►
From YouTube: 2022-02-09 meeting
Description
No description was provided for this meeting.
If this is YOUR meeting, an easy way to fix this is to add a description to your video, wherever mtngs.io found it (probably YouTube).
C
B
Yeah
I
did
have
I
mean
that
was
there
in
my
mind,
I
thought
of
doing
it.
It
delayed
it
because
I
knew
that
it's
we're
not
going
to
change
what
this
meeting.
So
it's
not
next
meeting.
So
thanks
for
bringing
this
up
I'll
I'll.
C
B
Okay,
I
think
let's
go
quickly,
go
to
these
items,
which
were
there
from
previous
meeting
returning
fade
pointer.
It's
on
me.
I
think
I'm
not
getting
time
it's
my
local
changes,
I'm
not
getting
time
trying
to
push
it
in
a
branch
and
probably
give
it
to
to
the
team
to
test
it.
Let
me
see
if
I
can
get
it
this
week,
matrix
implementation.
B
We
did
discuss
the
architecture
diagram
and
to
be
discussed.
B
B
And
let's
discuss
more
about
the
matrix
at
the
moment,
once
we
start
discussing
the
pr
from
the
issues,
I
know
there
are
some
things,
probably
probably
for
sr,
I
mean,
as
you
were
doing
matrix
leader.
This
lots
of
things
would
not
be
still
clear.
Let's
discuss
that
once
we
start
discussing
with
yes,
but
I
I'll
take
this
to
add
more
description
on
the
floor.
B
C
B
B
C
B
B
D
D
B
A
B
Normally
we
have
to
remove
these
people
individual
people
from
here,
and
then
we
can
see
if
you
really
want
to
try
it
out.
C
B
B
B
And
I
thought
they
would
be
joining
today.
They
wanted
to
contribute
some
of
some
libraries.
They
had
created.
Two
open
telemetry
sleepless
place,
I'm
not
sure,
because.
B
So
I
think
I'll
I'll
say
dotnet
is
there
which
is
more
mature,
but
the
doctor,
the
problem,
would
be
that
we
don't
have
a
api
layer.
We
don't
have,
they
don't
implement
the
specification
api,
so
probably
that
could
be
could
be
a
difference
to
really
understand
the
flow
and
how
they
are
doing
it,
but
dotnet,
javascript
and
java.
B
B
Javascript
is
then
dot
network
bit
easier
as
compared
to
to
see
how
they
are
implementing
it
and
but
but
the
way
I've
seen
javascript
and
java
doing
most
of
the
things
in
the
same
way.
So
they
they
have.
They
create
multiple
meters.
Each
meter
will
have
its
own
writer.
The
writer
is
going
to
stay,
store,
the
actual
metric
data,
and
it
maintains
the
aggregation.
Each
writer
maintains
execution
so
they're
doing
the
same
way.
The
code
is
structured.
The
same
way,
I'm
kind
of
following
these
two
and
python
python
is
also
a
bit
behind.
B
B
The
new
ones
which
we
have
implement
aggregation
storage.
This
is
something
I'm
doing
it.
I
think.
Once
this
is
completed,
we
will
have
more
of
a
complete
flow
for
creating
the
matrix
and
collecting
it,
because
this
is
the
piece
which
is
going
to
do
the
actual
metrics
too,
which
is
going
to
do
the
aggregation
and
store
the
metrics.
B
So
hopefully,
I
think
I'll
do
it
by
this
week.
The
bottom
lane
which
I'm
seeing
as
of
now
is
that
it
is
involving
lots
of
places
where
we
have
to
do
lock,
kind
of
that,
something
which
I'm
with
difficult.
B
I'm
trying
to
in
between
c,
because
we
need
to
maintain
basically
to
store
the
matrix
data
and
the
aggregations
we
have
to
maintain
a
hashtag
and
the
the
problem
is
that
hashmap
we
have
to
lock
so
every
time
everything
the
data
is
everything
a
new
new
measurement
comes.
We
have
to
do
a
log,
update
the
hashmap
and
then
then
do
the
aggregation
and
that
that
has
to
be
done
for
all
the
measurements,
and
that
has
to
be
done
even
when
we
are
doing
going
to
collect
the
measurement
from
there.
B
We
don't
have
any
data
structures
or
the
containers
so
trying
to
see
how
we
can
do
it,
but
just
in
case,
if
somebody
has
any
idea
a
better,
better
approach
or
existing
library
which
we
can
use
for
concurrent
dashboards
with
which
which
supports,
storing
and
erasing
it
should
support
both
of
them,
and
we
should
be
able
to
delete
that
delete
the
records
or
clean
up
the
records
and
you
should
be
able
to
add
the
unicorns.
B
B
D
And
I've
done
something
for
for
gpu,
not
for
cpu,
okay,
I'll
think
about
how
to
bring
something
called
cpu.
B
That's
the
complete
sdk
kind
of
thing
with
the
videos
we
just
want
to
use
one
separate
concurrent
hashmap
and
it's
very
difficult
to
really
get
the
pieces
from
that
and
isolate
isolate
that
hashtag
because
that
it
uses
dependent
header
files.
There
are
lots
of
hydrophones
which
it
is
using
internally,
which
is
also
coming
from
it's
difficult,
difficult
to
isolate
that
and
try
to
use
it,
but
yeah.
I
don't
want
to
over
or
engineer
or
over
optimize
as
of
now.
Let's,
let's
do
it
in
with
current
existing.
B
B
This
is
still
work
in
progress,
so
probably
by
the
end
of
this
week.
I
think
I
just
want
to
do
it
for
surely
by
this
week,
so
that
we
have,
we
should
be
able
to
have
an
end
to
it
flow
of
both
measurements
and
matrix.
B
C
B
B
B
B
There
may
be
some
changes,
I
I'm
not
very
sure
if
we
require,
but
that
makes
you
require
api
tubes.
Also,
it's
something
like
so,
but
I
don't
want
to
do
it
now.
Probably
we
can
discuss
it
afterwards.
D
C
That's
a
whoever
experience
differ
a
lot
than
the
native
versions.
A
B
D
B
D
Yeah,
I
think
I
I
don't
have
the
solution
yet
so
I
need
boost
both
headers
and
also
at
compiling
thrift.
So,
first
I
need
the
boost
for
compiling
thrift
and
then
I
need
boost
headers
for
using
thrift.
D
I'm
still
trying
to
find
a
way
to,
and
it's
I
don't
know
why
they're
using
boost
it's
like
they,
they
are
inheriting
from
one
class
like
non-copyable.
D
D
Yeah,
the
linking
for
building
this
thrift
itself,
but
when
you
consume
it,
you
are
using
some
header
from
thrift
that
header
uses
boost.
B
D
D
B
B
D
D
The
issue
now
it's
I'm
using
a
boost
from
rules
and
I'm
trying
to
get
that
one
working
to
try
to
have
so.
The
main
issue
in
the
beginning
was
in
atomic.
It
boosts
atomic.
It
tries
to
build
another
library
which
doesn't
link
incorrectly,
that
one
phase
and
but
we
don't
need
atomic.
So
we
can
just
use
boost
core
or
boost
to
that
doesn't
have
this
atomic
thing,
and
now
I
get
the
this
error
of
it
and
it
says
not
able
to
create
a
symbolic
link
to
some
file
and
there
is
no
information
on
that.
D
D
B
B
So
we
get,
we
get
the
measurements
from
so
from
the
instrumentation
library,
and
the
the
current
design
is
that
we
have
non-owning
pair,
which
we
use.
We
use.
B
B
Attributes
so
somewhere
in
our
sdk,
once
we
have
to
do
filtering
on
these
attributes
and
we
have
to
we
have
to
all
we
have
to.
We
have
to
store
these
attributes
in
our
sdk
and
we
we
have
to
do
some
processing
on
them.
So
each
of
these
measurement
activities.
B
We
have
to
make
a
copy
of
these,
because
this,
the
actual
memory
is
owned
by
the
instrumentation
library,
not
very
so,
once
we
start
showing
it,
we
have
to
create
a
copy
of
this,
and
that
is
a
concern
which
he
has
raised.
Why
do
we
are
creating
the
copy?
We
cannot
avoid
it,
but
probably
keep
it
like
this
and
he
realized.
B
B
Yes,
this
is
this
is
something
that,
if
somebody
comes
in
open
elementary,
therefore
they
want
to
see
the
status
of
metrics.
They
know
where
to
see.
But
what
are?
What
are
the
deadlines.
D
D
D
So
I
just,
I
think,
it's
better
to
keep
it
for
a
while,
so
we
can
see
which
tests
are
affected
the
most
I
saw
that
only
one
of
two
of
them
are
have
this
problem
very
often,
so
we
can
disable
the
alerting
for
those
ones
which
two
tests
did
you
see
it's
like
a
smoke,
some
smoke
test
with
s
being
locked
or
something.