►
From YouTube: 2022-04-13 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).
C
B
I
actually
I
need
a
pair
of
glasses.
I
I
wear
contacts,
but
I
realize
I
don't
have
a
backup
pair
right
now
and
you
know
I
should
get
one.
A
B
Nice
looks
like
the
the
oh.
This
is
messaging
semantics,
never
mind.
Let's
look
at
the
project
board,
I'm
like
it's
all
kind
of
filled
out
a
bunch
and
let
go
that's
the
other
board.
B
So
I
wasn't
able
to
make
it
to
the
spec
meeting
this
morning,
but
how
did
that
go?
C
C
The
second
thing
is
more
like
a
logistic
thing
that
I
reached
last
week
and
I
put
that
in
the
agenda,
so
this
is
try
to
clarify.
What's
the
purpose
of
this
meeting,
because
some
folks
believe
it
is
the
spike
meeting
for
impact
time,
though-
and
some
folks
think
this
is
the
instrumentation
meeting
so
tristan
and
I
actually
joined
last
week
with
the
understanding.
This
is
not
the
instrumentation
see.
This
is
the
spikesy,
it's
both
yeah.
So
so
I
have
a
pr
trying
to
clarify
what
it
is,
and
hopefully
we
can
get
agreement
on
that.
D
C
So
the
the
challenge
here
is:
if
it's
semantic
convention
sake,
then
people
who
join
this.
They
must
care
about
semantic
conversation,
and
maybe
I
should
just
quit,
because
I
I
care
about
the
api
sdk
semantic
convention
is
not
a
priority
for
me.
I'm
happy
to
help
to
facilitate
that,
but
it's
not
my
expertise.
C
If
it's
the
spike
sigma,
I'm
happy
to
facilitate
it
and
and
then
there's
another
very
related
question.
If
it's
everything
about
spike,
then
we
shouldn't
have
a
separate
mutinose.
It
should
be
merged
with
the
the
spec
sig
meeting,
which
is
8
a.m
in
the
pacific
time
in
that
way,
like
we're
running
kind
of
rotation.
So
if
we
have
certain
things
that
we
need
to
follow
up,
at
least
we
can
give
the
transparency
to
folks
who
are
not
in
this
timeline
like
the
atm
folks.
They
can.
C
They
can
see
what's
happening
here
and
we
can
also
see
what's
happening
there.
We
share
the
same
meeting
notes
in
this
way.
We
collaborate
more
so
it's
kind
of
we're
kind
of
in
the
out
spot
that
many
of
us
believe
this
thing
is
for
the
spec,
but
what
we
covered
here,
the
meeting
notes
and
topics
are
not
followed
up
in
the
spec
meter
unless
someone
explicitly
joined
that
meeting
and
put
something
in
the
agenda-
and
this
is
what
I'm
trying
to
solve.
B
We've
been
focused
on
instrumentation,
because
that's
that's
what
the
people
in
the
apac
time
zone
care
the
most
about
right
now
and
we
need
to
get
it
done
so,
and
there
hasn't
been
like
a
need
for
a
second
meeting
yet
so
I'm
fine
just
calling
this
a
apac
spec
meeting
and
just
saying
that
you
know
the
the
main
agenda
items
we're
trying
to
get
through
in
that
meeting
right
now
are
semantic
conventions,
if
that
makes
sense,
so
that
just
means
moving
the
meeting
notes
over
certain
take
meeting
notes
in
the
other
doc.
B
B
E
Yeah
previously
here
during
this
meeting,
we
had
a
gap.
That's
not
really
like
tc
members,
we're
not
joining
so
now
we
have
riley
joining
here,
but
the
riley's
role
here
is
mostly
about
like
facilitating
the
the
overall
stuff
here
right.
But
for
me
it's
like,
if
you
want
to
have
two
spec
meetings
right
now,
like
a
should
be
expect
dc
members
will
be
precipitating
both.
B
Yeah
I
mean
something
I'm
requesting
right
now
is
that
these
these
are
all
spec
meetings
right.
All
of
these
are
spec
meetings
and
there
needs
to
be
a
tc
member
present
in
in
all
of
them,
because
it's
not,
I
think,
it's
just
not
helpful
for
having
these
groups
move
along
without
some
connection
to
like
the
the
tc,
creates
a
situation
where
we
get
jammed
up
later.
B
When
we
go
to
submit
things
for
tc
review,
because
we
kind
of
like
just
pop
out
of
the
woodwork
and
surprise
everybody,
and
then
we
have
to
wait
to
get
a
review
so
and
unfortunately,
the
tc
members
who
are
interested
in
semantic
conventions
are
all
in
the
eu,
so
they
can't
come
to
this
meeting.
So
thank
you,
riley
for
for
being
here
yeah,
you
don't
have
to
have
a
strong
opinion
about
the
semantic
conventions,
but
if
you
can
just
help
make
sure
this
is
getting
pipelined
properly,
that's
that's.
That's
really
helpful.
E
Yeah
so
like
back
to
my
questions
like
I
said,
since
we
have
all
semantic
conventions
to
see
members
in
eu
time
zone,
does
it
even
make
sense
to
have
actually
to
call
this
4
p.m?
Meeting
as
a
spec
meeting
because
looks
like
we
are
not
going
to
work
on
so
many
conventions
here,
or
I
mean
it
any
like,
otherwise,
it
will
be
the
same,
the
same
problem
so
like.
E
Basically,
I
had
this
expectation
that
we
can
have
dc
member
and
we
can
reiterate
on
the
scope
for
we
won,
but
since
there
is
no
dc
member
who
can
you
know,
let's
say
make
a
call
here,
then
the
only
possibility
is
to
go
and
participate
atm,
meaning
to
to
be
able
to
talk
with
stc
members.
E
So
that's
that's
smartest
thing,
and
if
we
go,
if
you're
going
forward,
they're
going
to
have
the
same
kind
of
situation
and
there
is
no
tc
member
participating
at
4
pm
that
doesn't
make
sense
to
to
talk
about
such
things
here.
B
I
mean,
I
think
I
think
it
I
mean
we
need.
We
need
reviews
and
feedback
on
github
right,
that's
primarily
where
we
need
them
as
long.
B
E
Sure
yeah,
that's
a
really
good
point
then,
like
basically
probably
we
need
to.
We
need
to
figure
out
how
to
solve
this
problem
right.
So
we
have
a
people
and
we
have
a
community
participating
this
and
we
have
by
the
process
that
we
have
an
open-source
community.
We
should
be
dc
member
approving
all
this,
our
like
participating
and
going
forward
we're
approving
all
the
all
this
stuff
as
a
as
a
part
of
this
pacific
committee
yeah.
B
B
Yeah
anyways,
sorry,
my
cat's
being
crazy.
I
I
think
we
need
to
we
need
to
have
armin
and
carlos
and
anyone
else,
who's
interested
in
reviewing
our
work
being
like
prompt
and
present
with
reviewing
our
work.
I
don't
personally
feel
like
we
need
tc
members
to
like
guide
the
design
for
us.
B
I
think
we're
doing
a
fine
job
with
that,
but
we
do
need
this
stuff
when
we're
asking
for
reviews
of
our
work
for
that
to
be
be
propped
and
to
not
be
in
some
situation
where
we're
being
told
by
where
it's
just
like
hard
to
get
some
get
attention.
So
I'm
hoping.
F
B
By
having
riley
here
that
that
helps
solve
that
problem,
but
part
of
it
is,
is
following
up
with
the
other
tc
members
and
bringing
this
up
at
least
briefly
at
the
spec
meeting
at
8
am.
E
Right
but
I'm
like
actually
I'm
trying
to
to
solve
some
really
practical
problems.
So
we
have
this
old
tab
right
now
and
we
already
agreed
on
this
code.
There
yeah
and
it
was
approved
by
tc
members
several
months
ago,
but
the
problem
that
we
probably
discussed
last
time
is
that
since
we
already
have
it
stated
and
approved
it's
not
something
that's
widely,
I
would
say
understandable,
and
you
know
we
can
use
it
as
a
guidance.
I
mean
saying
that
we
resolved
all
the
items
in
this
scope.
E
It
doesn't
really
mean
that
we'll
be
having
this
support
from
dc.
In
a
way,
that's
we
can
start
like
announcing
https
spec
as
stable,
so
like
a
currently
I'm
just
confused
about
the
overall
kind
of
way.
So.
B
E
About
this
and
yeah
so
looks
like
the
only
way
to
to
be
on
the
same
page
and
to
like
you
know,
to
have
this
to
to
be
to
have
this
consistent
is
to
having
some
tc
members
actively
participating
in
the
discussions,
not
only
reviewing
the
proposals,
because
even
for
proposals
review,
it
takes
a
lot
of
time
from
dc
members
to
follow
up
and
just
to
understand
what
is
the
the
initial
goal
and
why
we
just
we
come
up
with
some
decisions
and
some
proposals,
and
we
also
saw
this
I've
seen
this
feedback
like
a
pushback,
for
example,
about
the
spam
links
right
in
bogdan's
feedback.
E
So
that's
that's
happens,
I
believe
just
because
we
don't
have.
We
didn't
have
this
like
an
open
precipitation
include
with
the
people
from
dc
committee.
So
that
that's
my
point
and
looks
like
if
we
proceed
the
same
way.
It
will
be
really
hard
to
do
the
same
thing
going
forward
because
we
have
a
lot
of
stuff
to
be
done.
E
Yes,
many
conventions
and
all
of
them
are
not
really
clear
and
like
they
are
similarly
clear
or
not
clear
as
spam
links,
for
example.
So
basically,
my
question
is
like:
if
I,
if
you
want
to
make
progress
on,
make
https
pack
stable,
which
meaning
I
need
to
participate,
let's
make
it
more
specific.
B
B
I
think
we
could
maybe
use
slack
more
for
some
of
this
discussion,
but
I
think
it
should
be
in
github
and
if
we're
working
on
like
new
design
ideas,
I
don't.
I
think
it's
it's
a
bit
moot
for
this,
getting
this
stuff
stable,
because
we're
basically
done
right.
So
it
really
is
just
a
matter
of
getting
these
last
things
approved
and
getting
them
to
verify
that
they
don't
see
any
missing
work.
B
So
I
think
the
action
item
is
to
to
have
a
follow-up
with
carlos
and
armin.
Maybe
on
slack
to
say
let
like
this
is
you
know
this
completes
this
otep
this
this
work
here.
B
B
I
think
again
making
sure
we
have
issues
that
are
tracking
everything,
we're
working
on
so
like
if
we
have
a
a
rough
design
dock
we're
working
on
in,
like
a
google
doc,
making
sure
there's
an
issue
linked
to
it
and
making
sure
we're
pinging
tc
members,
again,
maybe
in
slack
to
say
hey
like.
Can
you
review
this
design
dock
and
if
we
need
to
have
face
time
with
them,
then
using
the
the
tuesday
morning
meeting
meeting
for
that.
E
B
On
the
on
the
sp,
so
a
thing
we
didn't
do
here
that
we've
done
in
the
past
is
create
like
what
we
call
like
a
tracking
issue
so
that
that
could
have
looking
back.
That's
the
thing
we
could
have
done
here,
which
was
take,
take
the
set
of
goals
that
we
wrote
in
the
otep
and
just
like
have
them
in
a
spec
issue
with
a
checklist
where
we're
checking
them
off
as
we
go.
B
E
But
sure
for
sure
yeah
we
do
have
all
the
stuff
available.
The
only
the
only
goal
right
now
is
to
involve
tc
members
to.
E
And
make
every
everything
clear
for
everyone
and
explicitly
have
their
approval,
like
maybe
like
some
specific
person
from
tc
or
who
are
interested
in
some.
E
Like
to
have
them
specifically
approve
that,
did
you
think
that
they
are
like?
I
agree,
work
but
yeah.
We
have
this
in
this
otap,
and
this
in
this
board
represented
right
now,
and
we
can
probably
then
discuss
this
issue
during
atm
meeting
or
next
next
tuesday.
B
Yeah,
so
I
think
that
yeah,
the
action
item
is
like
after
this
meeting
you
know
to
to
ping
carlos
and
armin
and
slack
and
like
get
their
attention.
Just
just
to
reiterate:
hey
you
guys
are
the
ones
interested
in
semantic
conventions.
We
think
we're
just
about
done.
B
This
is
like
the
project
board
that
contains
what
we
think
is
the
last
items
we
need
for
stability.
Can
you
like
get
up
to
speed
on
this
and
approve
this
stuff,
and
if
you
see
something
missing
that
you
want
to
have
happen,
can
you
can
you
make
an
issue
for
it
or
talk
to
us
about
it?
B
E
B
And
I
think
we
should
add,
I
don't
have
a,
let
me
find
a
link
to
it,
but
ludmila's
pull
request,
open,
pull
request.
I
believe
that
resolves
all
of
this
remaining
stuff.
B
C
Just
from
my
curiosity,
is
there
any
agreement
from
army
or
cuddles
who
will
be
helping
to
drive
the
invitation
safe.
B
They've
they've
both
expressed
expressed
interest
and,
like
you
said,
they've,
both
listed
semantic
conventions
as
their
area
of
expertise.
So
so
I
think
they're
they're,
interested
or.
C
And
then
here's
my
suggestion
type
allow
me
to
interrupt
so
I
I
think
the
folks
who
are
interested
like
army
semantic
conventions,
carlos
semantic
conventions,
josh
suresh,
simon
though,
but
but
he's
out.
So
maybe
we
we
should.
We
should
give
him
more
time
and
then
tigger
the
schema
and
version
right.
So
he
needs
to
reach
out
to
four
of
them
and
don't
expect
a
reply
from
josh
zurich
because
he
he
has
family
affair
to
take
care
of.
C
So
at
least
from
these
three
tca
members,
someone
like
like
dennis
if
you're
driving
this
you
should
reach
out
to
three
of
them
and
at
least
I'll
say,
got
two
of
them
explicitly
expressed
they're
committed
they're
they're,
going
to
join
the
meeting
and
help
to
to
make
progress
here,
yeah.
So.
B
They
can't
join
the
meeting.
They
can't
join
this
meeting
during
the
wrong
time
zone.
Yep,
that's
another
problem
so
but
yeah
reach
reach
out
to
them.
Thank
you
and
please
do
it
in
the
in
the
instrumentation
channel,
not
as
dms.
If
you
don't
mind
just
I
think
that
would
be
better
and
yeah.
I
would
say
honestly,
don't
don't
reach
out
to
josh
he's
he's
got
some
serious
family
matters
right
now,
he's
not
he's
not
around
so
and
yeah.
B
I
think
that's
kind
of
like
the
next
step,
so
we
aren't
having
to
have
like
this
discussion,
and
these,
like
you
know
like
the
going
forward,
the
tc
will,
as
a
group
will
be
more
just
kind
of
aware
of
the
different
things
going
on
and
and
being
divided
up
and
committed
to
supporting
the
various
things
and
having
some
board.
That's
like
tracking
tracking
that
level
of
stuff
is
like,
I
think,
the
final
piece
of
this
puzzle.
C
Yeah
exactly
that
you
read
my
mind,
so
I
would
say
the
first
thing
is
to
identify
the
right
group,
how
people
otherwise
will
repeat
the
similar
thing
like
we
agree
on
something
within
a
small
scope.
Folks,
and
then
we
bring
that
to
the
larger
city,
and
then
someone
jumped
out
and
said
no,
so
once
we
have
the
the
right
side
help
folks,
especially
the
tc
members.
F
C
Sponsoring
this
the
number
one
thing
is,
we
should
review
the
board
and
make
sure
the
board
is
accurately
re,
reflecting
the
current
fact,
like
anything
that
we
we
think
we
should
address,
will
be
tracked
by
the
board.
Anything
else
shouldn't
be
there
yeah.
So
far
there
doesn't
seem
to
be
agreement
and,
and
seems
there,
there's
some
wider
topic.
People
are
even
talking
about.
C
Maybe
we
should
try
to
extract
the
attributes
that
are
not
specific
to
any
signal,
and
maybe
http
status
should
just
be
status
for
anything
like
logs
traces
and
metrics,
and
I
kind
of
agree
with
that.
But
do
we
want
to
block
on
this
until
this
problem
is
solved
or
we
can
just
go
ahead
and
release
http
and
who
should
agree
on
that?
So
so
those
are
the
things
we
probably
need
to.
First
get
some
alignment.
C
I
I'll
give
you
one
specific
example,
so
you
can
you
can
see
what
what
what
I'm
I'm
trying
to
point
out.
So
I
have
a
question
here.
The
current
http
scheme
is
already
documented,
but
my
point
is,
it
seems
your
uri
scheme
makes
more
sense
because
scheme
is
not
an
http
specific
thing.
If
I
have
iftp
do
I
call
that
iftp.skill?
C
I
think
it
wouldn't
make
sense.
But
how
do
we
agree
here
and
seems
like
based
on
the
reply,
seems
there's
a
different
set
of
people,
I'm
not
seeing
any
one
of
them
joining
this
meeting
and
they're
they're
saying
a
different
direction.
So
this
is
a
problem
we
have
to
generally
align
like.
Basically,
if
people
are
saying
we
don't
agree,
because
this
is
very
subjective
thing
right.
Anyone
have
their
own
personal
opinion.
But
if
someone
joins
say
we
we
don't
agree.
At
least
we
can
tell
them.
C
B
So
I
don't
think
we
can
say
that
a
meeting
is
the
place
where
agreement
happened.
Right
like
that
has
to
be
github
yeah
and,
like
I
think
we
have
to
avoid
saying
you
must
come
to
a
specific
meeting
if
you
want
to
have
a
say,
but
I
think
what's
missing
is,
I
think,
you're
totally
correct
more
than
in
any
other
area.
This
is
an
area
where
you're
going
to
have
like
a
split
vote.
B
B
For
backwards
compatibility
reasons
done
or
we're
going
to
switch
it
to
uri
as
a
breaking
change,
because
that's
really
important.
But.
C
Yeah,
I
agree
what
I'm
saying
is.
I
totally
agree
with
you
like
like
so
far.
Nobody
in
the
community
has
the
ultimate
power
to
decide
right.
I
think
it's
based
on
the
common
consensus.
B
Right
and
frankly,
I
don't
think
anyone
in
this
group,
part
of
my
french,
like
gives
a
about
those
little
details.
Mostly,
we
just
want
it
stable
and
done
yeah.
So
that's
that's
like
our
main
ask
is
like
just
just
pick
pick
something
please
so
they
they
all
seem.
B
Generally
speaking
like
like
the
work
like
the
the
harder
work,
is
the
work
we're
putting
proposals
on
which
is
stuff
like
span
structure
for
retries
and
redirects,
and
all
that
that
stuff-
and
you
know,
normalizing
things
when
the
optionality
is
actually
making
a
mess
of
it
like
we're,
putting
proposals
for
that.
But
for
a
lot
of
these
things,
it's
just
like
fine.
You
know
I
would
prefer
there
not
be
thrash.
B
That's
my
only
request,
but
again,
I
think
this
just
needs
tc
members
responding
quickly
to
this
stuff,
which
I
think
really
means
when
these
things
come
into
this
backlog
that
they're
getting
like
assigned
to
the
right
tc
members
who
are
then
like
responding
to
them
quickly.
You
know
like
right
now
that
issue
is
like
assigned
to
josh.
Like
I
don't
know,
josh
cares
and
he's
busy
yeah
so
like
it
should
be
reassigned
to
carlos
or
or
armin
yeah
and
you're.
A
C
Yeah
and
another
thing
is
once
there's
a
general
consensus
on
the
on
the
items
attracting
the
product
of
board.
Maybe
what
what
we
should
do
is
to
leverage
the
five
minutes
time
box
to
triage
any
new
issues
like
if
there's
anyone
saying
I
want
to
block
this
effort.
I
I
think
this
issue
xyz,
is
super
important
and
before
we
can
resolve
that,
we
shouldn't
release
stable,
then
they
should
come
to
the
spec
meeting
and
speak
out
otherwise,
like
we
don't
even
know
how
to
how
to
make
progress.
Anyone
can
come
and
stop
this
effort.
B
Well,
I
mean
people
can
make
comments
right.
The
only
people
who
can
block
something
are
approvers
in
the
tc
right.
They
can
request
changes,
use
the
request,
changes
feature,
which
means
we
have
to
change
something
and
should
do
that
in
perhaps
in
the
face
of
other
people
in
a
github
thread.
Wanting
wanting
something
done,
but
I
think
it
mostly
most
of
the
problems
we've
had
so
far
is
just
that.
C
B
Exactly
and-
and
so
that's
what
we,
I
think,
as
dennis
has
said,
like
that's
the
thing
number
one
thing
we
need
to
establish
right
now
is
that
for
these
things,
when
they
show
up
will
armin
and
carlos
or
somebody
else,
if
they're
not
going
to
do
it
like
like
reliably,
show
up
and
respond
to
these
things
and
should
get
them
assigned
to
them
when
they
come
in
and
that
in
general
the
tc
can't
just
say.
Well,
we
don't
want
to
deal
with
it
right
now,
because
it's
it's
too
late.
B
C
B
Yeah,
please
please
let
the
rest
of
the
tc
know
about
this
and
I'll
maybe
see
you
on
friday
to
talk
about
it
more.
C
B
B
E
Well,
liz
is
not
here,
but
I
trying
to
follow
up
follow
this
conversation.
It
was
a
good
conversation
between
ameren,
probably
I
mean
yeah
and
ludmila
so
looks
like
there
is
some
consensus
there,
but
overall
I
believe
it
will
be
good
to
have
more
still
still
more
attention
right.
Okay,
of
course,
there
are.
There
are
some
things
that
probably
should
be
clarified.
Yeah.
B
But
it
looks
like
tigran's,
responding
like
in
general
this
this
looks
very
active.
Yes,
it
doesn't
look
like
it's.
It's
being
hung
up
in
the
way
that
that
we
don't
want
it
to
be
hung
up.
E
Yeah
the
conversation
with
stigram
is
was
mostly
about
the
like
what
what
is
the
breaking
changes?
What
is
the
optional?
What
up
optional
means
so
can
we
like
remove
optional
attributes
in
the
next
version
or
now,
but
it's
still
experimental
so
can
we
just
remove
it
and
it
was
in
like
up
there?
E
They
were
mostly
discussing
this
server
underscore
name
attribute
and
like
changing
its
semantics,
is
it
good
or
not,
but
looks
like
they
had
some
some
agreements
and
some
consensus
on
this
and
all
the
other
changes
probably
just
needs
to
be
reviewed
by
some
other
people
like
umbreon,
armin,
sorry
or
carlos.
So.
E
B
B
E
Hope,
that's,
it
will
be
possible
to
resolve
all
the
all
the
open
questions
during
this
week
and
it
looks
like
next
next
tuesday
8
a.m.
We
just
need
to
agree
on
the
on
the
things
that
miller
is
doing
here
and
the
overalls
go
and
it
might
be
even
possible
for
us
to
make
it
make
everything
by
that
time.
B
Think,
but
we
think
we're
done
right
like
we
think
we
think
this
is
fine.
We
we
could
move
forward
with
fleshing
out
context,
propagation
and
some
of
these
v2
issues.
At
this
point,
like
we've
got,
we've
got
20
minutes.
We
want
to
do
that.
B
But
but
yeah
otherwise
we're
just
we're
just
trying
to
get
this
final
thing
over
the
the
finish
line
and
like
that
plus
this
action
item
of
pinging
garmin
and
carlos
and
slack
and
just
being
like.
We
think
this
is
the
last
thing
for
stable
to.
E
My
focus
right
now,
so
I'm
trying
to
be
more
focused
on
these
particular
items
and
you
know
make
it
make
it
done,
because
a
lot
of
things
that
we
have
for
like
in
our
in
in
our
team
and
the
the
like
another
teams
here
in
microsoft,
really
depends
on
the
status
of
the
semantic
conventions.
B
B
B
I
can
reach
out
to
michael
lee
who's
and
see
see
if
he
can
help
james.
I
don't
know
if
there's
people
at
at
atlassian
who
are
I
don't
know
how
much
sql
or
other
databases
you
all
use
over
there,
but
yeah
we're.
D
A
big
postgres
shop
yeah
in
our
team,
specifically,
we
don't
use
a
lot
of
postgres
though,
but
exactly
that's
why
we
need
subject.
D
F
D
Maybe
not
from
my
end,
but
I
can
have
a.
I
can
have
an
ask
around
see
if
anyone
yeah
nobility
team
would
be
interested
yeah,
presumably.
B
People
you
have
some
people
there
who
operate
postgres
and
thus,
like,
I
think,
that's
the
the
people
we're
interested
in
are
not
architectural
astronauts.
So
much
as
like,
like
people
who
who
have
to
operate
these
things.
Who
can
tell
us
like
this?
This
is
the
information
I
I
care.
If,
if
I
have
this,
I'm
good,
if
I,
if
I
don't
have
this,
my
life
is
hell.
Yeah,
yeah,
yeah,.
D
F
B
D
D
B
We
haven't
been
making
it
a
requirement
for
stability,
but
it's
more
that
it's
a
requirement
for
any
breaking
changes.
Post
stability
right,
that's
that's
when
that's
when
we
need
it.
B
Yeah
speak
the
devil.
So
no
we
don't.
We
don't
need
it
done
for
4v1
or
stability
for
this
stuff.
It's
just
I'm
happy
to
see
it
see
it
happening.
It's.
E
A
requirement
for
the
sdk
or
agents
right
to
in
order
to
operate
with
this
many
conventions
properly.
It
should
be
done
on
that
level.
B
Yeah
it's
a
requirement
for
once.
We
declare
something
stable.
If
we
go
back
in,
like
we
say,
http
is
now
stable
and
then
someone
is
like
actually
on
second
thought.
It
should
be
uri.scheme
instead
of
http.scheme
or
some
like
that.
Then
it's
like,
if
we're
going
to
make
that
change
after
we've,
declared
it
stable,
then
as
part
of
making
that
change
to
the
spec.
B
We
have
to
also
issue
like
a
schema
translation
like
that's
just
I'm
just
saying
like,
and
we
should
probably
get
that
written
down
in
the
spec
as
a
rule,
but
that's
like
the
rules
going
forwards
once
this
stuff
becomes
stable
is
like
you,
you
can't
change
the
stable
stuff
unless
you're
also
issuing
a
translation
that
allows
it
to
be
backwards
compatible
and
if
it's
a
change
that
cannot
be
implemented
through
our
schema
processor,
then
it's
not
allowed,
like
that's
that's
kind
of
how
we're
going
to
define
backwards
like
backwards
compatibility
going
forwards.
D
I
think
tigran
had
a
pr
up:
oh
just
got
merged,
actually
regarding
defined
semantic
conventions
and
instrumentation
stability.
D
So
I
think
it
actually
does
go
into
that
and
and
basically
explains
that
you
know
yeah
what
the
plan
is
behind
changes
to
the
semantic
invention,
yeah.
It
recently
got
merged
like
a
few
days
ago,
so
that
yeah
that
I
guess
that
helps
in
that
regard.
Right.
E
I
will
follow
up
on
with
this
yeah
then
just
understand
more,
but
yeah.
Actually,
a
really
good
point
so
like
for
us
to
make
in
order
to
make
spec
stable.
Should
we
also
come
up
with
a
schema
for
http.
B
Well,
we
I
think
we
already
have
one
right
like
this
stuff
is
defined
as
a
schema,
essentially
like
that's
what
the
semantic
conventions
are
for
the
spec
it's,
and
so
when
the
spec
gets
released,
it
has
a
version
number
and
that's
the
version
number
for
the
schema.
I
believe,
and
then,
if
there's
a
change,
but
that
happens
between
one
release
of
the
spec
and
the
next,
then
we
have.
B
We
have
schema
translators
now
in
the
in
the
collector
that
someone
could
install
if
they
want
the
old
version
instead
of
the
new
version.
Essentially
and
enough
information
comes
across
the
wire
to
identify
what
what
version
of
the
schema
is
is
being
used,
but
I
believe
it's
tied
to
the
the
semantic
convention
version
number.
I
believe
right
I'll
try
to
get
more
familiar
with
you.
B
We're
just
building
in
the
collector
for
the
time
being,
there's
no
reason
you
couldn't
build
it
as
a
as
processors.
You
know
for
the
sdks,
that's
just
a
huge
amount
of
work
so.
B
Yeah,
if
someone
doesn't
want
to
run
a
collector
okay,
but
I
think
I
think
we
could
add
those
on
like
like
if
someone
wants
one
of
those
and
wants
to
just
like
contribute,
one
like,
I
think,
that's
fine
but
like
I
think,
realistically
we're
just
gonna
stick
to
the
collector
for
for
doing
that
kind
of
stuff
and
yeah.
I
would
argue
that
it's
just
in
general,
it's
good
to
be
running
a
collector
in
your
pipeline
and
to
put
all
of
this
processing.
B
You
know
on
a
machine
that
isn't
running
your
your
application
servers.
So
I
don't.
E
Like
to
point
out
challenging
to
run
the
collector
all
the
time
right,
it
really
depends
on
the
configuration.
The
the
platform
that
you
are
using,
for
example,
like
in
some
kubernetes
environments,
is
really
hard
to
to
run
these
collectors.
E
Yeah,
so
sidecar
is
something
that
you
can
do
definitely
so
this
is
this
basically
something
that's
tied
to
your
kind
of
workload
in
this
case,
but
this
is
overhead
right
because
yeah
that's
additional
process
running
for
each
of
the
of
your
replica
and
if
the
like,
a
telemetry,
let's
say
the
the
amount
of
data
that
you
are
processing
or
you
are
standing.
F
E
B
B
Is
gmp
trash?
It's
trash
man,
I've
been
every
project.
I've
been
on.
That
starts
with
grpc,
eventually
rips
it
out
in
favor
of
http,
but
my
honestly,
my
biggest
problem
not
to
get
a
side
ramp,
but
my
biggest
problem
with
grpc
is
actually
dependency
management
in
that
project.
Often
they
make
a
lot
of
breaking
changes
and
stuff,
and
it's
very
easy
to
end
up
with
a
transitive
dependency
conflict
between
two
different
libraries
that
both
depend
on
incompatible
versions
of
grpc.
D
B
Yeah,
it's
it's
part
of
the
reason
why
I
know
we
had
to
care
so
much
about
that
in
open
telemetry,
that,
like
the
api
package
packages
and
all
of
that
stuff
that
we
ship
all
the
required
parts
of
open,
telemetry
cannot
end
up
causing
these
transitive
problems.
That's
why
we
need
super
hardcore
backwards,
compatibility
at
that
layer
and
stuff,
because
it's
like
grpc
specifically,
is
the
nightmare.
B
I
keep
running
into
around
transitive
dependency
problems,
and
so
I
just
know
how
nasty
it
would
be
if
you
ended
up
with
a
breaking
change
in
the
open,
telemetry
api,
and
you
had
a
whole
bunch
of
libraries
now
that
you
don't
control
third-party
libraries
that
are
using
open,
telemetry
with
incompatible
versions.
That's
my
nightmare.
F
A
E
B
Like
we
don't
have
to
use
it,
that
part
is
totally
optional.
As,
like
the
operator
application
owner
you,
you
can
pick
which
thing
you
install
there
in
your
app.
That's
the
that's
not
glued
to
some
web
framework
that
has
open.
E
B
I
recommend
otlp
over
http
proto.
B
Personally,
I
think
that's
that's
the
optimal
protocol,
but
if
you
like
jrpc
and
are
happy
with
it,
then
you
know
by
all
means
use
that
one
personally,
I
think
http
over
proto,
like
you
mentioned,
like
load
balancing
all
this
other
stuff,
like
I
just
feel
like
life
gets
easier.
We
don't
really
utilize
any
of
the
special
features
grpc
has
in
it
we're
not
doing
two-way
communication
or
right
stuff,
like
that,
you
know
so
I
just
I
personally
found
http
proto
over
http
to
be
like
less
of
an
operational.
E
And
another
another
problem
that
we
also
like
I
faced
here-
is
it's
probably
the
compliance
right.
So,
in
order
to
run
agent
and
some
cooperative
workload,
you
need
to
be
really
really
pushy.
I
would
say
to
to
to
like
defend
this,
this
decision
so
and
that's
another
problem.
So
that's
what
I'm
asked
like
what
what
yeah?
What?
What
is
the
story?
Let's
say
for
us
and
for
the
schema
version,
and
how
to
verify
how
how
to
yeah.
B
Yeah,
I
think,
just
realistically
we
we
don't.
We
only
have
the
bandwidth
for
for
maintaining
those
in
the
collector
right
now
and
if
somebody
wants
them
as
like
sdk
processors
in
some
language,
then
they
would
need
to
to
contribute
them.
But
I
think
the
flip
side
is
these.
Things
are
probably
not
a
heck
of
a
lot
of
work.
B
You
know,
but
that's
that
that's
what
I
would
say,
yeah
in
some
corporate
environments,
the
opposite
by
the
way.
It's
it's
very
easy
to
deploy
a
new
workload
like
the
collector
and
control
that
thing
and
configure
it,
and
it's
very
difficult
to
get
into
all
the
application
services
and
change
how
they
work
or
add
a
plug-in
or
something
like
that
so
yeah.
But
I
agree
we
need
to.
We
need
to
support
both.
B
C
D
The
with
the
proto
over
http
is
that
just
as
text
or
is
it
like
binary
in
the
http
binary.
B
Yeah
yeah
I
mean
grpc
is
just
all
the
goods.
In
my
opinion,
all
the
good
stuff
in
grpc
is
actually
http.
2.
htg2
fixes,
pipelining
and
fixes
binary
and,
like
all
of
this
other
stuff
and
grpc,
is
like
fine,
but
it's
just
this
extra
layer
of
stuff
that
kind
of
like
shifts
around
sitting.
On
top
of
that
thing,
and
I
find
that
in
that
extra
layer,
stuff
breaks,
things
happen,
trying
to
debug
that
code
is
there's
just
more
code
and
and
I
hate
code
so
when
I
have
like
between
that
or
the
option.
E
Actually
this,
but.
E
Sorry,
oh
yeah,
I'm
sorry
this!
This
load
balancing
problem
comes
not
from
jrpc
but
from
http
2..
E
This
stuff,
basically
capture
the
traffic
and
then
balance
on
the
application
level,
so
yeah
jrpc
is
basically
is
just
another
thing
which
just
like
allows
you
to
use
this
product,
buff
sterilization
and
all
the
other
yeah,
and
have
these
protofiles
and
whatnot.
But
yeah.
The
real
problem
is
just
http
yeah.
B
Yeah
you
need
to-
and
I
I
forget
I
should
know
this
aspect
of
our
protocol,
but
I
forget
what
our
defaults
are
around
stickiness
but
yeah
like
that,
the
answer
is
you
want
that
to
be
low?
That's
a
thing
you
right
like
in
general,
you
want
to
be
tuned
low.
The
the
overhead
of
reestablishing
that
connection
is
like
not
a
big
deal
but
yeah
totally.
It's
true,
though.
B
D
Yeah,
I
think
the
other
thing
that's
worth
pointing
out
that
does
it's
not
just
to
my
understanding,
not
just
for
converting
between
versions
of
the
spanning
convention,
which
obviously
it
can
do
that,
but
it
just
all
you
need
to
do
is
put
a
schema
url
in
and
it
will
grab
a
schema,
so
you
can
really
convert
between
any
semantic
convention.
If
you
wanted
to
define
your
own
one
or
you
know,
convert
between
diff
versions
of.
E
It
is
yeah,
I
totally
agree
and
yeah.
I
will
be
actually
interesting
to
know
how
people
do
these
semantic
conventions
and
changes
in
different
companies,
so
like
a
what
brazil
atlas
and
folks
do,
and
what
like
aws
folks
do
and
live
step,
maybe
so
like
what
is
the
story
there?
Because
for
us
we
just
need
to.
We
also
work
with
a
lot
of
many
customers
right
and
it's
really
hard
to
actually
make
everyone
running
on
the
same.
E
Let's
say:
version
of
semantic
conventions
and
even
still
semantic
conventions
themselves
are
not
stable.
It's
really
hard
to
convince
people
to
use
anything
any
kind
of
sdk
because
it
will
be
changed
going
forward.
So
it
will
be
good
to
know
all
these
stories
from
like
internal
stories,
let's
say
for
from
companies
who
are
actually
doing
all
these
things
just
to
learn
from
each
other.
D
Yeah
we're
still
in
the
process
of
trying
to
get
everyone
on
tomato
convention
and
we're
really
just
starting,
but
it's
it's
not
an
easy
thing
to
do.
Is
it
and
I
feel
like
there
needs
to
be
some
kind
of
reason.
There
needs
to
be
some
kind
of
benefit
that
a
developer
gets
by
moving
to
this
major
convention,
and
that's
there's
a
lot.
D
I
mean
there's
a
lot
of
things
you
can
do
but
yeah
there's
it's
it's
not
easy,
yeah
and
I
think
the
easier
it
is
for
the
developer
to
adopt
the
standard
hotel
like
offering
the
better,
for
instance,
I
think
stuff
like
the
micrometer
bridges
and
the
like
spring
cloud,
sloth
bridges,
because
they
can
just
switch
out
the
implementation
and
use
the
same
api,
at
least
in
the
java
world.
That
makes
things
easier
and
then,
if
you
can
provide
benefit
by
to
them
by
sending
a
standard
model
of
data,
you
know.
B
D
Whatever
you
do
in
the
back
end,
then
that's
yeah.
I
feel
like
it's
a
combination
of
all
those
things
making
it
easy
for
the
developer
and
also
making
it
beneficial
once
they
do
it.
Yeah.
B
One
of
the
big
value
adds
going
forward
that
and
the
ability
to
take
the
api
and
not
just
the
conventions
being
stable,
but
the
fact
that
the
api
has
no
dependencies
and
isn't
going
to
cause
these
dependency
conflicts
by
being
making
breaking
changes,
means
that,
like
library,
like
libraries
and
things,
can
start
instrumenting
so
like
application
developers
that
won't
see
as
much
of
a
change
but
for
people
who
write
open
source
libraries,
I
think
that'll
be
like
a
huge
win
when
they
realize
they
can
do
that,
because
that's
traditionally
a
big
pain
in
the
butt.
B
Yeah
yeah,
I
think,
if
we
solve
all
this
stuff
for
ourselves,
but
it's
extensible
then
being
able
to
like
teach
people
that
it's
like
yeah
like
you,
can
follow
our
lead
and
create
your
own
semantic
conventions
and
write
your
own
processors
when
you
update
them
and
all
of
that-
and
I
think
that's
going
to
be
valuable
also
when
we
start
having
all
the
signals
together
and
they
start
cross-correlating
with
each
other.
That's
that's
more
like
a
longer
term
benefit.
B
But
that's
that
that's
like
the
main
reason
to
use
open,
telemetry
metrics
right
is
that
you'll
be
able
to
connect
those
metrics
with
your
traces
and
start
moving
back
and
forth,
but
that
that's
it'll
be
a
while.
I
think
for
people
realize
that,
because
it'll
have
to
become
features
in
all
of
these
different
analytic
tools,
but
but
I
think
it's
all
coming
yeah.
E
Actually,
if
there
are
some
success
stories
like
using
so
many
conventions
or
doing
some,
you
know
upgrades
of
semantic
conventions
and,
like
I
just
reusing
them
somehow
to
read
this
like
a
blog
post
or
you
know,
I
have
seen
to
see
some
talks
about
this.
B
Yeah,
I
definitely
know
people
I
mean,
even
without
this
particular
processor,
who
do
exactly
that
right
who
have
used
the
collector
to
to
really
manage
smooth
out
their
observability
pipeline
where
they
have
this
really
fractured
landscape
of
stuff.
B
That's
all
emitting
different,
different
things
from
like
different
eras,
and
all
of
that
you
know
and
being
able
to
like
pipe
all
of
that
through
a
layer
of
collectors
that
then
normalizes
that
data
as
much
as
they
can
and
then
what's
coming
out
of
the
collector
is
now
just
like
one
protocol
with
like
mostly
normalized
stuff
I've.
Definitely
I
I
don't
know
who
I
can
talk
about
in
public,
but
that's
I've
definitely
seen
big
organizations
see
a
lot
of
success.
There.
E
Yeah
yeah.
We
also
do
exactly
the
same
problem
right
now
like
we
are
also
building
the
problem
like
a
platform
at
microsoft
and
we're
going
to
have
a
lot
of
different
teams,
sending
us
different
data
and
and
semantic
conventions,
something
we
can
make
it
stable
for
sure.
But
starting
from
some
time
there
will
be
differences
right
and
different
teams
can
can
send
us
different
data
and
different
versions
like
even
even
if
all
this
data
is
stable,
it
can
be.
It
will
be
different.
B
Yeah,
like
one
thing
I
predict,
will
be
useful
for
people
is
a
a
data
dog
translator,
because
that's
another
like
that
stuff
is
just
really
common
right,
like
there's
a
bunch
of
stuff
with
datadog
or
some
other
thing.
You
know
that
has
its
own
conventions,
that's
like
widely
used,
and
you
can
and
there's
a
receiver
for
that
data
in
the
collector.