►
From YouTube: 2022-02-23 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).
B
B
C
B
C
We
were
mainly
just
talking
about
anarag
was
with
us
as
well
he's
just
not
on
the
list
talking
about.
C
Basically
the
the
problems
we've
had
like
it
seems
like
this
semantic
invention
doesn't
get
a
lot
of
attention
and
a
lot
of
the
discussion
regarding
it
kind
of
just
fizzles
out
and
a
decision
can't
be
made.
I
think
that
has
like
obviously
slowed
down
anything
getting
too
stable.
B
Well,
we've
been
now
made
approvers
over
all
of
this
stuff,
so
we
do.
That
was
a
shift
like
we
have
had
authority
officially
handed
to
us
to
to
shovel
this
stuff
in
if
we're
satisfied
with
it,
so
I
think
it'll
be
a
bit
of
a
just
like
a
boomerang.
You
know
it's
sort
of
like
people
are
gonna,
look
to
us
to
say:
we've
we've
really
kicked
the
tires
on
this,
and
and
and
it's
good
to
call
this
stable.
C
Yeah,
that's
true,
that's
true
yeah!
I
I
did
mention
that
yeah.
We
came
through
this
that
that
obviously
will
be
a
help.
Yeah
definitely,
but
anarag
was
also,
I
think,
mentioned
well.
We've
like
there's
a
green
tick,
but
it
doesn't
necessarily
like
move
everything
forward
like
it's.
It
doesn't
necessarily
completely
solve
that
problem,
but
it
will
definitely
help
but
yeah
yeah,
because
it
seems
like
those
two.
C
Pr's
that
we
have
open
now
are
kind
of
just
like
I
know
dennis
you
were
having.
I
know
if
you
have
any
updates
regarding
that,
but
I
know
that
you
were
having
trouble
getting
kind
of
agreement
or
getting
replies
and
stuff
like.
A
That
yeah,
basically
the
we
are
the
same,
the
same
stage
right
now.
The
same
state
is
the
same,
so
no
progress,
unfortunately,
and
yeah.
Basically,
we
are
with
with
my
pull
requests.
A
It
is
in
the
same
state
for
three
weeks
now
I've
been
booked
on
twice
no
reply,
so
yeah
and
basically
the
main
question
that
we
discussed
last
time
and
probably
will
be
a
really
good
discussion
for
for
today's
meeting
is
how
to
make
progress
like
what
will
be
the
the
process
that
we
can
build
here,
at
least
within
this
sick,
or
maybe
we
can,
you
know,
borrow
some
another
prizes
that
other
people
have
here,
just
just
to
move
things
down
like
make
mistakes
done,
and
you
know
at
least
have
some
progress,
because
currently
I'm
just
stuck.
A
I
have
just
one
unresolved
discussion
and
there
is
no
reply
for
three
weeks
and
I
actually
yeah
I'm
not
really
sure
what
exactly
I
should
do
to
make
all
the
things
completed
and
the
merging
request.
Since
I
already
have
like
five
five
approvals
so
yeah
that
there
will
be
something
that
we
probably
need
to
solve
as
soon
as
possible.
A
So
also
have
some
had
some
internal
discussions
within
microsoft,
and
here,
like
I
heard
the
really
similar
experience,
and
this
actually
affects
you
know
our
plans
and
the
the
way
how
we
can
participate
a
lot
just
because
you
know
we
also
have
some
plans
that
we
want
to
align
with
all
the
things
that
we're
doing
within
this
group.
So,
for
example,
right
now,
we
want
to
like
like
make
sure
that
all
the
changes
that
we
plan
for
semantic
convention
spec
will
be-
or
at
least
we
can.
A
We
can
do
it
within
this
quarter
and
finalize
them
within
this
quarter.
So
we
can
rely
on
this
like
a
time
timeline
when
we
plan
another
things,
but
for
now
the
the
state
is
that
we
want
to
like
a
double
check.
Will
it
be
possible
actually
to
make
things
standard
and
looks
like
we
have
higher
like
a
higher
risk
here,
and
it
actually,
you
know,
affects
all
the
other
things
that
we
are
doing.
A
So
we
need
to
postpone
some
another
items
just
because
things
are
not
really
stable
at
the
moment,
so
we
cannot
take
them
for
this
iteration
and
since,
like
you
know,
for
example,
net
comes
with
early
release
cycles.
That
means
that
all
the
things
will
be
postponed
and
postponed
like
for
several
months
and
if
no
progress
it
will
be,
it
will
be
just
another
iteration
which
happens
next
year.
B
A
That's
unfortunate
well,
a
lot
of
different
things
are
connected
interconnected,
so
we
probably
need
to
have
some.
You
know
at
least
viable
process
and
to
make
it
clear
for
everyone.
B
Yeah
yeah
I'll
give
the
the
tc
some
of
this
feedback.
It
looks
like
there's
so.
B
For
let's
look
at
the
two
issues
we
have
open.
One
is
span
structure
for
http,
retries
and
redirects.
B
A
Yeah,
that's
right.
Basically,
he
just
you
know,
charming
and
asked
this
question
and
you
know
stated
that
he
has
some
doubts
about
the
usage
of
links,
but
we
discussed
this
three
weeks
ago,
yeah
but
yeah.
Since
then
there
is
no
reply
and
you
know,
since
the
conversation
is
isn't
completed,
not
resolved
unresolved.
Basically,
it's
in
block
state
right
now.
B
Yep-
and
it
looks
like
with
the
client
errors-
it's
maybe
similar
right,
I'm
trying
to
see
who
here
was
actually
thinking.
It
wasn't
a
good
idea
to
turn
them
off
everywhere.
Yeah.
C
There
was
a
fair
few
people
yeah
that
they
were
against
it,
but
like
I,
don't
think
that
that
pr
isn't
as
fleshed
out
and
as
dennis's
one
dennis
is
one
you've
got
like
you
know,
implementations
in
python
and
c,
and
everything
like
that.
I
haven't,
kicked
the
tires
as
much
on
this
one,
but
also
we
need
them
both
to
be
stable,
right,
yeah,
yeah,
yeah
yeah.
C
No,
exactly
I
was
just
saying
I
don't
really
mind
if
it's
one
way
or
the
other
with
this
with
the
client
errors,
like
I
really
just
made
the
pr
just
so
just
to
get
discussion
going
just
to
see.
B
Yeah
I
mean
I
I'm
actually
with
you
on
that
one.
It
seems
like
the
the
the
breakdown
there
is
sort
of.
It
seems
to
me
between,
like
pragmatists,
and
maybe
I
don't
know
what
the
right
word
is.
People
who
are
are
trying
to
stick
to
a
formalism
right
like
from
a
practical
standpoint.
B
C
But
it's
really
like
not
objectively
one
or
the
other.
Is
it
it's
just
a
matter
of
opinion
and
it's
hard
to
it's
hard
to
get
a
decision
and
get
agreement?
If
that's
the
case,
I
don't
know
if
there
is
a
formal
process
to
to
resolve
that
kind
of
thing,
but
yeah.
B
Yeah
I
mean
it's,
it's
it's
tough.
When
it's
it's.
I
think
these
kinds
of
design
decisions
are
tough
to
to
find
agreement
on.
You
know
without
doing
some
kind
of
extra
work,
but
I
think
I
think,
there's
a
good
case
to
be
made
to
to
change
it.
B
Anyways,
I
I
feel
like
I
can
raise
it
with
the
the
tc
for
dennis's
pr,
because
I
agree
that
one
has
just
just
been
stymied
yeah
by
by
a
lack
of
response
from
the
tc,
so
I
can
be
like
hey.
Can
you
guys
like
respond
to
this
and
like
get
her
done
and
going
forwards,
I
would
suggest
reaching
out
directly
to
the
people
involved
on
slack
dennis.
B
Can
you
get
this
this
put
in,
but
I'll
also
try
to
send
the
feedback
that
that,
like
the
slowness
here
is
like
actually
having
like
some
real
world
repercussions
and-
and
we
need
to
be
able
to
move
faster
on
on
getting
this
stuff
standardized
and
it
does
seem
like
what
happens
is
like
we
make
proposals
and
there
will
be
an
objection
from
a
tc
member
or
questions,
but
then
that
person
doesn't
follow
up
or
on
some
of
these
cases
it's
it's
a
decision.
That's
like.
B
B
Maybe
maybe,
as
like
a
third
prong
here,
is
bringing
this
up
in
the
the
tuesday
spec
meeting.
I
don't
think
we've
done
that
yet,
but
but
that
would
be
the
other
place,
I
would
bring
it
up
and
again,
maybe
on
slack
being
like
hey,
we
have
these
open
questions
about
links
like
what
are
we
supposed
to
do
here,
but,
but
I
kind
of
agree:
it's
bad,
it's
bad
form
for
people
to
to
block
stuff
and
then
not
not
follow
up
and
resolve.
B
I
know
bogdan
in
particular,
is,
is
stretched
really
thin
at
the
moment,
but
nevertheless
you
know
like
maybe
he
could
hand
this
off
to
somebody
else
or
something
like
that
if
he
doesn't
have
time
for
for
some
of
the
things
he's
doing.
A
Right
yeah,
so
I
was
thinking
about
you
know.
Some
kind
of
you
know
agreements
that
we
can
have
here,
for
example,
in
case
there
is
some
discussion
and
there
is
no
follow-up.
We
can
just
consider
the
discussion
resolved
and
kind
of
in
two
weeks
or
in
one
month
or
some
particular
time
frame
we
can
think
about-
or
at
least
we
can.
You
know,
have
some
another
process
that,
for
example,.
A
Of
a
ticket,
so
basically
it
it
shows
you
on
github
as
an
assignee,
so
an
assignee
kind
of
required
to
chime
in
like
within
one
week
and
just
to
you
know,
keep
the
discussion
going
forward
and
if
not
just
resolve
all
the
conversation,
or
you
know
doing
something
like
that
or
you
know
reach
out
to
a
tc
member
or
any
other
person
who
actually
left
the
command.
A
So,
like
I
said
you
know
some
some
kind
of
agreements
that
can
allow
us
to
to
keep
things
moving,
but
at
the
same
time
you
know
just
not
to
kind
of
prevent
others
to
to
express
their
their
their
concerns
yeah.
So
something
something
probably
should
be
done
extra
here,
just
because
all
the
things
that
the
current
states
just
just
doesn't
allow
us
to
to
move
forward.
B
Yeah,
so
I
would,
I
would
suggest,
as
a
protocol-
and
we
should
be
quicker
about
this
going
forwards
as
well
like
on
our
part,
which
is
to
keep
doing
what
we're
doing
and
and
making
prs
the
spec
if
they
stall
out,
because
someone
doesn't
seem
to
be
responding.
You
know
within
a
couple
of
days,
then
poking
them
on
slack
whoever's.
B
The
author
of
the
pr
should
poke
that
person
on
slack
and
see
if
that
works,
because
often
it
does
and
then,
if
that
doesn't
work,
the
follow-up
is
to
have
tc
members
who
are
kind
of
like
have
this
partic,
basically
going
to
another
tc
member,
but
I
would
suggest
in
general
armin
probably
because
he
is
paying
attention
to
the
the
semantic
conventions
in
general
and
carlos
is
another
friendly
tc
member
who
cares
about
this
stuff.
B
So,
and
also
tigran,
though
the
tigran's
a
little
busy,
so
we
could
try
reach
out
directly
to
the
tc
member
who's,
who's,
blocking
it
or
other
member,
and
if
that's
not
working
quickly,
asking
one
of
these
other
tc
members
to
be
like
hey,
can
you
can
you
resolve
this?
For
us?
Can
you
can
you
take
this
over
we're,
not
getting
a
response
from
this
other
tc
member
and
I'll?
B
Let
the
tc
know
that,
like
this
is
kind
of
what
we
want
to
do
and
that
we
kind
of
expect
in
general
to
have
like
a
quick
response
on
this.
But
you
know
if
one
tc
member
can't
just
basically
saying
you
know
either
really
care
about
it
and
pay
attention
or
let
it
go
and
pass
it
off
to
another
tc
member
to
make
a
decision
about
it
and
just
you
know,
be
satisfied.
You're
gonna
have
to
live
with
the
decision,
because
you're
not
you're,
you're
too
busy
to
pay
attention
to
it.
A
Yeah
something
like
that
that
probably
will
help
us
yeah
just
about
this
particular
item.
I
also
try
to
reach
out
to
tigran,
since
he
is
assignee
of
this
particular
pro
request
and
basically
their
response
was,
you
know
you
guys
need
to
figure
out
by
yourself
like
okay,
so,
basically
that
that's
that's
what's
yeah!
Definitely,
if
you
have
the
same
rules,
kind
of
we
can
follow
and
we
have
clear
understanding
how
it
happens.
A
Probably
will
help
all
of
us
just
to
you
know,
keep
them
smiling.
B
Yeah
yeah,
I
agree
for
for
issues
like
yours,
james,
where
it's
just
just
kind
of
a
mess.
I
think
maybe
bringing
it'd
be
hard
for
you
to
attend
the
tuesday
spec
meeting.
That's.
B
But
I
can,
I
can
put
it
on
the
agenda
for
the
next
spec
meeting
and
try
to
champion
it
and
be
like
look
we
practical
standpoint.
We
really
want
to
change
this
like
like.
Can
we
get
some
movement
here
and
and
if
people
are
really
like
have?
Basically
all
try
to
ask
me
like
do?
People
have
a
strong
practical
reason
for
not
changing
this
forget
about
the
formalism,
because
that's
not
actually
a
good
way
to
look
at
this.
B
We're
not
even
consistent
about
that
anyways
because
we
set
500s
on
the
client
and
it
would
be
silly
really
silly
not
to,
but
you
know,
do
people
have
like
a
practical
reason
why
they
want
this
stuff
set
by
default
like
it
actually
is
like.
Yes,
this
is
very
helpful
to
have
this
set
by
default,
and
I
want
to
suppress
them
not
not
enable
them
so
I'll.
B
C
It
it
doesn't
even
need
to
be
that
it
just
needs
to
be
like,
even
if
there's
some
kind
of
like
process
where
you
can
like
just
vote
like
one
way
or
the
other,
should
we
change
it?
Should
we
not
change
it
like
now?
We
can
just
just
objectively
make
a
decision,
because
I
mean
oh
yeah,
maybe
it
matters
more
for
other
people.
For
me,
if
the
span
has
an
error
status
or
not,
I
don't
really
care
what
I
do
care
and
what
atlassian
cares
about
is
the
semantic
invention
being
stable.
B
Yeah
and
it
could
be
the
the
response
could
be
like
look,
it's
you
know
it.
It
would
be
really
disruptive
to
change
this.
At
this
point,.
B
Yeah,
like
you
know,
are
we
we
haven't
declared
any
of
this
stuff
stable
yet,
but
nevertheless,
we
try
to
avoid
making
breaking
changes
along
the
way
that
would
really
screw
up
people's
dashboards
or
something
yeah,
and
I
could
see
this
maybe
being
a
very
surprising
change
for
people,
but
it
could
be
a
very
pleasant
change
for
people
because
I
suspect
most
of
these
errors
are
noise.
B
B
C
We
I'm
glad
we
at
least
can
not
told
you.
B
The
number
one
thing
is,
you
know,
there's
a
whole
suite
of
tc
members,
and
so,
if
you
can't
get
a
response
from
one
you
know,
and
if
you
talk
to
another
one
and
they're
like
give
you
something
unsatisfying,
like
tigrin's
answer
of
saying,
like
no
there's
no
choice,
you
just
have
to
get
bogden
to
pay
attention
to
this,
then
you
know,
I
would
say,
going
to
to
armin
or
another
tc
member
and
being
like
hey
like
this
sucks
like
like
we're
being
told
like
we
have
to
talk
to
this
guy,
and
this
guy
isn't
responding
like
what
the
hell.
B
I
think
I
think
it's.
I
think
we
should
be
like
yelling
at
people,
but
I
do
think
it's.
Okay
to
you
know
demand
that
that
the
tc
not
create
a
untangleable
knot.
B
You
know
the
only
answer
is
to
talk
to
the
guy
who
won't
respond.
You
know
like
like
they.
I
they
have
to
be
willing
to
to
sort
that
out.
You
know
so
I'll
give
them
that
feedback
and,
let's
just
try
on
our
end,
to
be
just
more
quick
and
aggressive
about
about
pinging
people
and
kind
of
raising
it
up
the
chain
when
we're
not
getting
a
response.
So
we're
not
waiting
a
month
and
then
trying
to
figure
out
what
to
do.
C
B
Of
I
mean
well
actually
in
this
case,
it's
like
the
problem
is
like
we.
We
have
like
a
a
tiny
linus
in
the
form
of
of
bogdan
right.
Yeah
part
of
the
reason
why
tiegen
or
someone
else
I'm
sure
doesn't
want
to
step
in
is
they're
like
oh,
it's
about
like
what
links
should
be
used
for.
Oh
man
that
is
like
that
is
such
a
like
bogdan
invention.
Like
you
know
you,
you
really
do
kind
of
need
his
his
feed
like
feedback
there.
B
You
know,
so
I
suspect,
there's
like
a
bit
of
that
going
on
with
that
particular
issue
right,
but
that's
still
not
acceptable.
Right,
like
the
whole
point
of
having
like
a
committee
of
tc
members
is
so
that
we
have
like
a
higher
bus
factor
right
like
I
don't
think
it's
I
I
think
I
do
think
it's
it's
like
the
model
we're
going
for
is
like
well,
if
one
of
you
can't
do
it
the
other
one,
somebody
else
should
be
able
to
pick
up
this
ball
like
like.
B
It
can't
be
that
that
that
just
picks
it
up,
which
is
the
thing
I
hate
about
the
the
bdfl
model.
It's
that
person
that
this
situation
always
happens
in
those
projects,
basically
there's
like
some
set
of
stuff.
That
gets
a
lot
of
attention
and
then
there's
some
issues
that
just
just
sit
there
in
the
back.
You
know
I
don't
like
going
into
a
project
and
being
like.
B
B
The
pile
isn't
it,
and
so
I
do
think
we
we
really
don't
want
open
telemetry
to
work
like
that.
I
do
have
some
hope
that
when
you
know
metrics
and
logs
are
kind
of
done,
it'll
be
a
little
easier,
but
but
I
do
think
this
is
like
been
kind
of
a
perpetual
problem
with
the
project
of
stuff
getting
jammed
up,
because
a
particular
tc
members
become
unresponsive,
so
the
metrics
is
getting
there.
Isn't
it
yeah
yeah,
it's
it's
really
getting
there.
B
It's
in
the
like
the
most
of
it
is
feature
freeze
and
being
implemented
in
a
number
of
languages,
so
we've
got
betas
for
everything
that
isn't
like
a
core
metrics
feature
so
like
all
the
the
fancy
stuff
around
integrating
it
with
trace,
exemplars
and
stuff,
like
that,
there's
still,
I
think
some
some
design
room
on
that
going
on
like
there's
still
like
that
stuff
hasn't
been
like
set
in
stone
yet,
but
but
everything
that
you
think
of
as
metrics
is
now
done
as
far
as
design
is
concerned,.
A
B
It's
just
a
matter
of
getting
the
betas
written
and
getting
getting
release
candidates
out
there.
B
But
again
I
actually
need
to
want
to
follow
up
with
that,
because
you
know
there's
like
proactive
maintainers
in
some
languages
who
are
like
you
know,
building
you
know
building
betas,
but
it's
not
totally
visible
to
me
like
like
how
do
we
have
a
deadline
for
when
some
of
these
are
going
to
be
ready?
You
know
you
know
it
does
feel
like
we
need
a
little
more.
I
would
like
to
see
something
we
can
say
publicly
about
this
stuff
and
around
when
it's
going
to
be
ready
in
each
of
the
languages.
B
A
Cool,
oh
yeah.
Actually
we
have
a
couple
of
more
items
in
this
old
tab
that
was
merged
recently,
oh
yeah
yeah.
So
if
you
don't
mind
I
can
share
so
we
can
maybe
start
the
discussion
today.
Actually
I
believe
we
discussed
this
last
time,
but
let's
do
it
once
again,
so
it
will
be
clear
for
erin.
A
So
here
in
this
old
tab
for
for
for
b1,
we
do
have
several
items.
Actually
four
items
so
error
status
is,
is
something
that
we
already
have
to
request
open,
retries
and
redirects
the
same,
but
we
do
have
a
couple
of
more
like
a
relate
required,
attribute,
sets
and
context
propagation.
So
for
this
one
we
had
a
short
discussion
last
last
week-
and
basically
here
we
have
like
this
issue
opened
by
the
miller.
A
I
was
not
able
to
follow
up
with
her
about
this
during
during
last
week.
I
will
do
this
this
week,
but
overall
idea
is
basically
at
least
define
or
actually
clarify
which
exact
attributes
are
required
on
a
client
side
on
a
server
side,
so
probably
something
that
we
can
do
as
a
first
step
here,
which
can
you
know.
A
Help
us
a
lot
is
to
do
the
following.
So
let
me
open
the
http
semantic
conventions
document,
and
here
we
have
the
following
statements.
A
So
we
do
have
all
these
attributes
and
required
is
just
http
method
right,
but
when
it
comes
to
like
sections
that
we
have
following
so
like
http,
client
and
http
server,
so
it
says
that
for
http
client
like
this,
this
statement
here
like
additional
tribute
requirements,
and
it
says
that
at
least
one
of
the
following
sets
of
attributes
is
required
and,
to
be
honest,
actually
none
of
these
exist
on
http
client
site
apart
from
http
dot
url.
A
So
there
is
no
schema
calls
target
whatever
what
not
by
default
right.
It's
just
a
string
url.
So
definitely
it's
possible.
You
know
to
parse
the
string
and
get
all
these
scheme
calls
and
target
and
whatnot,
but
only
url
exists
on
the
client
side,
and
this
we
have
the
really
similar
stuff
on
the
server
side,
saying
that,
like
at
least
one
of
the
followers
that
really
says
is
required-
and
you
know
this-
the
same
attributes
are
here,
but
the
order
is
different.
A
So,
on
a
server
side,
there
is
no
url
schemas
here.
Is
there
like
a
host
and
target?
Sometimes
host
is
not
available,
so
it
can
be
something
like
I
don't
know
server
name,
and
maybe
it's
like
a
port
or
host
like
a
hosted
port
horse,
name
and
port.
A
So
all
this
stuff
is
something
that's
probably.
We
need
to
clarify
further
just
to
make
it
super
clear
that
okay,
so
on
a
client
side,
we
do
have
this
as
required,
and
maybe
this
table
can
be
also
adjusted
saying
that
we
have
following
attributes
is
common,
and
this
is
the
kind
of
additional
attributes
that
we
expect
to
be
to
be
added
on
the
client
side.
So
when
spanking
is
clients
and
this,
but
this
should
be
kind
of
marked,
as
required
there
and
on
the
server
side.
A
A
All
this
stuff
actually
is
provided
here
in
this
issue
opened
by
led
miller,
but
it
like
there's
something
that
we
probably
so
ludmilla
provided
a
lot
of
different
information
here,
but
at
least
you
know
just
by
distinguish
by
this
thing,
like
a
pro
basically
splitting
these
things
for
client
and
a
server
will
help
us
a
lot
at
least
to
clarify
this
part
of
the
specification,
because,
basically
it's
it's
really.
A
You
know
it's
really
a
copy
paste,
but
with
some
differences
on
the
server,
so
yeah,
that's
the
first
thing
and
another
thing
we
also
have
here
in
a
roadmap.
Is
this
context
propagation
topic-
and
this
is
really
you
know
it
can
be
also
really
wide
discussion,
but
the
overall
idea,
actually
that's
something
that
will.
I
actually
also
will
be
interested
to
leverage
when
I
was
working
on
retries
and
redirects,
actually
how
to
propagate
contacts
between
tries
or
it
can
be.
You
know
we
can.
A
We
can
think
about
this
more
widely
asking
this
question
like
how
we
want
to
propagate
some
context
between
different
different
spans.
Should
it
be
like
a
baggage?
Should
it
be
some
another
kind
of
storage?
I
don't
know
like
a
some
some
ambient
context
or
whatnot,
so
it
can
be
useful
for
tries.
It
can
be
useful
for
like
a
designing
sessions.
A
It
can
be
useful
for
grouping
like
a
you
know,
to
do
some
kind
of
grouping
of
spans
that
correspond
to
some
logical
operation
or
logical
transaction
happen
happening
on
the
on
the
on
some
part
of
the
application.
So
this
this
is
a
really
wide
topic
and
there
is
no
clear
understanding
exactly
what
to
do
with
this,
but
at
least
we
can
start
with
it
right.
A
So
we
can,
you
know,
have
some
some
direction
here,
or
maybe
we
can
decide
to
to
move
it
out
or
like
I
just
remove
it
from
from
the
from
the
roadmap
for
v1,
but
there's
something
that
probably
needs
to
be
addressed
as
well.
So
yeah
we
have
two
more
items
actually
to
clarify
because
before
we
can
call
it
stable,
so
at
least
we
can.
A
You
know
so
the
the
the
really
simple
thing
that
we
can
do
is
just
to
move
it
to
the
vnx
section
and
that's
it,
but
at
least
there's
some
something.
That's
we
probably
need
to
take
care
of.
B
Yeah,
I
actually
thought
with
context
propagation
that
the
the
scope
of
that
was
at
least
what
the
scope
of
it
we
were
intending
to
tackle
was
tackled
as
part
of
the
retries
and
redirects
issue
right
around.
How
do
you
you
know?
Should
it
be
a
different
span
id
for
every
retry,
and
the
answer
is
like
yes,
if
possible,
though
there's
some
cases
where
it's
not
and
then
you
know
well,
I
guess
it's
not
quite
resolved
yet,
but
using
links
as
a
way
of
connecting
these,
these
retries
and
redirects
together.
B
My
impression
was
like
that
was
that
was
the
kind
of
context
propagation
we
wanted
to
to
deal
with
here.
I
don't
know
if,
but
if.
A
Yeah,
so
actually
my
my
initial
idea
was
exactly
the
same,
but
when
I
was
you
know,
working
on
retries
and
redirects
from
different
perspective
and,
like
you
know,
on
different
technology
stacks,
it
happened
that
it's
not
really
obvious,
which
exact
storage
to
be
to
use.
You
know
to
to
propagate
this
information,
so
sometimes
you
know
we
can
reuse
the
same,
like
http
request,
object
instance
and
you
know,
use
it
as
a
as
a
storage,
so
just
to
stash
some
information
there,
so
it
can
be
used.
A
We
can
take
it
from
there
on
the
secondary.
Try
and
this
can.
This
will
allow
us
to
you
know
propagate
some
in
some
information
but
yeah.
Sometimes
that
just
doesn't
work.
For
example,
if
you
cannot
reuse
the
same
exact
instance
of
http
request
object,
then
I
have
a
problem,
so
you
know
in,
for
example,
in.net.
A
There
is
a
concept
that
you
can
use
to
actually
kind
of
construct
this
ambient
context,
but
this
basically
leads
to
some
another
allocation
of
memory
and
definitely
something
that
we
want
to
avoid
from
the
efficiency
perspective
and
performance
perspective.
So
maybe
it
will
be
useful.
You
know
to
have
to
bring
or
to
provide
some
some
kind
of
general
recommendations
here.
A
So,
for
example,
if
you
have
this
case
here,
we
have
something
that,
like
you
know,
open
cylinder
baggage,
something
that
we
can
use
here,
and
this
will
be
the
the
main
choice
when
you
do
when
you
need
to
do
some
some
sort
of
these
things.
Alternatively,
it
can
be
something
else.
I
just
have
no
clear
understanding
at
the
moment.
What
exactly
we
can?
You
know
recommend
here
from
the
open,
telemetry
perspective,
so
but
yeah,
that's
something
that
probably
will
be
needed
to
to.
A
You
know
to
get
more
deeper
investigation.
B
Yeah,
yep,
okay,
yeah!
That's
that's
fair!
I'd
love
to
to
to
hear
your
proposals
on
that
stuff.
It
sounds
like
there's
it
certainly
at
least
a
chunk
of
what
you're
talking
about
that's
just
like
practical
stuff
that
comes
up
when
when
people
try
to
go
out
and
instrument
all
these
libraries
and
that
definitely
seems
like
v1
stuff
right,
any
kind
of
questions
that
are
coming
up
when
we
try
to
do
this
instrumentation
of
like
how?
B
How
am
I
supposed
to
do
this?
We
should
resolve,
and
maybe,
if
there's
things
that
are
a
little
more
like
just
outside
the
realm
of
open
telemetry's,
currently
functionality
we
can,
we
could.
B
I
would
suggest
we
punt
on
those
so
like
the
concept
of
like
a
user
session
is
something
that
seems
useful
for
open
telemetry
to
have,
but
we
we
don't
have
it
yet.
So
I
wouldn't
want
to
hold
up
getting
the
stuff
marked
as
stable
to
sort
out
what
user
sessions
should
look
like
in
open,
telemetry.
A
Right
yeah,
exactly
so
like
you
know,
at
least
we
can
start
from
some
some
best
practices,
not
inventing
something
else,
but
just
reusing
existing
concepts.
But
if
not,
then
we
probably
need
to
do
something
with
that,
but
yeah.
So
this
is
not
really
clear
and
definitely
requires
some
efforts
to
investigate.
But
this
one
is
pretty
much
obvious
and
we
do
have
a
lot
of
information
provided
by
ludmila
here.
Okay,.
B
C
A
So
if
people
have
some,
you
know
some
time
and
can
you
know
pick
some
items
from
here
that
that's
definitely
a
good
candidate
here.
B
Yeah
well,
it
sounds
like
ludmilla's
already
started,
started
an
issue
to
to
get
this
this
stuff
resolved
and
by
the
way
for
what's
actually
in
the
spec.
I
do
think
it's
a
little
more
like
what
you're
asking
for
like
it's.
B
Those
bullet
points,
I
think,
should
maybe
be
like
numbered
bullet
points
rather
than
like,
like
I
think,
there's
an
intentional
there's,
an
ordering
that
was
intended
here,
which
was
that
for
clients,
http
url
should
be
the
thing
you
put
there
and
for
servers,
scheme
host
and
target
should
be
the
thing
you
put
there,
but
then
there
was
this
like
long
trail
like
well,
there's
like
sometimes
you
don't
have
that
like
easily
accessible
or
you
have
to
concatenate
or
split
some
stuff
to
get
it
into
that
form
and
that's
expensive,
and
so
that's
where,
like
some
of
these
alternates
came
from
of
being
like
well,
if
you
don't,
if
like
it's
hard
to
give
fork
over
the
url
for
some
reason
on
the
client
then
hand
it
over
like
this,
I
don't
know
if
that's
right
or
wrong
like
maybe
we
want
to
be
more
firm
about
saying
like
no.
B
You
should
just
do
the
concatenation
here,
but
I
think
that
was
that's
why
the
the
orders
on
these
things
are
are
reversed
right.
A
Right
so
yeah,
maybe
like
the
way.
The
way
I
see
it,
it
might
be.
You
know
really
useful
if
you
can
just
have
this
as
a
like
a
require
required
attributes
on
a
server
and
we
can
put
them
to
the
table.
That's
the
same
way.
We
have
it
here,
yes,
and
we
can
say
okay,
so
this
is
required.
This
is
required
and
this
is
required,
but
we
can
also
have
something
like
you
know
this
index
here
saying
that
and
we
can.
B
A
We
need
to
have.
We
need
to
have
it
for
a
host
only
because
him
scheme
and
targets
are
always
there,
but
first
we
can
say
okay,
so
if
host
is
not
there,
then
you
need
to
use
this.
This.
B
A
This
one
is
not
there,
you
need
to
use
this,
so
basically
it
will.
It
will
clarify
the
things.
A
lot,
and
you
know
lots
of
different
attribute
sets
that
some
of
them
are
required,
but
you
know
there
is
no
like
a
clear
decision
how
it
should
you
should
like.
I
select
one
of
them.
So,
for
example,
here
you
there
is
not
really
clear
for
from
this
text.
You
can
decide
that
okay,
so
I
will
go
with
this
one.
A
It
is
there
and
it's
like
a
claim,
as
as
attributes
says,
is
required
and
we
have
it
here.
So
I
will
just
go
with
this
one.
B
B
Yeah
yeah,
it
would
be
good
to
know
I.
I
totally
agree
that
I
think
for
the
most
part,
the
spec
is
trying
to
say
what
you
want
it
to
say.
It's
just
not
saying
it
very
well
and
what
you're
proposing
sounds
better.
It
would
be
good
to
know
also
if
we,
if
there's
anything
in
here,
that
lou
mill
is
proposing.
That's
like
actually
a
breaking
change
or
a
change
of
some
kind
rather
than
a
clarification.
B
C
B
Yeah
I
mean
there
is
this
like
switching
issue
that
I
feel
kind
of
comes
up,
but
I
I
got
some
feedback
from
people
that
maybe
it's
okay,
but
I
do
sometimes
like,
like
we
don't
we
don't
put
in
any
s
type
fields.
We
tend
to
try
to
avoid
type
fields
or
fields
explicitly
for
for
switching
with
the
idea
that
you
should
kind
of
sniff
and
detect
the
presence
of
something.
So
like
is
this
an
http
span.
Client
span.
B
Well,
if,
if
it
smells
like
one,
then
it
is
one
right
like
if
there's
like
http
something
in
there
like
you
know,
then
it's
then
it's
an
http
span
and
that's
helpful
because
then
you
know
we
can
like
do
composition,
but
I
do
wonder
whether,
if
in
some
of
these
cases,
where
we're
creating
inefficiencies
down
the
pipe
like
like
it
would
be
much
nicer
to
say
like
if
it's
a
client
span,
http
url
has
to
be
there.
It
just
has
to
be
there.
B
If
you
got
to
concatenate
stuff,
to
put
it
there
too
bad,
that's
you
just
gotta,
do
it
and
then
that
would
be
like
like
a
much
more
reliable
field.
You
know
what
I
mean
like
I
don't
know
if
it's
true
or
not,
but
I
wonder
if
like
by
being
a
little
loose
here,
is
that
actually
creating
trouble
elsewhere,
yeah.
C
C
B
Then
yeah
yeah,
I
haven't
done
like
an
in-depth
investigation
here,
but
it's
just
a
general
design
smell
that
I
wonder
about.
Are
we
creating
when,
when
someone
wants
to
do
data
science
on
open
telemetry,
do
they
end
up
with
this?
Like
big
nest
of
you
know,
if
else,
clauses
and
and
all
of
this
stuff
in
order,
because.
C
There
is,
there
is
metric.
This
span
was
attribute
id.
Isn't
that
there's
a
just
like
http.client.com.
C
B
Yeah
so
so
we
group,
we
group
the
semantic
conventions
into
name
spaces
right,
and
so
it's
sort
of
like
if
you've
got
http
dot,
something
in
the
attributes.
Then
this
is
an
http
span
right
if
you've
got
db,
that's
something
that
it's
a
db
span
and
it
might
have
both
that's
why
we
didn't
go
with
like
a
span
type
because
maybe
it's
like
and
it's
like
literally
both
db
span
and
an
http
span.
If
it's
an
http.
B
Yeah
yeah
exactly
you
know,
it's
client,
you
know
it's
got
acp
stuff
on
there
and
that
compositional
approach
seemed
to
avoid
that's,
I
think
that's
fine
yeah
yeah,
I
I
I
totally
agree.
I
was
fourth
I
pushed
for
this.
In
the
past
I
was
like
one
of
the
people
I
batted
batting
aside
the
idea
of
like
a
span
type
that
some
people
are
asking
for
yeah.
But
since
then
I've
been
like
this
is
a
good
idea,
but
are
there
places
here
where
we're
we're
being
too
loose?
So
we're
not
supplying
enough
information?
B
And
I
don't
I
don't
know
the
answer
to
that
question
exactly,
but
this
is
like
in
a
situation
where
I'm
like
it
would
be
really
nice,
if,
like
you,
could
just
rely
on
url
being
there
like
you
know,
but
the
reality
is,
maybe
you
can't
so
yeah.
I
don't
know.
A
So
you
have
this
like
if
set
url
must
so
basically,
like
you
know
previously,
we
also
had
a
discussion,
so
must
is
not
something
that's
you
cannot
use
word
must
if
it
cannot
be
said.
So
it's
already
controversial
thing,
and
you
know
it's
also
like
a
about
the
required.
A
So
we
do
have
this
this
table
and
you
know
we
have
this
fields
or
attributes
and
some
of
them
are
required,
but
at
the
same
time
it's
true
http
url
is
not
required,
but
later
on
we
are
saying
this
is
this,
is
one
is
required?
So
you
know
there's
a
lot
of
different
statements
about
the
the
same
stuff,
but
the
statements
are
different
and
so
at
least
this.
B
B
Yeah,
maybe
we
could
clean
this
client
one
up.
Actually
the
server
stuff
is
a
little.
I
think
there
are
some
genuine
edge
cases
where
it's
like.
Okay,
you
may
not
actually
have
this
piece
of
information,
and
so
you
have
to
use
this
other
thing,
but
on
the
client
side,
all
of
these
formulations
could
be
concatenated
into
a
url
correct.
So
you
know
we
could
go
the
route
of
saying
like
if
it's
on
the
client.
You
just
have
to
put
this
in.
If
you
have
to
concatenate
some
stuff
to
do
that
too
bad
do
it.
B
I
don't
not
aware
of
an
edge
case
where
it's
like
literally
the
instrumentation
can't
hook
in
in
a
place
where
it
knows
all
this
information
all
at
once,
so
it
just
can't
or
something
like.
I
don't
know,
yeah
that.
A
Actually,
on
a
client
side,
you
know
the
the
way
how
http
client
work
is
already
like
a
required
thing
for
you
to
to
have
a
string
to
concatenate
it
to
be
able
to
at
least
use
http
client,
and
it's
not
really
depend
on
the
implementation.
We
have
it
for
java.
We
have
it
for
the
net
for
python,
so
all
the
stuff
you
need
is
is
is
to
provide
a
string
to
it.
B
Yeah-
and
I
don't
know
whether
or
not
some
of
these
things
popped
in
because,
like
okay
literally
like
the
go
http
client,
it's
like
actually
annoying
to
to
get
it
in
url
form
or
something.
But
I
don't.
I
don't
actually
think
that's
the
case.
I
think,
if
I
recall
it
was
more
about
people
trying
to
avoid
concatenation
costs,
but
it
could
just
be
like
a
super
edge
case.
That's
like
not
an
issue.
A
Http
url
yeah,
so
for
this
one
we
can
just
have
http
url
and
that's
it
so
basically
remove
all
this
stuff
from
from
the
specification
here
yeah
since
it's
already
like,
I
said
that
it
must
and
it's
required,
so
we
can
just
put
it
as
is,
and
for
server
side
since
we
already
like.
I
know
that's
on
the
server
side,
there
is
no
url
as
a
string,
but
we
have
these
different
components,
because,
basically
that's
how
http
protocol
works
right.
A
So
on
a
server
side,
you
have
just
all
the
things
as
a
separate
things
like
you
you're
you
have
host
as
a
healer.
You
have
like
a
target.
You
have
scheme,
so
all
this
stuff
on
the
server
side,
I'm
naturally
provided
by
any
any
http
server.
Just
because
that's
the
way
how
http
protocol
works-
and
here
we
can
just
have
so
okay,
this
is
the
those
are
required
on
the
server
side
and
that's
it.
But
in
case
you
don't
have
host,
it
happens
right.
A
So
you
have
like
a
no
host
name
or
something
it
can
be
like
a
server
name,
and
it's
is
if
there
is
no
server
name,
that
it
should
be
like
a
iep
or
you
know
whatever
it
is.
We
have
all
these
things
written
here
as
well,
yeah.
So
basically,
that's
that's
mostly
about
the
clarification
and
removing
this
contribution.
A
Yeah,
so
definitely,
for
example,
I
can
start
working
on
this,
but
I
would
say
that
maybe
it
will
be
good
for
me
to
focus
on
this
stuff,
so
we
can
work
on
this
in
parallel,
but
this
one
should
be
definitely
good.
If
someone
can
pick
pick
it
up.
A
She's
willing
to
I
will
I'll,
pin
her
today
tomorrow,
so
she
she-
I
didn't,
see
her
like
on
these
meetings
for
a
while,
but
I
heard
that
she's
really
busy
with
her
and
other.
You
know
projects
so
I'll,
be
here
just
just
to
figure
out
like
what
is
the
availability
of
her
right
now
and
if
not,
I
will
let
you
know
somewhere
inside
probably
yeah.