►
From YouTube: 2021-02-12 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
B
I
don't
know
if
anybody
else
is
gonna
show
up,
but
yeah.
The
spec
has
gone
1.0
for
tracing.
A
B
You
have
on
that.
I'm
gonna
do
another
page
that
specifically
talks
about
how
context
and
the
process
dictionary
are
used.
All
right.
A
Well,
that
kind
of
under
the
horde
is
like
yeah
super
value,
especially
because,
like
especially
in
the
the
elixir
community
there
is,
there
are
huge
chunks
of
it
that
happily
just
use
frameworks
and
have
no
idea
about
like
they
know
that
there's
otp
under
the
hood
right,
but
it's
abstracted
away
from
them
and
so
yeah
they're
all
like
no,
it's
100
immutable.
B
A
Yeah
and
it's
an
important
it's
like
it's
a
pleasant
distinction
because
you're
right
these,
these
people
are
they're
what
they
want,
don't
use
the
process
dictionary
and
and
but
nonetheless,
it's
there
and
it's
getting
used
like
like
another
example.
There
is
elixir
like
you've
got
the
co
dollar
callers.
B
A
Right
and
you're
like
well
when,
when
one
task,
you
know
you,
when
you
spin
up
a
task,
you
keep
track
of,
who
spun
it
up,
and
then
that
gets
used,
for
example,
if
you're
using
mocks
to
mock
out
a
call
for
something
it.
It
knows
that
there's
every
chance
that
you,
you
spun
up
a
task
to
do
that
thing,
so
how
the
hell
does
it?
A
B
A
A
It's
all
like
invisible
most
of
the
time,
which
is
great
but
yeah,
it's
like
so
it's
going
to
be
fun,
coming
up
with
the
right
way
of
saying
to
everybody
in
the
documentation.
You
know
it's
like
yes,
we're
using
the
process
dictionary,
but
no
you're
not
so
this
is
not
violating
the
advice
to
don't
use
the
pro
the
process
dictionary
of,
if
that's
still
ringing
areas,
don't
worry.
It's
like
yeah.
B
A
B
A
All
right
I
mean,
is
there
okay?
So
if
I
click
edit,
this
page
great,
it
just
goes
straight
through
to
github
and
then
it
says
404
I
think
because
well
I
think
it's
because
it's
it's
been.
It's
either
been
uploaded
from
a
branch.
A
B
A
B
A
B
Look
at
this
deck
is
released
and
we
get
stacks
up
so
just
letting
people
know
so
they
can
give
comments.
A
B
Yeah,
I
think
it's
nice
to
look
like
yeah
one
of
the
others.
One
of
the
javascripts
go
java.
A
Yeah,
so
there
was
that
there
was
a
thing
in
the
like
somebody
mentioned
like
they
ripped
out
something
everyone's
like.
Oh,
no,
I'm
going
to
miss
that
and
I
was
more
kind
of
like
I'm
fine.
I
can't
remember
what
the
what
the
controversy
was
about.
What
was
it
might
have
staled
out.
A
This
makes
me
sad:
1797,
tail-based,
sampling,.
B
A
Somebody
will
climb
yeah,
I
mean
in
the
the
honeycomb
universe
is
more
on
the
that
thing
where
they
just
try
and
do
a
probabilistic
one
where
it's
you
know
you
just
look
at
the
you
know
if,
if
the
trace
id
modulo
some
number
that
you
picked
as
zero.
A
Then
great
you
know
the
you,
you
include
it
and
then,
and
then,
if
you,
if
those
moduli,
if
you
line
them
up
nicely,
then
you
know
the
easy
example.
Is
you
know
one
samples
at
ten
to
one
another,
one
samples
at
100
to
run?
Well,
you
know
when
you
get
that
the
rarer
one.
Well,
you
know
the
less
rare
they
they
they
sorry,
the
more
prevalent
one
will
definitely
be
present
because
you,
you
chose
the
moduli.
A
A
A
A
B
A
B
A
The
I
mean
some
of
these
scaling
things
geometric
or
worse.
You
know
like,
as
far
as
the
like
the
coordinating
humans
part
of
thing
is
concerned.
It's
it
was
super
ambitious.
Just
with
the
the
tracing
and
I
was
worried
at
times
we
weren't
going
to
make
it
just
because
it
was
it's
so
hard,
wow,
yeah
logs.
A
What's
the
angle
with
the
metrics,
though,
the
I
get
to
ask
on
behalf
of
the
audience
parts
of
the
audience
who
don't
know
what
the
plan
is
for
metrics?
A
B
A
B
New
new
brainpower
and
new
people
working
on
it
so
hopefully
it'll
come
together
pretty
soon.
Now
that
tracing's
1.0,
okay.
A
Cool
because
there's
this
I
mean
yeah,
like
the
charity,
mages
fans,
would,
you
know,
heard
a
billion
times
that
you
know
in
theory
like
if
you
you
know,
if
you
roll
up
details
of
what
you
just
executed,
you
end
up
with
this.
You
know
a
bunch
of
span
attributes
from
that.
You
can
derive
logging
messages
about
that.
Yeah.
A
C
A
Derive
you
know
metrics
that
just
you
know,
throw
away
most
of
the
detail
and
just
say
look.
I
got
a
hit
yeah
on
the
limited
number
of
tags
that
we're
willing
to
keep
right
and.
A
Or
whatever
it
is
but
yeah.
So
in
theory
you
would
just
you
know,
have
everybody
you
know
going
with
child
span
and
then
hooking
on
derivation
functions,
but
that's
not
how
a
lot
of
programmers
think
I
think
that
way.
But
every
time
I've
tried
to
explain
that
to
somebody
else
they
go
garth.
Will
you
stop
doing
that?
You
know
so
so,
if
they're
just
making
something,
that's
a
lot
more
kind
of
prescriptive,
then
that's
cool.
A
A
B
B
A
proto
buff
for
stack
traces
and
errors,
and
that
was
really
useful
because
none
of
these
back
ends
can
parse
early
stack
traces.
So
having
a
common
way
of
writing
them
is
really
useful.
So
I've
been
kind
of
pushing
that
and
right
now
stack.
Traces
are
just
strings
in
open
telemetry,
but
I
think
the
goal
is
to
have
something
that
covers
that
kind
of
use
case
and
has
structured
stack,
traces
and
stuff
and
there's
exceptions
that
you
can
add
the
spans
right
now.
But.
A
Yeah
I
I
hooked
up
like
the
first
open
telemetry.
I
hooked
up
was
actually
in
python
and
yeah
sitting
in
the
lambda
function,
using
honeycomb's
standard
output
to
honeycomb
scraper.
A
So
you
just
you
just
hook
up
a
a
particular
an
unusual
twist
on
the
console
reporter,
which
is
you
know,
bleating
out,
honeycomb
format.
Events
instead
of
you
know
raw
open,
telemetry
events
as
in
a
kind
of
a
json
format.
It
works
just
fine
and
yeah.
One
of
the
things
I
noticed
was
every
time
I
crash
something
out.
I
I
get
a
little
exception
blob
in
a
honeycomb
like
what
the
hell
is
that
it's
like
turns
out.
Open
telemetry,
has
exceptions
yeah.
A
I
never
knew
yeah,
it's
there,
yeah
so
yeah.
Now
I'm
gonna
have
to
go
back
and
make
sure
that
when
that
that
we
get
the
same
kind
of
an
event
out
of
open
telemetry.
B
I
just
noticed
yeah
I'm
trying
to.
I
want
to
get
record
exception
into
this
zero
six
release
and
just
discovered
that
the
pull
request
fails
because
of
a
bit
of
code.
I.
A
Yeah
so
late
week
after
next
I,
if
my
my
current
sprint,
I
guess,
is
the
an
easy
word
goes
well
that'll,
be
when
I'll
be
going.
You
know
what
I
think
now
is
the
time
I'm
gonna
try
to
pull
that
in
either
that
or
I'll
do
it
in
we're.
Trying
that
base
camp
thing
called
shape
up
it's
a
different
way
of
organizing
your
work.
A
You
know,
instead
of
it
being
driven
by
kind
of
a
breakdown
in
advance
and
estimate,
everything
and
then
get
very
frustrated
and
thump
the
table
a
lot
and
yell
that
engineers
have
to
get
better
at
estimating.
You
know,
because
your
statistical
roll
up
is
indefensible.
The
my
cto
just
looked
at
that
goes.
You
can't
do
that.
It's
order
of
login.
It's
sorry!
It's
log!
It's
it's!
A
A
a
log,
normal
distribution,
just
like
yeah,
because
he's
a
data
scientist,
the
yeah
you've
got
like
these
six
week
cycles
and
you
end
up
with
a
two
week
cool
down
and
that
that's
when
you
pull
things
like
switching
from
open
sensors
to
open
telemetry,
so
I'm
really
looking
forward
to
it
rip
it
out.
You
know
put
another
one
in
so
you'll
you
you
could
reasonably
expect.
If
I
do
do
that,
but
there'll
be
the
usual
flurry
of
pull
requests
which
will
invariant.
B
B
Nice
yeah
yeah
because
yeah
considering
that
pretty
much
release
release
candidate
for
1.0
is
0.6.
So
let's
get
some
bug
reports
on
this
one
and
then
make
a
stable
release.
A
Look,
I
I
don't
think,
there's
other
other
than
you
know
badly,
suggesting
open
error,
capture
and
open
analytics.
A
B
A
So
stoked
to
see
this
documentation
page,
that
is,
that
is
super.
Is
there
a
redirect
from
like
that?
The
path
is
dox
erlang,
but.
B
B
So
all
right,
I'm
gonna
go
get
0.6
for
at
least.