►
From YouTube: 2020-10-20 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
This
one
may
be
maybe
shorter.
I
guess
we'll
see,
can
wait
a
few
more
minutes.
A
B
B
B
B
B
B
B
Okay,
I
guess
we
can.
We
can
probably
just
get
started
if,
if
john's
gonna
come,
we
can
always
catch
him
up,
but
he
was
a
little
confused,
so
he
might
not.
B
He
might
not
make
it
on
time.
Yeah.
I
think
it's
far,
let's
see
so
the
things
that
I
had
for
the
agenda.
Do
you
see
the
agenda
document.
A
B
Yeah,
so
I
had
sort
of
those
three
items
were
there
any
that
you
wanted
to
add
to
this
list
for
this
meeting.
A
Not
really,
I
think,
go
as
the
other
sij
usually
goes
through
the
issue
or
the
prs
that
can't
be
the
same.
We
want
to
do
it,
but
I
don't
really
have
a
specific
item
other
than
that
I've
been
mostly
working
on
some
of
the
issues
within
today.
B
Cool
oh
looks
like
he's
stuck
out
there.
B
B
B
All
right
made,
it
hey
guys,
sorry,
hey
hi.
Now,
where
is
yeah,
it's
always
a
always
a
little
janky
to
try
to
organize
on
these.
These
types
of
things
like
getter
is
not
that
great
and
one
big
google
calendar
is
not
the
best
way
of
you
can't
really
subscribe
to
individual
events.
I
don't
think
I
don't
know
just
the
whole.
The
whole
setup's
a
little
odd
still
from
the
hotel,
org
side.
B
Okay,
so
I
put
I
put
those
three
items
in
there
for
today's
meeting.
Were
there
other
things
that
you
wanted
to
add
to
that
or
sort
of
questions
that
you
had
that
we
should
put
on
here.
B
Okay,
cool
yeah
could
be
could
be
a
quick
meeting,
but
that's
always
always
nice,
let's
see
so
the
the
issue
that
was
opened
there
or
the.
I
guess
the
spec
change
is
sort
of
a
is
about
the
api
in
the
absence
of
the
sdk,
probably
around
what
the
global
default
will
be
as
well.
B
I
guess
it's
sort
of
or
tangentially
related
to
the
global
default,
but
the
bigger
question
is:
if,
if
someone
isn't
using
the
sdk
or
theoretically,
if
someone's
using
a
different
sdk,
you
know
technically,
someone
could
implement
their
own
version
of
the
open,
telemetry
sdk
and
still
use
the
api.
B
B
I
think
it's
just
the
propagator
that
currently
is
in
the
sdk,
the
chrome,
closing
apis.
A
Should
be
fun,
I
mean
I
originally.
I
think
that
the
the
whole
baggage
should
be
independent
from
the
any
propagate
or
any
kind
of
format.
B
B
But
given
things
like
this,
it
seems
like
the
division
of
api
and
sdk
is
less
around
an
implementation
of
traits
and
more
around
a
division
of
sort
of
implement,
sdk,
specific
implementation
details,
and
it
kind
of
sounds
like
they're
saying
the
baggage.
Propagator
is
part
of
the
api
and
not
part
of
the
sdk.
B
Yeah
and
I
more
importantly,
it
should
still
be
possible
to
propagate
it.
Right,
like
you,
shouldn't
need
to
have
an
sdk
installed
right,
yes,
and
if,
if
that
is
the
case,
then
it
sort
of
requires
that
this
thing
exist
in
the
api
somewhere.
A
C
B
Like
that,
but
they
don't
they
don't
ever
mention
them
in
sdk,
which
it
seems
very
vague
about
that,
like
it
seems
like
the
spec,
is
sort
of
written
for
for
optimizing
for
a
single
file
rather
than
because
it
seems
like
a
lot
of
these.
Things
really
should
be
split
into
baggage,
api
and
baggage
sdk,
and
really
that's
true
with
all
of
them
like.
Basically,
every
one
of
these
files
should
be
split
out
into
api
and
sdk
and
they're,
not
which
just
makes
some
of
this
confusing.
So
one
option
would
be.
B
B
B
A
Yeah,
I
think
it
would
be
nice
to
have
a
issue
around
it
to
clear
it
clarify.
I
mean
if
other
sig
may
have
the
same
problem.
B
B
Yeah
I
mean,
on
the
one
hand
it
it's
nice
to
leave
certain
things
open
to
language
level.
Implementation
details
I
mean
one
option
is
just
to
ignore
this
part
of
the
spec
and
just
not
follow
that,
but
it
seems
like
at
least
opening
an
issue
may
be
the.
B
C
At
the
moment
it
sounds
like
it
should
live
in
the
api,
including
the
propagation,
if
we
just
follow
how
they
yeah.
How
is
that
out
at
the
moment.
C
Even
the
even
the
other
propagators
might
like
the
or
the
other
yeah
I
mentioned
in
the
context
talks
about
having
them
in
the
api.
If
I.
B
B
I
don't
know
it's
a
little
confusing
okay.
Well,
I
guess
I
can
do
that.
B
One
okay,
so
then
the
second
one
is
sort
of
related
in
do
we
do
we
need
a
top
level
like?
Basically,
our
file
layout
can
still
be
what
it
is
currently,
but
do
we
basically
want
to
pub
use
at
the
top
level?
B
C
B
Exactly
because
yeah,
and
really
even
there,
especially
with
the
new
with
the
the
sort
of
higher
level
pipeline
installers,
there
really
isn't
even
any
interaction
with
the
sdk
module
there,
maybe
for
things
like
config,
if
you're,
if
you're
customizing
things,
but
if
you're
using
something
default,
you
may
never
even
reference
it.
C
It's
true
it's
interesting
because
then
the
propagation
is
part
of
the
sdk
right
now
for
us.
So
maybe
then
it
should
live
in
the
api,
but
anyway
yeah.
B
Well,
I
get
so
yeah,
that's
a
good
question.
I
think
it
would
really
just
be
removing
the
api,
so
we
could
like
go
literally
flat
into
the
entire
namespace,
even
trace
and
metrics.
They
just
flattened
everything
into
hotel.
I
think-
and
that
maybe
is
a
I
guess
that
would
be
sort
of
a
second
question
of
do.
We
want
to
like
fully
flatten
everything
out,
but
if
we
wanted
to
flatten
out
up
one
level
it
could
just
be
open.
Telemetry
trace
is
currently
open.
B
A
B
B
Right
yeah,
this
was
just
sort
of
an
artifact
of
the
the
standard
project
layout
that
open
telemetry
recommends.
B
Yeah,
do
the
preview
private
modules
yeah
or
not
even
yeah-
I
guess
private
modules
yeah
it
there's
some
yeah.
I
guess.
If
we're
gonna
do
this,
I
guess
goes
doing
it
at
this
stage
anyway,
so
it
we're
still-
and
they
have
far
more
usage
than
rust-
probably
does
so
in
terms
of
continuously
breaking
everybody's
api,
because
the
last
one
already
was
a
fairly
big
set
of
breaking
changes,
but
I
think
yeah
as
I
I
guess,
that's
sort
of
a
you
know
have
either
of
you
run
into
things
that
would
be
large.
B
A
A
Actually,
it's
not
just
fine,
it's
unavoidable,
because
the
specification
it
changes
a
lot
right
and
you
have
to
follow
the
specification.
So
whenever
the
specification
decides
to
change
your
name
on
something
we'll
have
to
change
it
along
with
it,
and
that
would
definitely
would
cause
breaking
changes.
B
Correct
but
the
spec
is
freezing
now
I
think
there's
some
issues
for
spec
freeze,
but
I
think
it's
basically
done.
B
No
yeah
yeah,
it's
basically
the
point.
I
think
the
what
they're
trying
to
go
for
is
ga,
but
if
there's
anything
horrible,
that's
found
they're
still
going
to
do
the
breaking
change
and
not
go
to
open
telemetry
too.
B
Yeah,
that
would
be
a
problem,
but
yeah
after
I
mean
I
imagine
that
this
ga
well,
I
guess
I
don't
know
how
long
this
period
is
going
to
last,
but
yeah
with
the
spec
freeze.
There
shouldn't
be
too
many
of
those,
but
I
think
now
is
probably
a
reasonable
time
to
be
doing
things
like
module,
yeah
or
changes
of
course.
B
C
A
B
A
C
B
C
C
Okay,
there
is
like
something
you
can
put
in
the
cargo
tunnel:
yeah.
Okay.
I
can
possibly
make
a
pr
for
that.
B
Okay,
well,
that's
a
that's
a
good
catch,
because
we
definitely
definitely
want
that.
B
Noted
cool
yeah,
I
guess
the
last
one,
do
we
well.
I
guess
I
had
another
thing
that
just
popped
up.
B
B
Okay,
do
we
want
to
do
an
official
sort
of
minimum
supported
rest
version?
This
seems
somewhat
common
for
the
more
core
infrastructure
things
to
have
some
guarantees.
Do
we
want
to
do
this
now?
Do
we
want
to
do
this
at
a
later
time?
I
mean
there
would
be
some
changes.
I
think
I
think
some
of
the
things
are
basically
147..
C
B
That's
a
good
question
so
well,
I
was
looking
at
let's
see.
B
A
B
They
support
all
the
way
back
to
142.
our
sdk,
the
yeah,
the
d
so
open,
telemetry
itself,
the
crate
supports
142,
but
not
all
the
exporters
do
that
could
be
a
default.
We
add
just
to
sort
of
whatever
be
at
parity
with
this
ecosystem.
We
don't
necessarily
have
to
do
that,
but
it
would
be
helpful
for
people
in
the
tracing
ecosystem.
B
Okay,
I
think
it'll
just
mean
some
changes
to
the
way.
I
think
it's
just
something
in
jaeger
and
something
in
the
otlp
exporters.
I
don't
think
it's
a
huge
amount
of
work.
I
was
just
poking
around
trying
to
use
one
of
the
old
compiler
versions
and
see
what
would
break
so.
My
guess
is
that
it's
not
a
huge
amount,
but.
C
B
Yeah
we
can.
I
can
open
this
as
an
issue
and
we
can
look
in
to
see
how
much
work
that
would
be
if
it's
really
just
a
few.
If
it's
just
a
few
lines
of
code,
then
that
seems
fine
yeah.
If
I
think
there
was
something
with
grp
cio,
where
they
don't
support,
so
there's
some
question
about
how
far
back
they
do.
This
could
also
be
something
that
is
a
per
crate
level.
So
maybe
we
have
the
otlp
exporter
have
a
different
minimum
supported
version,
and
that's
that
could
be
fine
too.
B
B
Stop
on
that
too
I'll
open
an
issue
for
this
and
start
investigating
it,
and
if
it's,
if
it's
a
reasonable
amount
of
work,
we
can
just
do
that.
B
Okay,
the
next
one
is
yeah
you're.
Looking
for
metric
stacks,
the
the
last
one
that
was
that
just
popped
into
my
head
was
there's
in
the
global
package
we
use,
we
use
arc,
read
write
lock
for
things
like
getting
a
new
tracer.
B
C
B
B
C
B
Yes,
maybe
this
is
a
like
a
profiling
question.
B
B
Yeah,
the
other
yeah,
the
other
question
is
sort
of
our
stance
generally
on
dependencies
and-
and
it
may
just
be
one
where
it's
sort
of
a
call
per
dependency,
but
some
crates
draw
a
harder
line
on
external
solutions.
Like
the
nice
thing
about
arc
read,
write
lock.
Is
that
it's
available
in
standard,
so
there's
no
there's
no
additional
dependency.
That's
that's
added
here.
I
would
assume
that
we're
following
at
least
the
this
crate
can't
be
one-o
if
it's
dependencies
or
aren't
1-0
or
it's
public
dependencies
or
know.
C
B
Which
doesn't
seem
like
a
good
solution
so
yeah?
I
guess
that
would
be
yeah
yeah.
I
guess
there's
not
necessarily
something
to
discuss
there.
It's
it's
more,
there's
some
profiling
that
I
think
we
should
be
doing
around
performance
there,
where
I
suspect
that
there
could
be
issues
for
some
people
that
that
is.
We
could
also
just
recommend
that
people
take
a
basically
just
get
a
tracer
once
when
the
app
boots
and
use
that
repeatedly.
C
B
B
Yourself
yeah
it
very
much.
Does
there
were
some
issues
that
were
opened
around
removing
the
name
requirements,
so
you
could
just
say,
get
a
sort
of
unnamed
tracer
for
because
there's
sort
of
two
consumers
here
as
well
right,
they're,
sort
of
middleware
developers
who
are
going
to
be
where
it
does
make
sense
to
say
this
is
you
know,
you're
using
whatever
rocket
middleware
version
three
and
then
there's
the
users
where
it
really
doesn't
make
sense
at
all.
B
But
and
and
now
they're
exported-
and
things
too,
which
just
gets
really
weird
yeah
true,
so
I
don't
know
the
whole
name.
Tracer
thing
I
was
never
a
fan
of,
but
I
think
was
in
there
for
some
vendor
specific
reason.
C
C
B
And
oh.
B
Yeah,
so
that
that's
probably
fine
the
yeah,
there's
sort
of
questions,
I
think
just
more
more
benchmarking
in
general,
perhaps
is
something
that
could
be
useful
to
do
just
to
see
where
the
kind
of
slower
performance
issues
are.
B
There's
also
at
some
point
you
can
get
into.
If
a
lot
of
these
traces
are
basically
the
exact
same
trace
done,
you
know
thousands
of
times
a
second.
If
it's
some
sort
of
high
performance
web
server,
does
it
make
sense
to
go
into
like
an
arena,
allocator
or
other
ways
of
basically
reusing
memory
and
sort
of
other
things
that
profiling
and
benchmarking
may
lead
us
toward.
B
But
yeah
I
don't.
I
don't
really
have
concrete
performance
issues
yet
before
benchmarking,
so
not
not
too
much
to
discuss
there
yeah
those
were
those
were
basically
the
things
that
I
was
thinking
about.
Yeah
are
there
other
things
we
wanted
to
go
over
today.
B
Yeah,
there's
lots
of
issues
yeah,
I
guess
looking
over,
are
all
the
issues
clear
right
now
like
or
do
we
need?
Are
there
any
that
aren't
any
issues
that
we
know
of
that
aren't
in
in
github?
Currently.
A
B
The
the
grpc
trace
bin
header,
one
yeah,
let's
see
spec
639.,
it
looks
like
it's
open.
B
B
It
can
just
be
done
manually.
Currently,
you
can,
just
you
know,
use
the
normal
headers
yourself.
A
Yeah,
I
mean
yeah
kind
of
like
that.
It's
like
the
it's
about
the
what's
the
status
of
our
tracing
component.
I
I
see
there
is
an
issue
like
272
which
related,
but
I
was
wondering
if
we
have
like
some
kind
of
detector,
to
gather
those
matrix
and
export
it
in
kind
of,
like
some
promises,
format.
B
Yeah
yeah,
I
can
update
that
one.
A
B
B
Yeah
sort
of
a
yeah-
I
guess
we
could
open
that
and
discuss
that
in
an
issue.
I
don't
know
if
other
people
I
mean
at
some
point,
I
think
it's
probably
people
that
want
to
use
statsd
or
some
other
exporters
or
some
other
metrics
libraries,
or
I
think
the
aws
and
gcp
sort
of
open
census
variants
kind
of
have
their
own
things
that
people
might
be
using.
But
I
really
don't.
I
don't
know
what
the
rust
community
is
using
right
now
outside
of
prometheus
there's.
Definitely
that
seems
like
the
most
common.
A
A
B
Yeah,
that's
a
that's
a
tough
one
that
one
or
maybe
you
could
look
around
and
see
what
other
and
presumably
other
languages
have
to
deal
with
this
too.
So.
A
B
There
might
be
others
who
can
sort
of
resolve
some
of
these
questions
as
well.
Okay
other
than
that
yeah.
I
think
all
the
issues
that
I
have
at
least
are
in
there.
So
some
of
the
other
discussions-
I
guess
we
can
just
keep-
keep
going
in
the
issues
themselves.
B
All
right
sounds
good,
we'll
yeah.
I
guess
that
that's
it
we
can
is
the
is
the
two
week
cadence
still
reasonable,
or
should
we
is?
Has
it
not
been
enough
time
where
we're
we're
still
sort
of
working
through
the
last
two
weeks
worth
of
things?
I
guess
maybe
we
could
just
do
something
I
could
just
ask
before
you
know
in
two
weeks,
so
you
just
ping
and
the
sort
of
sync
up
in
the
channel
and
see
if
it
makes
sense
to
talk
again.
A
C
Yeah
seems
fine
to
me:
I'm
not
yeah,
not
too
sure.
Otherwise,
the
meetings
are
short,
so
it
doesn't
hurt
you
there.
B
Can
gotta
have
your
gotta?
Have
your
attendance
counted
cool?
Well,
thanks.
Everyone
we'll
talk
later.