►
From YouTube: 2021-03-17 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
B
B
D
F
G
Yeah
there
are
folks
that
in
the
pandemic
bought
high
def
cameras
and
a
lot
of
things
I
I
should
do
that.
You
know,
but.
F
They
even
have
those
monitors
that
attach
to
like
usbc
and
they're,
just
like,
like
just
attached
like
they're
portable
monitors,
those
are
pretty
cool.
F
F
Yeah,
I
think
some
of
them
might
have
can
get
their
own
power,
so
maybe
an
advantage
there
for
those
ones.
The
self-powered
ones.
C
Really
a
modern
setup
is
you
have
a
usbc
hub
and
it
has
one
cable
to
a
usbc
charger
that
should
put
out
100
watts,
and
then
it
has
one
cable
to
the
computer,
who
both
charges
through
that
cable
and
puts
out
video
at
the
same
time
and
then
another
cable
to
a
monitor
that
both
gets
powered
by
this
cable
and
gets
video
signal.
At
the
same
time,
through
the
camera.
F
Yeah
I
have
the
cal
digit
ts3
plus,
which
does
that
it
doesn't
work
well
with
our
the
windows
machines
we've
been
assigned,
so
I
only
use
it
for
my
home
device
but
yeah,
it's
just
one
usb
well,
one
thunderbolt,
cable
and
then
yeah
supplies.
You
know,
there's
a
display
port
out,
there's
like
four
different
or
like
five
different
usb
outs.
There
headphone
another
thunderbolt,
it's
great.
C
F
F
C
No,
no
thunderbolt,
yes,
but
I
mean
the
actual.
Is
it
the
the
name
of
the
switch
of
the
usb.
F
G
Yeah
all
right,
I
think
I
share
here
the.
G
All
right,
I
I
put
the
the
things
for
status
that
I
think
it's
opportunity
for
us
to
have
a
quick
chat
here
and
actually,
I
kind
of
I
think
I
miss
that
whiteboard
discussions
with
people
in
the
office.
G
So
we
can
have
quick
discussions
to
kind
of
flush
out
the
issues
here
and
I
think
it
will
be
a
bit
more
productive
than
the
just
going
back
and
forth
on
github
for
some
of
the
issues.
G
So
I
I
would
like
to
ask
because
robert
has
the
pr
for
the
status
and
zach
made
a
comment.
Do
you
guys
have
a
way
forward?
I
mean
both
robert.
There
is
a
path
that
you
see,
ford
or.
A
G
A
G
Cool,
so
I
I
I
just
want
to
be
sure
that
this
is
something
that
is
kind
of
not
blocking.
You
have
a
path
forward
and
you
can
adapt
to
to
to
be
in
the
way
that
zach
mentioned
there
in
the
comments.
A
F
Okay,
yeah
it's!
It
might
be
hard
to
construct
because
it's
a
combination
of
auto
instrumentation
and
using
the
sdk
yeah.
A
I
know,
but
I
think
that
you
know
there
may
be
more
changes
like
that
in
future,
so
I
think
it's
good
to
know
how
to
test
our
stuff
and
maybe
even
create
some
little
markdown
file
whatever.
Even
on
my
I
don't
know,
you
know
my
some
gist
or
whatever
just
if
there
will
be
more
issues
like
that,
so
we
can
quickly
run
through
the
same
issue,
because
I,
what
I
understand
is
that
similar
changes
can
be
similar.
Problems
can
be
whenever
we
change
this
internal
stuff
right.
F
Yeah,
basically,
if
the
even,
if
there's
internal
apis
so
for
everyone's
kind
of
knowledge,
this
is
a
development
limitation.
We
have
on
our
side
and
this
kind
of
surfaces
it
where,
because
we
kind
of
do
both
well,
at
least
on
the
datadog
side.
This
doesn't
really
apply
for
the
this
automation,
but
we
have
a
nuget
package
which
people
can
compile
against
and
use,
and
then
there's
also
the
auto
instrumentation
package.
F
But
we
rely
on
the
same
assemblies,
and
so
even
internal
assemblies
are
prone
to
like
missing
method.
Inception
stuff
like
that,
because
there
is
a
constructed
scenario
where
a
auto
instrumentation
like
1.20
is
run
and
the
user
has
a
newer,
1.21,
nuget
package,
and
so,
if
there's
any
apis
that
were
removed,
then
we
find
a
missing
method
exception.
So
until
that's
resolved,
we
kind
of
have
that
same
limitation
in
terms
of
the
changes
we'll
take
back
from
the
open,
telemetry
repo
that
remove
apis.
F
It's
anything
that
is
in
yeah,
that's
a
question:
it's
anything!
That's
in
datadog.trace
assembly
or
yes,
anything
in
data.trace
assembly.
That's
getting
called
from
the
data.trace
clr
profile
are
managed,
so
any
calls
made
across
that
boundary,
even
if
they
are
internal.
B
F
C
Something
we're
trying
to
solve
ourselves.
It's
it's
not
realistic.
It's
because,
if
you're
a
corporation,
you
can't
just
update
your
application,
and
so
essentially
it
would
mean
that
people
cannot,
if,
if
really
forced
it,
that
means
people
can
never
update
the
tracer,
because
updating
an
application
for
a
corporation
is
sometimes
a
no-go
like
it
is
what
it
is.
C
Actively
working
on
it,
I
mean
in
in
the
long
term,
if
we
can
use
activities,
this
problem
goes
away
right,
but
in
the
in
the
midterm
it
is.
G
This
is
something
that
perhaps
I
I
don't
want
to
put
more
work.
We
already
have
a
bunch
of
pr's,
but
this
is
perhaps
something
that
we
could
use
some
tool
to
kind
of
expose
the
apis,
and
if
there
is
change,
we
kind
of
break
some
tests.
You
know
so
at
least
call
people's
attention
to.
H
F
Greatly
benefit
us,
since
that's
really
a
problem
and
not
a
hotel
problem,
so
maybe
I'll
just
add
that
to
a
backlog,
that's
a
really
good
idea.
G
Cool
cool,
the
other
big
pr
that
and
that
I'm
saying,
because
of
the
this
related
to
through
the
start
of
swing,
is
that
change
is
about
trace
id.
G
So
I
think
we
need
to
be
care
of
the
same
limitations
in
api
changes
that
were
were
mentioned,
but
this
is
just
a
scalar
there,
you
long
and
so
perhaps
it's
simpler,
but
do
we
have
anything
regarding
trace
id?
I
I
would
ask
zach
to
take
a
look
at
that
pr.
It
touched
a
lot
of
places,
but
I
think
it's
one
of
the
kpis
to
ensure
that
we
can
keep
the
flow
going
back
and
forth
going
well.
You
know.
G
All
right
by
the
way,
I
think
I
should
have
done
the
the
announcement,
but
zach
is
officially
a
maintainer.
Now
you
know
very
cool.
B
G
C
G
So
I
I
would
love
to
have
zach
and
increase,
because
me
and
gregorette
did
some
urges
in
the
past,
but
I
want
to
see
you
guys
clicking
the
merge
button
too.
G
Okay,
so
so
for
regarding
the
text,
I
did
just
asking
for
reviews
for
the
time
being
and
if
we
need
something
else,
we
will
discuss
from
that.
I
think
on
the
jaeger
code
review,
there
is
a
discussion
about
the
the
copyright
headers
and
I
don't
recall
this
from
the
top
of
my
mind.
I
don't
think
you
are
required
to
have.
G
Then
we
are
required
to
have
the
proper
nodes,
but
I'm
not
100
sure,
but
the
only
thing
that
I
would
say
is
that
if
we
are
gonna
do
anything
about
this,
we
should
do
in
bulk.
You
know
it's
not
a
problem
for
integration.
I
think,
because
it's
kind
of
isolated
as
they
had,
but
I
don't
think
we
should
do
this
piece
mew,
you
know.
If
we
do,
we
do
for
the
whole
ripple,
we
add
some
tool
to
verify
and
require
the
header
or
we
don't
do
for
for
the
whole
thing
you
know.
G
So
I
don't
know
if
there
is
any
open
telemetry
that
is
not
using
the
headers,
but
I
don't
think
that
is
a
requirement.
I
can
have
a.
G
G
G
Thanks
robert,
I
I
that
that
that's
good
information,
so
let's
postpone
this-
let's
not
add
them
now
and
eventually,
when
we
add,
then
we
have
to
enable
some
check
to
ensure
that
everything
has
it.
You
know-
and
we
do
invoke
not
at
this
time
all
right.
G
G
All
right
and
this
week
first
time
it
was
not
me
doing
the
pool
for
upstream
it
was
great
that
david
was
doing
it.
I
I
have
gave
some
instructions
to
him
and
he
improved
it
on
top
of
them
and
david.
I
will
ask
you
one
thing:
can
you
create
a
pr
adding
to
the
docs
with
the
instructions
updated.
G
So
then
we
can
share
this
burden
of
sure
doing
the
the
pull
for
upstream
from
with
everyone
you
know,
so
other
people
can.
G
And
one
thing
that
I
noticed
when
I
was
changing
the
readme.
I
would
like
to
remove
the
readme
to
the
docs
to
folder,
because
azure
pipelines
doesn't
it.
There
is
no
way
that
azure
pipelines
can
avoid
ci.
G
If
you
are
changing
the
readme
on
the
root
it
needs
to
be
on
docs
and
github
recognize
the
readme
if
it's
under
the
docs
folder.
So
I
would
ask
somebody
to
move
there.
They
read
me
to
to
the
docs
folder,
perhaps
david,
when
do
this
dark
change
and
then
we
can
avoid
ci
for
for
doc
chains.
You
know
so
we
saved
some
our
carbon
footprint
in
the
in
the
azure
pipelines.
G
F
One
update
from
my
end
is
right
now
we're
sort
of
in
a
code
freeze
because
we're
preparing
a
release,
but
after
that
I
plan
to
port
some
of
the
pr's
that
in
this
repo
that
are
applicable
to
the
data
dog
side-
and
I
think
I
already
worked
with
paolo
to
get
those
lists
of
pr.
So
I
think
we're
good
on
the
definition
of
those
I'll
just
be
doing
those
later.
G
G
H
No
complaints
but
I'll
be
out
next
week.
So
if
you
need
something
from
me,
reach
out
this
week,
all
right.
F
If
we
don't
have
anything
else,
I
have
also
prepared
something
really
small.
Let
me
just
refresh.
G
G
C
So
essentially,
it's
like
an
integration
between
profiler
and
and
the
tracer,
where
you
have
your
traces
and
then
you
know
whatever
you
call
them,
but
essentially
the
the
activities,
the
the
spans
and
then
for
each
one.
Like
four
four
don't
know
it
would
be
activity
the
corresponding
thing,
essentially
for
each
one.
C
You
say:
while
this
activity
was
actively
executing
and
over
time,
it
may
or
may
not
be
actually
executing
all
the
time,
because
if
an
activity
was
essentially
on
a
thread
and
then
it
got,
is
it
made
a
remote
call,
so
it
wasn't
actually
running,
but
every
for
the
time
that
it
has
been
executing
what
time
was
spent
on
which
methods.
H
Yeah
so
as
far
as
the
way
that
that's
been
handled,
I
mean.
H
H
C
H
So
this
usually
happens
in
the
case
of
probably
wcf
client
instrumentation
would
be
a
good
example
where
one
method
will
initiate
the
asynchronous
call,
but
then
there's
another
method
that
will
ultimately.
H
Finish
it
yeah,
because
it's
the
old,
older
style
of
async,
where.
H
Methods,
and
so
that's
where
you
have
to
track
state
using
some
other
context,
propagation
mechanism.
C
That
makes
sense,
but
whatever
whatever
mechanism
you
you
choose
right,
you
have
the
span
and
then
say
you
solve
the
problem
of
when
you
started
and
then
to
finish
it.
So
that's
that's
good.
Now
you
have
a
customer
who
has
uses
your
profiler
and
now
you're
like.
I
know
that
dynatrace
does
this
at
least
they
claim.
C
Although
the
profiling,
I
I
couldn't,
get
it
to
work
honestly,
it
seems
to
to
switch
itself
on
for
for
a
short
period
of
time
and
then
switch
itself
off
and
but
essentially
the
idea
is
that
you
you
have.
You
have
spence
right
and
then
you
look
at
one
span
and
whatever
make
it
start
to
finish
right.
It's
called
even
like
a
consider
simple
case
right
where
it
is
the
like.
Some
I
entered
entered
some
library.
The
library
did
something
and
finished
some
library.
C
It
may
or
may
not
have
asynchronous
points
in
there,
and
but
this
is
one
spam
but
method
level
tracing
is
a
heat
offered
by
by
people
is
when
I
say:
okay
now,
this
span
actually
spent
three
milliseconds
in
method
full
and
five
milliseconds
in
method
bar
and
like
three
more
milliseconds
in
methods,
foobar,
okay,
right
and
and
and
so
so
you
have
profiling.
That
tells
you
well
overall,
in
this
last
minute,
you
spend
three
seconds
in
method:
football
right,
but.
H
H
C
Not
not
to
spend
for
each
method,
but
for
each
span
that
you
have
from
the
tracer
anyway,
you
say
where
the
time
of
the
span
went
in
which
methods
yeah.
H
So
we
try
to
take
the
stance
we
just
report
the
times
for
for
the
spans,
and
if
you
want
to
break
things
down
further,
then
you
would
have
to
ultimately
add
more
instrumentation
to
get
more
fine-grained
timing
on
things.
Alternatively,
you've
got
the
this
separate
thing
that
we
call
the
thread
profiler,
where
you
can
run
for
a
period
of
time
where
it's
doing
it's
using
those
profiling
apis
that
we've.
C
C
Yeah
from
from
that,
you
get
aggregated
from
that.
You
know
how
much
time
you
spent
in
which
methods
over
a
particular
point
in
time
that
you've
chosen,
but
I'm
thinking
about
genuinely
connecting
information
gets
out
of
the
tracer.
It
was
information
that
gets
out
of
the
profiler
and
I
don't
see
a
problem
with
doing
it,
but
I
do
see
a
lot
of
problems
with
doing
it
fast.
H
Yeah
so
so
there
was
something
interesting
mentioned
in
the
collector
pr
that
paulo
mentioned
before,
and
I
believe
noah
was
the
one
that
left
a
comment
on
that.
Pr
that
talks
about
some
of
that
information
coming
through
event
pipe
and
potentially
having
the
ability
to
associate
that
to
activities.
C
C
Profiling
apis,
you
already
control
the
thread,
and
then
this
red
local
storage
already
knows
which
span
is
running
because,
whatever
you
do,
you're
very
likely
using
thread
local
storage,
and
so
if
there
was
some
easy
way
to
access
the
manage
thread.
Local
storage
from
from
the
profiling.
H
C
Yeah
yeah,
I
had
really
weird.
This
is
probably
not
not
very
productive,
but
if
people
wanna
hear
about
really
funny
stories
about
this,
like
I
can
tell
you
some
when
I
was
playing
around
with
this.
Like
things
just
freeze
like
you,
you
call
you
call
from
the
thread
that
does
the
profiling.
C
You
call
managed
code
and
it
works
perfectly
fine,
but
then
three
seconds
later
in
the
middle
of
something
it
just
freezes
yeah,
and
I
wish
I
I
was
just
so
curious
to
know
why
I
like,
I
just
learned
not
to
do
it,
but
I
really
would
love
to
actually
learn
about
the
mechanism.
What
causes
it
to
freeze.
H
Yeah,
my
c
plus
plus
skills
weren't
sharp
enough
they're
using
some
sort
of
precondition
stuff
on
each
of
the
profiling
methods.
That
can
then.
C
Yeah,
the
problem
is
that
the
bits
where
things
deadlock
later
like
so
you,
you
call
managed
code
from
this
thread
and
you
call
manage
code
and
you
return.
You
continue
running
and
then
some
it
does
a
bunch
of
work
and
then
it
freezes.
So
it
freezes
a
certain
point
in
time,
but
the
work
that
it
was
doing
when
it
froze
has
already
been
done
between
the
freezing
point
and
the
time
that
you
called
managed
code
and
yet
managed
code
is
what
triggers
it.
So
if
you
never
called
manage
code,
then
it
never
happens.
C
So
what?
What?
What
really
is
weird
is
that
it's
not
like
next
after
you
call
next
time
you
do
this
kind
of
work.
You
freeze,
okay,
fair
enough,
you
change
some
state.
No
you!
You
do
that
particular
kind
of
work
several
times
and
then
you
freeze,
so
you
didn't
really
change
the
state.
You
just
created
some
sort
of
favorable
condition
for
a
deadlock
that
didn't
exist
or
something.
H
Yeah,
there's
something
about
the
some
of
the
things
that
get
left
on
the
thread.
Once
you
call
into
managed
code
that
causes
something
weird
to
happen:
yeah
and
yeah.
I
I
don't
understand
enough
to
know
what
that
is.
Yeah.
G
So
one
thing
on
that
note,
and
chris
already
mentioned:
I-
I
think
that
from
our
side
at
least
a
lot
of
the
stuff
that
we
potentially
could
do
through
event
pipe,
we
are
going
to
be
looking
at
doing
via
the
collector.
You
know
I
I
know
that.
Probably
that
doesn't
work
for
you,
your
scenario
greg,
but
this
is
where
we
kind
of
plan
to
invest.
G
If
we
need
more
information
from
event
pipe
stuff,
you
know
we,
you
already
have
the
protocol
there
right
now,
it's
just
metrics,
but
just
metrics,
and
I
don't
want
to
sound
it.
It's
it's
a
great
thing.
It's
already
absolutely
yeah,
you
know,
but
if
we
have
more
information
there,
we
we
plan
to
kind
of
build
on
top
of
that
event,
pipe
implementation.
C
Yeah,
anyway,
and
by
the
way,
so
the
diagnostic
source
stuff
I
just
got
bogged
down
and
like
formalities,
was
sharing
this
I
like
good
bureaucracy.
So
I
hope
that
in
the
next
few
days
it
will
be
cleared
up
and
I
can
share
it
and
we
can
start
using
it.