►
From YouTube: 2021-12-20 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
B
D
C
I
mean
it's
a
kind
of
understanding,
but
we
can
go
through
that
letter.
Yes,
one
style
that
is
agendas
or
no.
No,
no,
we
can
discuss
it.
Then.
D
Okay,
I
sure
we
can
discuss
that.
I
just
had
one
agenda
probably
to
talk
about
what
agenda
would
probably
if
we
can
review
the
dll
changes
which
has
been
done.
D
So
probably,
I
think
it
would
be
good
to
really
go
through
that
and
try
we
can
see
if
it
makes
sense.
I
mean
if
it's
a
cleaner
approach.
I
I
know
the
dll
support
has
its
own
issues.
D
If
you
want
to
support
it,
but
would
be
good
to
really
go
through
the
changes
and
decide
whether
should
we
should
we
include
these
changes,
or
should
we
wait
for
more
people
to
interest
from
more
more
from
the
community
and
then
decide
on
that?
I
mean
in
case
it's
a
very
straightforward
and
simple
change.
I
think
we
can
go
with
that,
but
it
looks
bit
complicated
and
it
would
be
difficult
to
manage.
I
think
you
can
wait
further
for
more
interest
from
the
community.
D
D
Yeah,
okay,
you
are
saying
in
the
includes
he
has
included
some.
Yes,
the
source
files
inside
the
header
include
location
right.
No,
no!
He
is.
D
D
D
D
D
D
Okay,
probably
I'll
I'll,
also
go
through
that
this,
this
yeah,
I
totally
agree.
I
think
this
looks
not
good
approach.
If
you're
going
to
do
this,
so
probably
let's
comment
it
and
I
think
I'll
also
put
my
comments.
Let
me
go
through
the
changes
and
probably
let
me
put
the
comments
and
to
tom
to
your
answer.
I
think
I
was
expecting
that
if
this
is
set,
then
we
should
be
using
all
everything
in
header
only
so
only
if
this
is
not
enabled,
then
the
source
should
be
used.
D
D
C
D
C
So
if
you
could,
if
you
can
scroll
up
a
bit
yeah
so
currently,
as
you
can
see
the
documentation
that
we
we
must
need
to
support.
Like
all
the
drops,
some
lost
value,
histogram
and
explicit
bucket
histogram.
D
D
Yeah
yeah,
so
it
is,
if
you're
talking
about
the
new
specification,
we
are
asking
for
the
new
specification
or
the
old
implementation.
I
mean
this
is
the
current.
This
is
a
new
specification
which
we
are
seeing
here.
Oh.
D
D
D
So
I
think
it
should
be
good
for
me
to
start
continuing
that
aggregation
api,
and
I
think
that
that
should
include
everything
that
all
the
aggregators
you
know
no.
D
D
D
Yeah
anything
else
we
want
to
include
today
or
we
can
just
go
through
the
open
issues
or
the
pr's
and
the
issues.
D
Yes,
that
should
be
atomic,
ideally
this
this
not
this
one.
I
think,
probably
that
we
can
start
with
that.
We
wanted
to
talk
about
this
before.
Do
we
before
deciding
on
anything,
so
this
one.
Yes,
this
is
shut
down
yeah.
The
only
concern
I
had
was
that
this
this
variable
this
or
this
is
this
variable-
would
be
used
in
the
actual
hot
path
in
the
actual
export
functionality,
so
whether
it
would
have
any
performance
degradation
if
we
are
ever
in
during
every
export.
D
D
D
D
Yes,
the
only
so
okay,
so
the
only
issue
I
felt
was
that
if
we
have,
if
we
call
shutdown
at
the
time,
an
export
is
happening,
so
shutdown
becomes
true
or
or
by
the
time
the
shutdown
becomes.
True.
Exporters
will
not
see
in
that
state
of
truth
and
it
goes
further
and
then
they
actually
exported
has
got
shut
down,
but
we
are
still
trying
to
export
so
so
to
protect
that
we
definitely
should
use.
This
should
be
atomic.
B
D
That
but
more
important
to
have
shutdown
and
export
called
simultaneously.
D
So
we
can
try
to
use
the
lock
free
mechanism
right
that
which
is
which
we
have
tom.
I
mean
in
our
api.
D
Yeah
this
is
based
on
the
tight
loop,
so
this
is
definitely
lock
free.
We
try
to
lock,
but
we
don't
really
wait
for
lock
to
acquire
the
lock.
B
D
So
I
think
probably
we
should
definitely
support
it.
Let's
try
to
use
this,
and
if
we
can
have
some
benchmark,
I
don't
know
whether
we
already
have
benchmark
against
any
of
the
exporters.
We
don't
have
it
any
other
test
for
the
benchmarks
to
really
compare
if
something
some,
if
we
have
some
degradation,
but
anyway,
I
think
we
can
definitely
implement
the
mutex
and
probably
protect
it
under
that,
at
least
at
least.
If
we
can
use
this
matrix,
I
think
we
should
be.
We
should
be
safe
enough
spin,
lock,
more
things
here.
D
D
D
D
Yeah
and
okay,
we
can
go
through
other
issue,
other
pr's,
probably
we
have
already
started
discussing
if
something
we
need
to
know.
D
D
Basically,
we
are
still
installing,
even
if
we
don't
use,
if
you
don't
have
any
exporters
with
the
default
open,
telemetry
cmake
build,
we
still
somehow
just
this
json
package
is
getting
installed,
so
it's
just
probably
we
can
see.
A
D
D
I
mean
why
does
it
fail?
It
will
fail
for
standard
center
library.
D
This
is,
I
think,
so
this
is
still
under
progress.
I
hope
so
it's
work
in
progress,
so
I
didn't
review
it.
D
A
D
D
D
D
D
D
This
is
good
to
merge.
I
mean
feel
free
if
we
have
to
go
through
this.
I
think
if
you
have
any
comments,
but
I
think
josh
has
already
approved
it
since
I
think
it
should
be
good.
No
major
changes
at
this
required
threat
required
boost.
Federals
still
had
some
concern.
I
had
put
put
commented
on
this:
why
do
we
need
it.
D
I
think,
let's
restart
old
ones,
probably
we
can
ignore
it.
He
shows
I
created
these
two
issues.
These
are
not
very
urgent,
but
I
think
it's
good
if
somebody
really
want
to
pick
any
one
or
some
any
any
time
in
future.
It's
just
that
for
windows,
platform
and
apple
platform.
D
I
mean,
depending
on
curl
main,
is
not
a
good
idea
when
both
of
these
have
their
own
for
http
client
as
part
of
their
sdk
in
the
language
itself.
So
when
I
net
is
there
cf
network?
Is
there
so
and-
and
our
current
open,
telemetry
c,
plus
plus
already
provide
provide
this
flexibility
where
we
can
plug
in
any
http
clients?
D
So
I
think
it's
probably
in
future.
I
think
it's
good
to
go
to
remove
that
dependency
for
both
mac
and
windows,
so,
let's
just
just
create
it
so
that
we
can
track
it
in
future.
B
D
I
mean
the
reason
is
that
net
native
library
would
be
something
more
more
to
stay.
I
mean
in
general,
people
will
be
reluctant
to
include
any
external
libraries.
First
of
all
I
mean
being
in
windows.
I
I
may
want
to
use
anything
which
is
part
of
already
part
of
the
windows
library
instead
of
being
dependent
on
something
from
external.
D
So
that
would
be
one
of
the
thing
like
specifically
for
like.
If
it's
a
etw
exporter
I
don't
want
to
use,
I
don't
think
etw
is
dependent
on
girl.
So
that's
a
bad
example
like
on
windows.
If
I'm
going
to
use
gypkin
exporter,
I
really
don't
want
to
use
anything
about
anything
apart
from
json
external
json
library,.
D
Yeah,
so
basically
it's
already
there.
I
mean
it's
something
like
if
we
see
the
code
here
line
factory,
so
it's
already.
We
don't
need
to
do
anything.
There's
no
effects
condition.
It
will
just
see
if
any
of
the
http
library
is
present.
It
will
try
to
use
that
if
it
won't
find
any
http
library,
it
will
fail
with
the
symbols.
Unresolved,
error.
D
D
Again
on
me,
I'm
I
have
to
see
this
I'll
see
that
these
all
are
metrics
support
for
instrumentation
library
and
these
two,
this
one
already.
We
have
a
pr-
and
this
also
owen,
is
working
on
the
pr.
D
D
D
Which
one
is
this
okay?
This
was
not
introduced
by
me.
Let's
see
make,
and
this
is
happening
again
happening
on
a
particular
compiler
for
gcc
7.5
right,
yeah
7.5
and
it
is
reproducible
consistently.
So
now,
probably,
we
may
add
a
patch
specifically
for
this
compiler
and
then
let
it
work.
D
Yeah,
that's
all
we
have
anything
else
we
wanted
to
discuss
today.
D
Okay,
I
think,
if
nothing,
I
think
we
are
good
and
then
we
are
not
going
to
meet
next
week.
I'll
cancel
the
meeting
for
next
week
and
then,
let's
meet
next
year,
have
good
holidays.
Everyone.