►
From YouTube: 2020-05-28 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
A
A
B
C
C
A
B
A
B
A
A
B
Yeah,
well,
it
is
just
us
so
we
can.
We
can
go
over
pretty
much
whatever
like
I,
often
kind
of
do
this
spec
sig
just
run
down.
We
can
do
that.
We
can
skip
that
if
you
yeah
just
tell
me,
what's
gonna,
be
most
useful
for
you
and
yeah.
B
Yeah
yeah
I'm
familiar
with
a
lot
all
this
stuff.
There
yeah.
C
B
Like
the
biggest
difference
is
kind
of
like
the
modeling,
like
a
New,
Relic
kind
of
has
this
like
a
transaction
model,
you
kind
of
aggregate,
the
spans
up
in
process
or
I,
think
they're
we're
calling
or
you're
calling
them
segments.
B
You
kind
of
an
aggregate
that
was
up
in
each
process
and
kind
of
build
a
trace
out
of
them
or
generate
span
events
and
then
kind
of
send
them
up.
I
would
say
like
the
biggest
difference.
Is
that
open,
telemetry
kind
of
has
the
dapper
model?
So
really
the
biggest
thing
you're
ever
really
worried
about
is
a
span
and
the
current
span,
and
you
know,
as
long
as
it
has
proper
trace.
Iv
parent
ID,
like
it
all,
gets
stitched
up
and
the
trace
is
kind
of
eventually
consistent
in
your
back
end,
okay,.
A
B
I
was
a
very
big
advocate
event,
actually
I.
Think
I
think
the
model
made
a
lot
of
sense.
You
know
when
that
agent
was
first
written
and
you
know
it
was
I
think
it
was
started
back
in
like
2007
and
it
was
really
about
tracing
your
rails.
App
distributed
tracing
was
probably
not
a
thing
you
were
concerned
about
and
it
just
kind
of
like
mild.
Oh
requests
going
through
kind
of
your
rails,
application,
yeah.
A
Sure,
because
I
remember,
when
push
came
out
the
party
can
you
use
you
that
could
actually
give
it
some
sort
of
insight
up
the
case.
You
were
doing
John
Simon's
that
we
will
probably
the
first
company
to
try
to
do
to
Michael
architectural
approach.
Mike
authorities
approach
to
around
that
because
think
of
a
big
monolith,
which
is,
are
you
doing
back
then?
Okay,
take
a
hard
time
figuring
out
where
things
are
breaking.
B
B
A
B
So
I
I
think
he
would
be
well
well
prepared
for
something
like
that
and
yeah,
knowing
that
it's
kind
of
just
this
span
model
like
the
really
you
just
need
to
create
the
right
spans
and
then
the
next
thing
is
just
like
annotating
them,
so
adding
the
proper
attributes
to
them.
So
I
think
I
think
what
were
these
called
parameters,
segment,
params
or
something
on
the
New
Relic
agent?
But
there's
this
a
whole
list
of
semantic
conventions
to
follow.
I,
see
Erin
popped
in
good
morning.
Hey.
A
A
A
B
D
Yeah
there's
traces
with
many
spans
and
then
there's
different
subcategories
of
spans
and
things
like
that
yeah,
because
there's
a
lot
of
massaging
that
the
the
agent
does
ahead
of
the
everything
getting
submitted.
So
that's
a
lot
more
of
the
that's.
If
anything,
that's
the
biggest
secret
sauce
of
the
data
dog.
That's
not
really
that
secret
cuz!
It's
in
the
open
source
agents,
so
yeah
a
lot
of
there's
a
lot
of
pre-processing
and
correlation
that
goes
on
before
things
get
submitted
up
towards
data
dog
directly,
so
yeah.
B
D
There
really
aren't
many
closed
source
things
with
David
on
the
anything
like
the
things
that
are
public
are
pretty
much
all
the
server-side
processing
tools,
all
that
stuff,
pretty
much
everything
that
actually
makes
the
magic
happen
on
the
other
side.
So
but
everything
else
is
this
open
core?
It's
a
lot
of
open
stuff,
open
source
stuff
that
we
manage
it's
kind
of
insane.
C
D
Yeah,
like
it's
it's
in
the
it's,
not
even
in
the
tens,
it's
in
the
hundreds
in
terms
of
projects,
it's
pretty
intense,
we
yeah,
we,
we
have
a
spreadsheet,
still
the
track,
a
lot
of
it
and
quite
a
few
of
the
move.
Rna
they're,
given
away
your
closed
down.
Some
of
them
are
just
Forks
that
stuff,
because
a.
D
B
D
A
D
Yeah
I
mean
the
the
newer
VM.
The
newer
yarv
gives
you
a
lot
more
detail
now,
so
you
can
actually
introspect
that
from
the
Ruby
side
of
things
and
not
incur
major
performance
penalty
anymore,
but
a
lot
of
the
like
the
tracing
stuff,
that's
baked
into
Ruby
is
pretty
darn
slow
because
it's
mostly
in
reveal
and
and
not
in
seal,
and
so
that's
why.
D
Rust
to
do
to
that
particular
level
of
profiling,
yeah
they're,
one
of
the
first
like
ship
trust
projects
that
ever
known
as
that.
D
D
D
So
they've
got
profilers
for
CPU
heap
allocations
and
all
that
other
stuff
to
actual
trace
events
like
within
the
VM
itself,
so
it's
pretty
interesting
profiling
set
up
so
we'll
see
if
open
telemetry
ends
up
integrating
them,
I,
don't
know
if
there's
an
RFC
for
it,
I
didn't
find
one
so
I,
don't
know
if
that
has
ever
actually
been
true
discussion
about
it.
I
know
if
you've
seen
anything
Matt
from
the
the
discussions
from
open
telemetry
on
that
side
of
like
if
there's
been
any
talk
about
profiling.
B
B
Way
to
track
the
amount
of
time
that
a
fan
spends
working
on
its
own
self.
You
kind
of
subtract
out
all
of
the
child's
time
and
there
are
ways
to
kind
of
try
to
do
this
as
you're
tracing,
but
I
think
the
approaches
are
fairly
naive
and
like
pretty
involved
and
in
the
end
you
kind
of
get
an
approximation,
but
I
don't
think
it's
exact,
actually
I
think
you
would
actually
need
profiling
to
get
those
things
to
a
level
of
accuracy.
That
makes
a
lot
of
sense.
B
D
Get
even
higher
fidelity,
you
know,
cuz
like
all
tracing
and
even
open
senses
and
stuff
we're
doing
or
just
like.
Let's
just
you
know
time
everything
going
on
inside
of
your
apps
based
on
the
time
signature,
you
can
kind
of
spot
the
the
slowness
and
stuff
like
that,
but
to
gain
additional
detail
having
profiling
will
tell
you
like
part
of
why
it
is
slow,
so
yeah.
B
B
B
D
Know
30,
oh
yeah,
if
you
could
do
that
at
least
or
check
and
see
what's
up
with
the
the
milestones
there,
I
haven't
paying
a
lot
of
attention
to
because
I'm
scrambling
to
get
all
the
stuff
together
for
shevkov,
which
is
next
week
and
stuff.
So
it's
like
other
projects
that
have
like
slid
back
into
I,
approve
you
and
then
I've
got
the
bigger
webinar.
Just
after
that,.
B
Yeah,
okay
share
my
screen.
B
B
B
B
B
D
That
that
that
was
quite
quite
the
thing
to
tackle
like
there
are
other
ones
that
are
gonna,
be
a
lot
simpler,
I
think,
like
the
rack,
implementation
can
be
done
better,
like
what
are
the
other
low-hanging
fruit
that
we
could
make
a
thing
for
is
to
sort
of
standardize
the
rack
adapter.
So
that
way
it
gets
pulled
in
through
either
Sinatra.
B
D
Just
rack
and
plain
in
general
that
way
it
becomes
more
of
a
groups:
dependency
yeah.
If
you
grab
Sinatra
it
automatically
just
injects
the
rack
adapter,
which
is
the
way
it
should
just
work,
or
at
least
we
should
hook
rack
directly
and
then
there's
not
much
else.
That
Sinatra
needs
to
do
on
top
of
that.
Yeah.
D
B
Okay,
yeah
I
haven't
looked
at
our
rack
stuff
for
a
long
time,
but
I'm
pretty
sure
that
it's
not
our
portion
I
was
thinking
more
Sinatra
I,
don't
look
at
our
Sinatra
in
a
while,
but
I'm
pretty
sure
it's
not
otto
injecting
drag
no.
D
It's
not
it
does
it's.
It
Auto
injects
its
own
rack
adapter,
so
we'd
probably
want
to
just
figure
out
what's
different
between
that
and
the
rack
one,
and
then
we
would
have
to
refactor
the
the
the
Sinatra
adapter
to
basically
just
depend
on
the
rack,
one
for
starters
and
then
later
on
anything
specific
to
Sinatra
from
there.
D
Basically,
if
you
see
Sinatra
then
go
ahead
and
inject
the
rack
adapter
so
make
it
a
direct
Jem
dependency,
so
grab
so
now
tray
you
get
rack
for
free
and
if
there's
anything
in
particular
that
Sinatra
needs
for
ya,
like
the
render
template
call
and
stuff
like
that,
that's
where
you
can
layer
that
end
or
we'd
have
to
expose
an
API
for
that
to
work.
That
way
at
least
any
improvements
in
the
rack
adapter.
D
The
rack
middleware
are
there
and
then
adding
in
stuff,
like
the
the
exceptions
and
all
that,
that's
where
you
can
put
in
the
exception
middleware.
So
then
there's
that
point
you
would
get
all
the
other
stuff
for
free.
It
would
just
that's
effectively
how
it's
been
done
for
like
the
DD
trace.
Gem
is
the
rack.
Adapter
is
peppered
into
a
lot
of
other
stuff,
so
anything
and
it's
that's
framework
related
just
pulls
the
rack
adapter
in
automatically.
So
this
Sinatra
rails
I
think
Konami.
D
B
Yeah
all
that
sounds
good
to
me.
It
sounds
like
you
have
an
idea
of
the
requirements
around
this.
What
would
make
it
best
and
most
easy
to
use
and
kind
of
pretty
good
understanding
of
how
it's
working
in
DD
traces?
Would
you
want
to
write
up
an
issue?
Maybe
with
that
yeah
I
could
write
up
an
issue
for
that.
We
do
that
after
the
call
cool,
oh
yeah,.
B
And
then
yeah
Michael
I
know
you
were
interested
in
trying
to
get
started
on
something
mm-hmm
I
do
think.
Like
you
know,
any
of
these
instrumentation
adapters
that
have
been
added
in
would
be.
You
know
like
a
good
first
thing,
so
you
have
to
do
it
now,
but
if,
if
you
wanted
to
look
through
these
and
like
just
pick
one
and
comment
on
it,
when
you
do
I
can
assign
it
to
you,
if
you
have
a
mind,
let
me
know:
okay.
B
A
A
B
Cool
yeah
I
would
like
to
have
like
a
better
like
milestone
definition,
but
I
just
don't
know
what
to
put
in
it
right
now,
just
because
I
think
like
we
don't
have
a
ton
of
people
working
on
things
and
I.
Think
the
people
who
are
currently
here
are
already
mainly
working
on
things
or
selecting
things
so
yeah.
D
B
We
were
really
we
were
a
little
more
structured
a
while
ago
and
we
had
milestones.
Then
we
just
started
kind
of
like
kicking
cans
down
the
road
and
that's
what's
kind
of
here
for
the
most
part
and
then
since
then
a
lot
of
new
stuff
has
been
added,
but
it
hasn't
really
been
prioritized.
So
I
would
say
there
is
some
newer
stuff,
that's
probably
more
important,
but.
B
Francis
Francis
from
Shopify
I
think
he
has
a
lot
of
opinions
on
this
and
I.
Think
Shopify
is
trying
to
actually
use
this
stuff
a
little
bit
more.
So
I
feel
like
it's
good
to
have
him
involved
in
milestone
Ignace,
because
we
can
make
sure
that
we're
adding
stuff
that
they
would
use.
You
know,
that's
that
somebody
will
use
so
like
I,
think
that's
kind
of
my
main
drivers
like
if
somebody's
gonna
use
it
you
should.
We
should
add
it
in
if
it's
just
kind
of
like
a
nice
to
have
like
I.