►
From YouTube: 2020-07-02 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
B
A
A
B
A
A
D
B
D
A
C
So
he
left
run
I'm
styling
from
Brazil,
maybe
sometimes
it
to
be
a
little
bit
hard
for
me
to
follow
along,
but
I'll
try
to
follow
along
as
much
as
possible.
My
journey
with
open
telemetry
started
just
yesterday,
I
think-
or
it
was
the
day
before
yesterday,
because
at
work
we're
watching
a
quite
critical
project
and
had
zero
perceiver
bility.
The
first
test
was
really
frustrating
because
things
were
happening
in
the
one
knew
why
and
books
that
don't
happen,
production
not
staging
in
tests.
B
A
C
D
D
B
D
C
B
B
We
were
gonna
talk
about
Obon
and
how
we
can
instrument.
Their
main
thing.
I
wanted
to
say
was
that
if
Obon
doesn't
half
away
right
now
to
attach
some
metadata
to
a
job,
we
should
talk
to
I'm,
pretty
serious,
like
mainly
one
guy
behind
Obon.
You
should
reach
out
to
him
and
get
get
him
behind
what
we
would
need
for
telemetry
to
execute
events
that
contain
this
metadata
and
we're
able
to
use
it
as
a
link
to
the
span
in
the
worker.
B
D
C
D
B
A
Know
I
keep
having
to
fall
back
to
explicitly
painting
passing
around
span
contexts
or
you
know,
trace
headers,
even
if
they're
they're
not
going
on
an
HTTP
transaction,
you
know
like
I'll,
smuggle
them
back
up
as
macer
tagged
in
a
web
page
or
something
just
so
that
I
can
get
the
the
connectivity
down
and
especially
if
you've
got
the
job
getting
created
over
here
with
one
process
and
once
that
our
code
and
then
it's
getting
shoved
into
a
queue.
Yeah.
A
A
B
C
A
In
my
heckinger
round
on
the
on
the
plug
side
of
things
like
I,
seriously,
think
I'm
going
to
end
up
having
to
go
back
to
the
sort
of
open
census,
stop
plug
mechanism
where
you
just
put
yourself
in
the
code
path
as
early
as
you
can
and
then
what
I'll
add
this
time
is
monitoring
the
process,
so
I
could
catch
it
if
it
dies.
Oh
yeah.
B
B
E
C
B
C
B
A
B
Yeah
the
reason
they
use
it,
it's
just
faster
than
a
map
and
want
to
be
as
little
overhead
as
possible
when
doing
this
telemetry
stuff,
then
well,
probably
wouldn't
be
a
noticeable
difference.
Just
using
a
record
and
record
just
Earling
doesn't
have
anything
for
likes
trucks,
so
the
only
way
to
make
sure
we
don't
miss
type,
something
is
to
use
a
record
instead
of
a
map,
a
map.
You
know
if
I
span
ID
without
an
underscore
you'd
be
getting
a
new
map
and
nothing
would
tell
you
so
it's
nice
to
just
use
a
record.
A
A
A
B
D
B
There
will
be
views,
but
it's
also
attempting
to
optimize
for
the
stance,
D
side
people.
So
what
I've
been
arguing
for
is
that
creating
a
instrument
and
recording
to
it
can
have
a
sort
of
a
default
view
and
you
can
have
those
all
enabled
or
disabled
configurative
lis.
And
so
you
create
a
counter
and
you
start
recording
to
it.
And
you
have
the
an
exporter.
It's
gonna
be
pushing
that
those
counts
automatically
without
creating
a
view.
B
B
D
D
Right,
but
we're
still
in
that
situation,
where,
like,
if
you're
not
declaring
what
you
want
upfront,
then
you're
holding
on
to
everything.
That's
one
of
the
problems
of
like
why
I
haven't
why
it's
been
a
long
time
coming
for
me
to
actually
tackle
adding
summary
support,
because
every
all
the
time
series
do
having
to
be
recorded
and
in
order
to
enable
to
support
the
percentiles,
because
if
you
define
to
Windows
like
I,
can't
it's
hard
to
keep
that
data
and
just
operate.
Those
two
windows
off
of
that
same
recorded
set
of
data
in
our
language.
B
B
D
I
guess
it
doesn't
make
a
difference.
I
guess,
you're
doing
the
same
thing
here.
I
will
see
the
way
of
telling
it
what
you
want
tag
wise.
D
F
B
D
B
B
Yeah
I
think
it
has
a
better
chance
actually
working
in
open
telemetry,
because
the
upfront,
the
metrics
you
define
and
use,
are
more
powerful
than
open
census.
Where
it's
you
define
the
metrics
and
they're.
Basically,
nothing
and
then
you
define
the
views,
so
might
be
able
to
map
telemetry
metrics
to
open
telemetry
instruments.
D
Yeah
I
was
just
trying
to
figure
out
like
what
I
would
be
capable
of
doing
with
the
British
Library
yeah
beyond.
Just
saying
like
here,
drop
this
in
and
go
get
some
spans
do
cuz,
there's
also
I,
guess
the
topic
I
was
alluding
to
this
morning
as
well
of
like
how
do
you
tell
in
that
library?
How
would
you
tell
it
like
these?
Are
the
tags
I
want,
because
you're
gonna
have
to
have
an
upfront
knowledge
of?
D
What's
in
the
events,
which
is
fine,
like
I,
think
that
that's
totally
good,
like
I,
think
that,
like
for
most
of
these
libraries,
that
I
just
thought-
oh,
this
is
kind
of
what
I
was
alluding
to
earlier
is
like.
Most
of
these
libraries
are
gonna,
be
black
box,
except
for
using
telemetry
events,
and
so
we
have
to
rely
on
like
the
metadata
and
stuff
that's
in
there
to
operate
off
of
to
like
kind
of
wire.
These
things
up
together.
So
I
am.
We
are
gonna,
have
to
figure
out
some
mechanism
and.
D
Like
you
could
continually
just
say,
like
yeah,
here's,
the
application
like
span
like
create
spans
for
everything
that's
available
and
then
out-of-the-box
like
you.
Just
by
default,
you'll
get
durations
bands
on
all
those
things,
and
then
you
could
pass
in
something
that
you
would
be
able
to
say
like
okay,
for
this
particular
event,
though,
I
also
want
these
tags
and
then
I
can
just
map
those
up
together.
D
So
that
it'll
add
those
because
I
also
want
to
get
people
like
the
same
thing
that
we
have
I
think
would
be
cool
to
support
the
same
things
that
we
have
in
the
metrics
definitions.
No,
having
like
the
transforms
and
being
able
to
like
provide
functions
that,
like
deeply
pull
out
the
things
that
you
want
and
reshape
that
map,
and
this
into
the
tags
and
names
that
you
want
really
I,
think
that's
a
powerful
functionality
that
I
would
like
to
heat.
D
A
C
A
A
So
the
you
know,
Here
I,
am
writing
the
same
kind
of
code
to
pull
attributes
out
of,
like
all
of
the
semantic
attributes
for
open,
trace
out
of
Tilsit,
Tesla
environments
and
plugs
and
so
play
cons
and
so
on
and
so
forth
and
yeah.
The
the
more
library
you
can
bring
for
that.
The
the
better
like
I'm
I
can't
be
the
only
person
trying
to
wire
up
Hackney
style.
D
Yeah
and
partnerships
in
any
conditions
to
you
right,
because
we're
trying
to
map
to
like
hotel
specific
things
and
like
everybody
like
every
issue
to
be
client
like
calls,
are
put
status
or
something
in
a
different
spot,
and
we
have
to
worry
fry.
We're
gonna
have
to
do
that.
Mapping
up
anyway,
so
I'd
like
to
figure
out
a
way
to
make
that
easier.
I
didn't.
D
Constraints
that
you
want,
based
on
like
the
type
of
requester,
where
it's
going
so
like
if
I
have
a
client
that
is
talking
to
like
one
service
and
I'm,
expecting
like
five
millisecond
deadlines
on
that.
But
then
also
this
service
happens
to
have
like
oh
I'm,
also
like
making
calls
and
downloading
like
100
Meg's
I
want
those
like
distribution
buckets
to
be
drastically
different
between
the
two
and
I
might
want
like
totally
different
tags
and
stuff,
and
so
it
was
a
way
to
get
around.
A
A
A
D
Like
if
you're
used
to
grams,
basically
like
it
records
everything,
and
then
when
you
tell
it
to
like,
you,
asked
for
like
the
p99,
but
it
like
is
able
to
efficiently
be
like
a
sliding
window
of
time
to
be
able
to,
like
you
could
say,
like
over
five
minutes
and
it
like.
Does
it
really
efficiently,
yeah.
A
Gonna
get
there,
you
know
it's
a
bit
like
dynamic
sampling
like
I'm,
totally
gonna
get
there,
but
every
time
I
get
close
to
needing
it.
Something
changes
like
honeycomb,
saying,
hey,
tell
you
what
we'll
give
you
two
months
as
long
as
you
don't
hit
twenty
million
events
and
I'm
like
well.
That's
not
gonna
happen
this
week,
so
I
guess
I'm
not
dynamic
sampling.
This
week,
I'll
just
keep
shoving
out
spans
one
to
one
one.
D
Just
you
just
basically
call
and
apply
on
it
with
the
mesh
with
the
measurements
and
then
there's
the
keep
drop
thing.
I,
don't
know
if
that's
going
to
be
pertinent,
like
it
could
be
like
in
a
sampling,
I
guess,
but
the
tag,
tags
and
tags
values,
though,
is
like
the
big
thing,
because
the
tag
values
lets
you
produce
a
function
that
accepts
the
metadata,
and
then
you
can
do
whatever
you
want
to
return.
D
A
D
B
B
I'm,
hoping
is
well
I,
think
the
main
thing
is
gonna
be
first
or
the
biggest
part
is
gonna,
be
understanding
it
I,
don't
think
the
initial
code
is
gonna,
be
too
bad.
Once
I
wrap
my
head
around
what
all
they're
the
concepts
are
so
like
we
probably
yeah
I
can
speak
better
about
it.
Next
week
after
I've
gone
through
this
bag,.
B
D
Think
we
need
to
like
boil
the
example
like
if
it's
Phoenix
like
it
just
needs
to
be
strictly
like
Phoenix,
with
open
telemetry
installed
and
that's
an
example
of
how
its
wired
up
and
then
we
can
do
that
separately
for
like
each
concept,
but
I'm
really
hesitant
on
about
you,
Tristan
of
like
adding
like
all
these
third-party
libraries
that
are
like
more
subjective
things
of
like
I,
like
this
library
and
I,
think
it's
good.
But
it
doesn't
actually
matter
to
like
open
telemetry
itself
and
providing
an
example
of
it.
A
Does
that
make
sense,
yeah
that
that
was
part
of
what
made
it
a
mess
they
like
with
it?
Shawn
I
had
added
my
internal
Greg
had
added
vapor
too,
which
is
like
a
norm
driven
configuration
thing
to
make
it
easier
to
dynamically
suck
environment,
varial
variables,
out
of
either
the
environment
or
a
dot
in
file
or
whatever,
and
that
was
I
mean
it
wasn't
just
for
what
I
saw
Z's
had
called
shits
and
giggles.
It
was
to
try.
E
A
A
The
command
line,
but
you're
right
it
ended
up
sort
of
it
was
hard
to
distinguish
what's
what's
actually
important
for
telemetry,
and
why
and
what
isn't?
And
and
so
I
found
myself
when
I
was
copying,
that
to
the
plug
gateway,
which
had
not
yet
been
instrument,
and
so
I
was
concentrating
my
efforts
there,
which
is
why
I
keep
rubbing
up
against
Tristan's
work.
A
The
I
was
copying
all
of
that
stuff
in
and
then
having
feeling
the
that
I
had
to
write
a
lot
of
documentation
to
explain
which
parts
we
intended
people
to
copy
in
which
parts
to
not
yeah
so
yeah
I
I
think
it
would
be
better.
If
we
just
leave
out
the
leave
out,
the
vapor
find
other
way
of
solve
for
the
problems.
A
D
A
And-
and
he
was
had
this
wonderfully
ambitious
idea-
I
loved
it
of
deliberately
rebasing
so
that,
even
if
we
had
to
kind
of
go
back
in
time
to
update
the
versions
of
the
libraries
that
we
would
kind
of
move
the
commit
to
the
before
picture
so
that
there
was
still
a
nice
sharp
dividing
line.
As
of
this
commit,
there
is
telemetry
in
there
right
wow.
That
sounds
like
an
awesome
lot
of
fun.
D
B
D
D
B
B
A
C
D
Let
us
know
I,
guess
I
think
that's
one
of
things
we
want
to
see
is
like
start
getting
feedback
and
getting
word
out
as
to
who
is
actually
starting
to
use
it
in
production
and
how
you're
feeling
about
it.
Stability
and
all
that,
because
yeah
like
like
I,
would
like
to
start
rolling
it
out
on
some
things
as
well.
So
everybody's
feedback
is
awesome.