api.
This.
A
A
You
you
can
you
can,
but
it's
harder,
so
it's
just
for
convenience.
Ultimately,
it's
just
reporters.
B
Convenience
yeah
that
I
I
personally
like
that
approach
of
saying
you
have.
I
mean
it's
something:
I've
tried
to
take
with
open
telemetry,
but
I
agree.
It
runs
into
problem
with
defining
the
convenience,
but
the
idea
that
you
want
to
have
like
a
model,
that's
coherent,
and
you
want
to
solve
that
in
some
place,
but
making
that
model
completely
coherent
often
makes
it
a
little
unwieldy,
and
then
you
make
it
easy
for
people
to
use
by
adding
like
just
convenience
layer
on
top
of
it.
That
simplifies
this
simplifies
the
semantics
yeah.
A
And
that's
awesome,
that's
good
yeah,
and
also
that
means,
even
if
we
made
a
mistake
like
we
made
the
wrong
assessment,
like
actually
people
use
a
particular
scenario
more
often
than
we
anticipated.
We
can
add
that
later
without
having
to
break
the
entire
thing.
If
we
build
things
like
using
the
layered
approach,
we
have
the
fundamental
layer
that
is
flexible
and
also
that
that
gives
us
flexibility
if
today
we're
making
the
wrong
we're,
making
the
correct
judgment.
But
after
three
years
situation
changed,
we
have
that
flexibility.
A
I
I
think
the
challenge
is
most
folks
understand
that,
but
they
were
struggling
with
worse,
the
right
balance
and-
and
this
is
very
subjective,
so
so,
okay,
I
think
if
you
look
at
the
community
in
general,
like
there's
a
lack
of
clear
ownership
responsibility,
so
everyone
can
chime
in
and
put
their
idea
and
when
they
have
different
ideas,
we
don't
have
it
it's
basically
based
on
the
the
social
pressure
yeah
and
basically,
we
kind
of
agree
like
like
based
on
the
majority
like
agreed
on
something.
Then
they,
the
minority,
decide.
Okay.
A
B
That's
kind
of
how
things
do
work
I
mean
I
do
feel
like.
There
is
some
amount
of
structure
right
like
we
do
have
spec
approvers
and
tc
members
right.
That's
the
ultimate
authority
for
for
this
stuff,
but
I
feel
like
especially
with
metrics
we've
identified
that
the
it's
a
little
unbalanced
in
terms
of
like
the
amount
of
expertise
represented
in
in
those
groups.
Right,
I
think
josh
is
the
only
tc
member
who
actually
has
like
a
strong
background
in
metrics,
currently
yeah.
B
So
maybe
that's
part
of
the
problem.
So
if
you
took
a
bunch
of
experts
through
different
opinions
and
put
them
on
tc,
it
wouldn't
solve
anything.
I
suppose,
but
I
do
agree,
there's
an
issue
with
getting
people
to
bottom
line
things
that
can
be
a
little
different,
difficult
in
open
source.
B
What
we're
missing
and
like
come
back
with
some
justification
right
like
we
need
to
add
these
things
and
like
because
here's
an
example
of
like
other
stuff
that
has
it
that
people
like,
or
here's
an
example
of
feedback
from
users
about
why
they're
missing
it
or
here's
like
just
an
abstract
justification
for
like.
If
you
don't
have
this
data,
you
can't
perform
this
kind
of
monitoring.
You
know
just
whatever
it
is.
B
A
A
I'm
I'm
I'm
still
happy
to
work
on
that
and
just
feel
like.
If
it's
just
just
a
few
of
us,
I'm
I'm
a
little
bit
worried
about
like
what.
If
I'm
I'm,
not
a
matrix
expert
right,
and
I
treat
this
as
an
important
project,
and
I
I
fear
that
if
you
don't
have
enough
feedback,
I
might
make
some
wrong
decisions
or
or
in
general,
there's
a
ferry
in
the
communities.
Nobody
is
going
to
make
decision
that
pr
is
going
to
stuck
there
forever
and.
B
Yeah
I
mean
sometimes
it
is
scary.
I
mean
I
I
feel
like
I
get
you
know
put
in
this
position
for
things
like
the
last
one
for
me
was
like
versioning
right
like
we're,
gonna
hit
1.0,
but
we
don't
actually
have
strict
definition
of
stability.
B
Long-Term
support
guarantees
versioning,
you
know
any
any
of
that
stuff,
but
the
community
seemed
like
they
were
fine,
just
kind
of
declaring
1.0
without
it,
because
they
were
feeling
pressure,
and
so
I
had
to
like
push
for
it
and
it
was
like
fine
people
like
okay,
fine,
we
should,
but
that's
still,
the
end
result
was
like
if
I
didn't
actually
write
all
those
pr's
they
weren't
going
to
get
written
and
I
could
get
feedback
on
the
prs.
B
Then
we
at
least
have
something
that
that
I
would
say
is
like
confidently
serviceable.
It
probably
will
not
be
perfect.
It
probably
won't
never
be
perfect,
but
that's
that's
like
the
ultimately.
The
only
approach
we
can
take.
The
other
approach
is
the
dictator
approach.
Where
someone
is
a
super
expert
or
not,
and
they
just
pick
an
idea
and
shove
it
over.
Let's
mention
the
the
jerk
mode
of
running
oss
yeah,
and
that
I
mean
to
be
fair
that
if
that
person
really
does
have
a
good
intuition,
it
can
be
fine,
it
can
work.
B
I
don't
want
to
say
there's
a
morality
story
here
where
that
doesn't
work,
but
I
think
in
our
case,
the
approach
that
works
best
is
there
needs
to
be
a
champion,
and
that
champion
needs
to
be
proposing
ideas
but
synthesizing
the
the
feedback
into
oil.
But
there's
got
to
be
someone
at
the
center
who's,
who's,
writing
the
prs
and
proposing
the
ideas
or
yeah.
I
agree
it
doesn't
it
there's
no
way
to
generate
momentum,
and
this
is
maybe
just
how
people
work
on
some
level.
B
A
One
example
remember:
last
week
you
have
a
conflict,
you
won't
be
able
to
join
so
I
yeah.
I
brought
the
topic
that
we
we
can
move
the
meeting
to
8
a.m.
Okay,
but
one
thing
I
missed:
it's
not
clear
who's
going
to
move
that,
because
I
I
thought
I
I
don't
own
that
meeting
so
I
was
thinking
in
this
meeting
like
if
folks
join
I'll,
just
mention
like
riley
will
just
move
that
meeting
and-
and
please
tell
me
it's
okay.
B
Is
there
a
group
conscious
conscience
that
feels
this
is
the
correct
way
to
go,
and
then
you
act
on
behalf
of
that
group
conscience
to
make
the
change
and
any
member
of
the
group
can
do
that
so
long
as
they
are
acting
in
accord
with
the
group
conscience
on
behalf
of
the
group,
as
opposed
to
dictating
to
the
group
what
should
happen?
Yeah
and.
A
Yeah
yeah,
so
I
think
currently
it's
becoming
hard.
You
remember
in
the
spec
meeting
where
josh
from
google
mentioned,
like
hey
the
the
agenda.
Dog
has
had
multiple
meetings
agenda
mixed
up
and
he
would
want
to
split
the
data
model
part
and
he
got
an
agreement.
People
are
saying
oh
yeah,
like
like
when
we
put
my
name
in
the
attendees
list.
I
actually
put
that
in
the
wrong
spot
because
that
that's
reserved
for
the
next
meeting
sure
in
the
agenda,
but
he
never
achieved
that.
After
that
we
got
the
agreement
and
he
got
support.
B
And
then
there's
a
question
of
like
how
do
you
move
forward
with
it
and
I
think
maybe
there
that's
where
we
have
to
one
there's
like
literally
empowered
like?
Can
you
actually
move
the
meeting
you
know
or
do
you
have
to
bother
somebody
else
to
do
it
and
then
the
other
part
is
maybe
just
developing
this
conscience
of
service.
B
In
the
name
of
the
group
and
group
conscience,
where
you
say
hey,
you
know:
we've
we
appear
to
have
seized
consensus,
so
I'm
just
gonna
do
this
as
the
person
who
proposed
it
or
tap
someone
else
to
do
it
and.
A
I
think
there's
a
lack
of
clarity.
He
he
does
not
have
enough
permission.
That's
one
thing
number
two
thing
is,
he
doesn't
even
know
who
has
the
permission
to
give
him
permission
like
last
time,
yeah,
I
think,
josh
and
and
told
him
hey.
It
seems
I
joined
a
meeting
and
you
were
there,
but
you
disappeared.
B
A
B
Is
like
yeah,
this
is
so.
This
is
an
area
where,
where
we
don't
actually
have
access
spread
out
sufficiently,
which
is
around
zoom
right
like
sergey,
I
think
is
the
only
person
I
know
who
has
has
that
and
we
actually
need.
I
think
we
need
to
spread
out.
B
Part
of
it
is
like
zoom
makes
this
very
difficult
to
do
like
it's
difficult
to
give
out
like
host
privileges,
to
lots
of
people
in
like
a
generic
way,
but
maybe
we
can
set
it
up
to
ensure
that
in
any
given
meeting
there's
a
number
of
people
either
one
of
two
things:
either:
there's
a
mechanism
for
people
to
be
able
to
like
log
into
the
host
account
or
there's
a
mechanism
to
give
out
co-hosts
or
host
privileges
to
a
number
of
people
who
are
likely
to
be
in
each
meeting
just
so
that,
if
something
crazy
happens,
someone
is
either
there
or
it
could
be,
there's
a
list
of
people
that
could
be
grabbed
to
help
all
right,
because
this
has
come
up
several
times.
B
It
came
up
in
a
another
metrics
meeting.
I
think
data
model
where
someone
wasn't
on
mute.
Oh
it's
on.
B
B
And-
and
this
is
just
so-
we've
we've
just
this
is
an
example
where
actually,
we
need
to
spread
out
the
host
privileges
and
yeah.
So
so
that's
just
the
thing.
I
think
we
can
bring
this
up
with
the
gc.
A
Yeah
and
my
suggestion
would
be
people
who
have
the
access
to
schedule
meeting.
That
means
they
should
have
the
power
to
host
me.
A
We
have
ten
minutes
left,
I'm
gonna
share
my
screen
and
and
go
back
to
the
topic.
This
is
supposed
to
be
yeah,
so
so
we
have.
We
have
five
issues
since
last
week,.
A
Okay,
you
see
the
issue
like
it
seems
like
a
lot
of
mirror
faces
like
riley,
tigran
and
carlos,
and
we
just
need
more
and
we
need
a
way
to
recruit
more
people
and
retain
them
and
also
have
a
way
to
make
make
sure
they
feel
they
feel
comfortable.
I
I
figure
we
start
to
have
the
problem
like
people
who
have
been
in
this
like
on
these
products
for
longer
time.
They
know
how
it
works
so
either
they
kind
of
setting
a
low
expectation
or
they
just
figured
out
the
the
process.
A
B
B
B
You
know,
there's
the
the
gc
contains
a
lot
of
people
who
are
very
helpful,
but
but
in
kind
of
like
a
limited
domain
like
I,
you
know
I'm
on
the
gc,
so
I
see
the
work
they
do,
but
it's
sort
of
like
defending
the
project
from
crazy
stuff
coming
in
from
the
outside
or
dealing
with
corporate
politics
and
other
things.
So
they.
B
Job
at
that,
but
that
group
is
not
really
as
collective
is
not
focused
on
day-to-day
management
of
project
and
like
how
does
that
work,
and
how
does
that
go?
You
know,
like
sarah
novotny
is
awesome,
but
she's
not
paying
attention
to
the
project
in
that
form
right.
That's
not
her
role
in
the
project
and
really.
A
B
The
official
role
in
the
project,
for
that
is
actually
the
technical
committee
and
you
can
actually
see
that
in
our
charter
right.
The
governance
committee
draws
up
the
charter
for
the
technical
committee.
The
technical
committee
executes
it
sort
of
like
the
way
you've
got
a
board
and
a
c-suite
right
technical
committee's
a
c-suite
governance
committee's
board.
B
So
technical
committee
is
the
group
that
should
be
executing
all
this
stuff,
but
the
technical
committee
is
mostly
people
who
these
days
are
either
absent
or
they're
present.
But
heads
down
writing
a
lot
of
code
and
involved
a
lot
of
engineering.
We
don't
have
people
on
the
technical
committee
who
have
the
time
or
the
orientation.
B
It
seems
to
be
focusing
more
on
like
a
project
management
role
and
on
these,
like
kind
of
community
issues,
and
when
I
do
try
to
expand
that
encourage
like
go
to
other
people
and
try
to
encourage
them
to
take
on
that
role.
Unfortunately,
the
people
I'd
like
to
see
do
that
kind
of
imply
to
me
like
because
they
tend
to
be
more
like
managers.
You
know
they're,
like
I'm
already
busy.
You
know
like,
like
I'd,
love
to
see.
B
Joshua,
do
some
more
of
that,
but
you
know
he's
he
feels
maxed
out
I
kind
of
get
the
impression
you're
sort
of
maxed
out
a
little
bit.
So
that's
like
my
other
concern
is
the
experienced
people
who
I
think
could
do
that
role
that
I
seem
to
to
already
be
committed
to
the
project
to
the
degree
that
they're
able,
otherwise
they
probably
would
be
doing
this
stuff
already,
and
you
can't
exactly
ask
new
people
to
do
this
stuff.
So
right,
that's
just
a
genuine
concern.
I
have.
A
Have
some
idea,
so
I
think
josh
is
very
busy
because
he
has
a
smaller
team
in
google.
Yes,
he
made
a
lot
of
code
changes
in
the
java
and
c
plus
plus
repo,
which
I
I
think
the
ideal
case.
Is
it
shouldn't
they
should
find.
Someone
like,
I
know,
someone
from
his
team
or
also
very
smart
developers.
They
should
take
more.
A
A
The
memory
usage
is,
is
unbounded,
which
is
scary
like
whenever
we
receive
something
we
just
throw
that
in
the
task
pool
and
expect
the
system
right.
I
see
this
as
a
source
of
I
fixed
all
these
things,
but
now
I
think
it's
relatively
stable,
so
I
just
leave
the
entire
thing.
I
don't
even
have
time
to
join
that
sick
meeting.
Yeah.
B
Yeah,
so
we're
we're
growing
it,
but
slowly
you
know
that's
just
the
thing,
so
I
think
it's
on
us
to
continue
to
do
this
and
I'm
gonna
try
to
find
more
time.
Unfortunately,
I've
I
felt
these
past
couple
weeks.
I've
been
supposed
to
be
running
things
and
I've
been
too
distracted
by
a
lot
of
other
stuff.
So
I
actually
feel
like
I'm
behind
myself
on
on
this
role,
but
I'm
gonna
try
to
fix
that
next
week
because
same.
A
Based
on
url,
this
one
seems
like
folks
are
asking:
they
won't
be
able
to
access
the
the
url
from
the
sampler.
I
I
think
we
never
define
that
this
seems
to
be
related
to
the
semantic
convention,
but
more.
The
question
is
the
sampler
is
designed
to
be
super
fast,
so
it
doesn't
have
to
have
access
to
all
the
input
parameters,
but
we
never
specified
that
sampler
must
have
access
to
a
subset
of
things
right.
Well,
I
mean
it.
B
A
A
B
But
we
struggle
to
make
that
a
requirement.
I
mean
we
can
make
it
a
recommendation,
but
you
know
we
run
into
like
practical
scenarios
all
the
time
where
it's
not
practical
to
to
start
the
span
at
the
moment
that
you
know
so.
I
don't
quite
know
how
we
solve
some
of
those
issues.
You
know
there
could
be
an
additional
api
at
some
point
that
triggers
this
stuff
or
re-triggers
it.
But
I
don't
know
what
that
is.
B
A
B
B
A
A
B
B
My
my
my
s,
my
guess,
I
would
tell
him
that
the
second
one
is
is
safer.
He's
correct,
like
stay
away
from
name
spaces
that
open
telemetry
is
already
using
and
use
a
namespace
that
is
specific
to
the
project
or
the
organization.
That
is
clearly
the
safest
thing
to
do,
and
there's
nothing
harmful
with
doing
it.
That.
B
Yeah-
and
I
would
just
instead
of
just
say,
put
company
names
could
put
company
or
project
names.
A
This
is
more
like
some
something
that
folks
should
do,
but
it's
it's
not
done
yet
it's
basically
the
module
version
like
the
iphone
version,
the
library
version.
So
currently
there's
only
the
yes
library.
B
A
A
B
The
classic
example
of
text
map
is
things
like
whatever
that
java
protocol
is
it's
like
a
java
messaging
protocol
that
isn't
compatible
with
http
character
set.
B
A
A
I
think,
depending
on
the
scope,
we
might
say,
hey
like
he,
he
would
want
a
serializable
span
that
can
be
passed
across
different
contacts
in
the
same
process.
But
later
I
suspect,
if
we
take
this
approach,
folks
might
say
I
want
to
have
a
definition
of
how
to
civilize
spend
across
different
processes
across
different
languages,
and
it
doesn't
seem
to
be
it
it.
It
doesn't
seem
to
be
the
mission
of
open
climate.
I
think.
B
A
B
A
B
No
yeah
I
mean
we
have
otlp.
Is
our
serialization
standard
for
spans?
You
know
like
you
can
use.
You
can
serialize
it
into
otlp.
If
you
want,
you
could
write
a
you
could
write
a
propagator
right
like
you
could
create
a
a
you
could
create
a
custom
propagator
that
serialize
the
entire
span.
B
You
know,
there's
nothing
to
say
you
can't
do
that,
but
I
don't
know
that
I
suspect
that
won't
solve
their
problem.