►
From YouTube: 2021-08-04 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).
A
C
C
B
C
C
B
C
C
Yes,
yes,
I
mean
so
so.
Data
dog
has
hosting.
B
No,
our
entire
team,
all
the
like
tracer
teams,
we're
all
remote,
so
datadog,
already
kind
of
had
a
hybrid
of
fully
like
on-site
teams
and
fully
remote
teams.
So
I
I
joined
in
2019,
and
so
we
already
had
that
entire
team
was
distributed.
C
I
see
yeah
so
yeah
we
have
the
the
folks
in
in
eastern
europe
and
besides
the
focus
in
eastern
europe
for
having
me
here.
D
C
C
C
Okay,
once
more,
the
the
work
that
actually
robert
and
david
have
been
doing
have
been
most
on
the
sdk
side
relatively
simple
stuff,
but
they
are
updating
the
environment
variable
support.
So
we
don't
need
to
have
that
code
on
our
side.
C
We
are
also
on
the
plc
branch
taking
dependencies
with
dot
net
six,
because
the
only
new
get
package
that
the
sdk
is
really
easy
now
and
that
allow
us
to
pick
up
those
chains.
They
they
require
dot
net
six
6.,
since
we
are
not
going
to
be
releasing
kind
of
in
the
next
few
months.
That
seems
fine
to
me
and
it's
just.
I
think
that
we
need
to
document
on
the
plc
branch
and
make
sure
that
everything
is
working.
C
I
was
gonna
another
thing.
We
are
having
some
issue
with
the
otlp
exporter,
but
we
are.
C
We
are
not
able
to
get
it
working
with
the
collector,
and
I
was
hoping
if
somebody
that
perhaps
had
played
with
this-
and
I
think
the
team
here
today-
none
of
us
have
been
playing
too
much
with
that.
So
I
don't
think
we
can
kind
of
give
any
insight.
C
But
I
asked
david
for
him
to
take
a
look
at
this
and,
let's
see
if,
by
the
end
of
the
week,
we
have
some
resolution
about
that,
because
one
thing
that
I
think
we
should
be
doing
for
the
pocs
and
in
general,
because
the
oto
lp
export
is
the
main
exporter
for
open
telemetry.
C
I
think
we
should
be
doing
the
tests
the
integration
tests
eventually
against
otlp
on
this
branch.
You
know
so
we
we
need
to
fix
that.
But,
as
I
said,
I
think
the
two
of
us
here
today.
Not
nobody
has
an
insight
about
that.
I
was
just
hoping
to
see
one
thing
that
I
did.
Oh
maybe
so,
let's
let's
wait
for
him
and
I
asked
that.
C
I
was
just
mentioned
that
we
are
having
some
issue
trying
to
use
the
ot
otop
exporter
and
I
I
was
just
hoping
to
see
if
somebody
had
some
insight
about
that,
because
we
want
to
move
the
integration
test
to
the
otp
exporter
kind
of
the
open
element.
We
may
export
it,
but
we
are
having
some
trouble
running
that
with
against
the
collector
against
some
configuration.
C
So
if
you
have
any
insight
about
that,
let
us
know
I
asked
david
to
take
a
look
and
he
hopefully
maybe
have
some
update
by
the
end
of
the
week.
But
if
you
know
something
already,
maybe
alan
knows
something
about
this.
Let
us
know,
please
is.
E
It
really
with
the
otlp
explorer.
I
thought
the
pr
was
mentioning
something
about
jaeger,
not
working.
C
I
I
think
I'm
I'm
more
interested
on
the
old
tlp
exporter,
but
but
maybe
because
there
is
grpc
two
for
jager.
Maybe
it
suffers
the
same
issue.
You
know,
okay,
I
I
I
really
focus
on
the
oklp,
but
it's
possible.
You
know
I
I
I
didn't
follow
up
the
details
of
the
issue,
so
it's
possible.
E
C
So
if
you
could
take
a
a
look
at
the
issues,
okay
on
the
agenda
notes
you'll
be
helpful.
C
Yeah
cool,
so
we
had
from
last
week
some
the
questions
and
follow
up
about
removing
the
strong
name.
I
did
kind
of
dive
into
the
memory
lane
and
end
up
remembering
what
I
I
had
done
in
the
past
regarding
that,
and
that
was
actually
the
case
for
windows
device
bar,
because
there
was
a
exploit
that
load
assembly,
you
could
load
bytes
and
then
you
could
run
in
the
device
bar.
C
C
I
don't
think
the
machines
will
be
running
with
device
guard.
That's
why
I'm
saying
that
I,
I
don't
think
I
ever
encountered
a
a
customer
that
was
running
their
servers
with
device
guard
and
they,
if,
if
that's
a
scenario
that
requires
its
own
handling
of
signature
of
any
assembly,
that
you
wanna
not
only
an
assembly,
any
any
file,
any
library
that
we
want
to
be
allowed
to
belong.
So
I
I
consider
that
to
be
a
very
specialized
scenario
that
we
don't
need
to
invest
at
this
time.
C
Any
looking
more
around
that
the
only
other
thing
that
I
I
try
to
find
related
and
I
look
at
this-
was
about
the
old
thing
that
it's
possible,
and
this
is
pretty
old.
It's
because
I
I
don't
remember
which
version
but
dotnet
started
to
bypass
a
strong
name,
verification,
and
it's
quite
old.
I
think
three,
five
or
four
something
like
that
start
to
have
that
by
default,
because
they
trust
the
the
zone
where
the
assembly
comes
from
and.
C
That
that
is
not
a
problem.
I
did
a
quick
test
in
the
poc
branch
with
the
binding
redirection
application,
because
you
can
re-enable
that
set
and
requires
the
verification
per
application
or
via
the
regis
for
a
box,
but
I
did
the
test
for
application
and
the
thing
is
still
works,
even
without
the
assemblage
that
we
load
from
instrumentation
being
being
not
signed,
not
strongly
signed.
So
I
think
we
are
good
in
that
regard.
I
still
put
on
the
the
bottom
of
the
notes.
C
I
still
have
to
verify
the
iis
and
botnet
course
and
irons
for
this.
I
hopefully
get
this
done
by
next
week
and
I
think,
regarding
the
strong
name
and
this
combination
of
strong
name
and
the
devops
scenarios,
we
are
kind
of
really
find
a
spot
that
we
can
do
the
devops
scenario
and,
as
I
said
last
week,
we
we're
probably
not
doing
this
anytime
soon,
but
we
know
that
we
have
paths
to
move
forward
on
that.
E
C
C
But
rasmus
should
be
back
just
next
week,
so
if
I
have
time
I
will
try
to
fix
that
and
perhaps
get
somebody
to
review
and
merge.
You
know
I
I
can.
I
can
push
definitely
on
top
of
his
work.
So
if
I
have
time
I
will
fix
that
and
try
to
push
this
chain
this
week.
So
we
can
keep
moving
the
poc
branch,
especially
because
that
one
is
pretty.
C
C
One
thing
that
I
would
mention
for
people
that
are
interested
in
in
the
metric
side:
they
are
asking
for
feedback
about
the
api.
C
That's
gonna
make
2.06
and
it's
a
relatively
trivial
question,
but
I
know
that
the
people
from
their
own
time
take
the
decisions
about
those
things
very
carefully,
so
they
are
basically
trying
to
get
some
input
about
the
number
of
overloads
that
they
should
have
out
of
the
box
to
support
a
number
of
dimensions
for
maps
you
know
kind
of
because
they
know
that
there
is
a
performance
penalty
for
when
you
have
a
variable
number
of
parameters,
so
they
try
to
have
those
overloads
that
take
a
specific
number
and
they
are
wondering
how
high
they
want
to
go.
C
You
know
from
our
perspective,
I
said
actually
that
six
is
pretty
good,
because
resources
don't
count
as
those
dimensions.
C
So,
even
if
you
have
a
four
five
that
come
from
resource,
then
having
six
it's
a
pretty
general
case,
we
do
have
some
time
series
with
many
more
than
that,
but
that
is
not
typical
at
all.
You
know
so
having
overloads
for
six
from
our
perspective
is
good,
but
they
are
asking
for
feedback
if
you're
interested
on
metrics
and
have
a
different
feedback.
It's
your
chance.
I
I
put
a
link
to
the
issue
on
the
slack
after
the
meeting,
so
if
anyone
has
some
input
to
that
jump
on
the
issue.
C
I
think
the
biggest
challenge
you'll
be
right
now
because
they
have
releases
that
are
not
as
frequently
I
mean
the
sdk,
and
we
do
have
some
space
to
ask
for
releases
for
us
if
you
need
things
to
be
fixed,
but
I
think
that
perhaps
is
a
a
bit
of
not
right
now,
but
I
think,
as
we
come
to
kind
of
hey,
this
is
really
the
the
the
thing
that
we're
gonna
do
and
we
start
to
get
ready
to
release
alpha
or
something
like
that.
That
may
become
a
risk.
C
You
know,
but
other
than
that
I
so
far
I
didn't
see
any
problem
you
know
kind
of
that
is
not
that
we
don't
have
a
workaround.
We
do
have
the
limitation
of
the
devops
scenario.
For
for
now
we
don't
have
any
automatic
solution
or
simple
solution
for
those,
but
we
know
that's
possible,
you
know.
So
I
don't
know
if
we
get
this
successfully
for
something
we
can
look
at
perhaps
invest
on
some
dotnet
that
can
help
generating
dividing
overrides
to
get
the
versions
to
satisfy
the
devops
scenario.
C
But
we
know
that
we
have
work
around
them
so
far.
Everything
that
I
tried
in
that
case
is
working,
so
I'm
feeling
pretty
good,
I'm
really
so
that
I'm
not
seeing
any
blocker
for
that
at
this
time,.
E
Okay,
so
the
sdk
is
focusing
on
metric
support
at
this
point
in
time,
but
a
lot
of
the
emphasis
has
been
on
the
api
side
of
it
and
only
a
little
bit
on
the
sdk's
handling
of
metrics.
E
E
Should
we
be
spending
some
time
going
through
those
that
don't
have
sdk
instrumentation
for
already
to
enable
some
metric
generation
based
on
the
conventions.
C
I
I
think
that
it's
a
very
good
idea.
I
would
just
say
that
kind
of
if
we
look
to
to
the.
E
C
C
You
know,
even
in
a
sense
that
things
like
this,
the
net
six
api
should
be
frozen
this
week
next
week,
but
the
open
telemetry
usually
provide
light
wrappers
around
it,
and
just
because
of
going
back
and
forth,
I
I
think
it's
better
if
we
start
to
work
with
those
rappers
or
conventions
that
open
the
limited
sets
when
they
are
really
less
changeable.
You
know.
E
Okay,
yeah
and
then
yeah.
The
other
thing
is
that
it
might
be
the
case
that
most
of
the
libraries
that
we're
targeting
for
the
alpha
probably
have
sdk
instrumentation
for
already.
C
Yeah
some
of
that
stuff,
and
we,
since
we
can
load
I'm
trying
to
find
a
term
I'm
calling
those
source
instrumentation,
but
I
think
our
preference,
where
is
available,
we
should
try
to
load
the
source
instrumentations
like
we
are
doing
for
asp.net
core
on
the
plc
branch.
We
should
always
try
to
do
those,
but
there
will
be
some
that
we
are
going
to
still
be
doing
bytecode
instrumentation
and
for
those
we
we
we
should
have
kind
of
good
targets.
I
I
wanna.
C
I
definitely
I
think
everyone
here
wants
to
avoid
the
case
that
we
go
through
the
trouble
of
creating
some
bite
code
instrumentation,
and
then
we
discovered
that
there
is
a
source
code
instrumentation
you
know,
so
we
should
look
carefully
at
those
to
avoid
kind
of
having
to
do
some
work
for
bike
code
instrumentation
and
not
and
then
later
discovering
that
oh,
we
can
use
the
source
instrumentation,
so
yeah.
I
think
I
think,
especially
for
database
stuff.
C
We
we
have
to
pay
a
lot
of
attention
because
there
are
some
interfaces
and
types
that
we
could
use
for.
That
and
typically
drivers
for
a
specific
database
are
implemented.
On
top
of
those,
so
we
should
review
those
to
be
sure
that
kind
of
we
have.
C
There
is
a
sweet
spot
that
we
can
take
one
class
one
type
and
perhaps
even
with
source
instrumentation
over
a
bunch
of
things,
but
for
right
now
I
think
we've
been
kind
of
careful
about
choosing
a
small
number
of
bytecode
instrumentation
that
we
know
that
we
don't
have
alternatives.