►
From YouTube: 2020-08-20 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
Hey,
I
guess
we
can
get
started.
Probably
everybody
for
today's.
My
updates
and
agenda
items
are
just
that.
I
haven't
got
time
to
do
this,
I'm
supposed
to
do
which
is
release
rebar
3.
So
we
can
merge
the
repos
and
fix
the
context
api.
A
A
B
Yeah,
so
I
was
looking
for
any
guides
or
anything
at
all
about
good
practices
on
contributing
with
package
that
instrument
libraries.
There
is
a
small
guide
on
the
opening
to
limit
specification,
but
I
believe
it
just
says
about
naming
it
doesn't
say
much
about
anything
else,
and
I
see
that
we
have
on
opening
telemetry
being
organization
lots
of
packages
that
do
just
that.
So
I
want
to
know
if
there
is
any
anything
that
I
have
to
follow.
If
I
want
to
contribute
to
that
organization.
A
I
mean
I,
I
guess
the
main
thing
is
nate.
I
mean
naming
it's
not
that
important,
but
we
do
kind
of
follow
a
convention
of
open
telemetry
before
the
application
name.
Besides
that,
no,
I
would
be
happy
to
er
if
you
have-
or
you
start
an
instrumentation
library
and
then
want
to
move
it
over
to
the
github
organization.
B
Okay,
so
there's
another
thing
that
I
want
to
to
talk
about
if
you
want
to
create
a
new
hp,
extractor
or
injector,
to
support
other
vendors.
Is
there
any
other
any
guides
about
this,
and
would
it
go
on
the
open,
telemetry
being
organization,
or
is
it
something
that
should
be
separated.
A
B
Okay-
and
the
last
thing
I
want
to
talk
about
is
that
I
haven't
really
seen
any
discussion
on
trace
fields
on
open
telemetry,
or
at
least
I
could
be
looking
up
for
the
wrong
keywords.
So
I
got
didn't
get
any
reads.
Is
it
fine
to
to
add
a
new
feature?
That's
not
in
the
specification
in
the
open
parameter,
api,
open,
kilometer
length,
api.
A
Yeah
I
mean
it
depends
on
what
is
meant
by
feature
part
of
the
reason
for
spam.
Processors
is
so
we
can
do
stuff
like
that
like
if
you
need
to
have
something
happen
when.
B
So
there
is
about
this
spam
processor.
I
I
really
read
about
it
yesterday
and
in
the
specification
it
at
least
for
me.
I
understood
something
that
should
go
before
this
pain
exporter
and
not
really
something
that
should
happens
on
spanx
so
that
you
can
outrage
spend
so
maybe
I
just
didn't
quite
get
it.
A
There's
a
start
processor
that
you
that
will
run
on
the
span
when
it
started
you're
trying
you're
wanting
to
do
some
modification
to
the
span
when
it's
and
it
yeah
it's
doing
it
at
start
time
acceptable.
You
can
also
do
it
at
end
yeah
actually,
but
because
it'll
just
pass
through
it
and
then
you'll
pass
it
to
the
exporter.
B
So,
at
least
for
this
trace
field
feature
that
I
have
in
mind.
I
think
processor
is
a
good
fit.
If
I
understood
correctly,
okay
makes.
D
A
B
Yeah,
so
there
is
at
least
in
honeycomb:
there
is
a
function
in
their
sdk.
That's
called
add
field
to
trace
that
when
you
call
that
it
will
at
the
field
that
you
want
to
add
to
all
trades
all
spans
inside
the
trace.
So,
for
example,
you
started
hp
request
and
it's
been
traced,
but
just
in
the
at
the
end
of
that
requires
you'll
get
the
user.
You
authenticate
the
user
or
anything
like
that.
B
You
can
add
a
few
to
trace
and
it's
it
will
read
proactively
affect
all
the
spans,
but
I
think,
with
its
main
processor,
it's
possible
to
do
that
on
spain,
starch
I
can
see
if
there
is
any
kind
of
trace
field
and
but
on
its
stable
or
anything
like
that,
and
then,
when
spain
ends,
I
can
see
if
there
is
anything
new
and
just
merge.
I
think
yeah.
E
A
B
A
Be
I
mean,
I
guess
if
you
pipeline
did
pipeline
export,
because
you
want
to
want
to
write
your
own
exporter
if
you
want
to
go
to
the
open,
telemetry
collector
and
then
the
honeycomb
so
because
then
you'd
be
duplicating
so
probably,
but
also
just
the
function.
That
did
this
would
make
sense.
I
mean
as
a
separate
librarian
feature,
because.
B
A
B
B
D
B
A
Yeah
and
there
might
be
a
correlation
context-
slash
baggage,
future
feature
that
makes
this
easy
as
well,
because
baggage
is
where
you
would
carry
something
like
the
user
id,
but
there
isn't
currently
a
way
to
like
force
its
values
into
a
span
as
attributes,
but
that
will
come
eventually.
B
So
there
is
their
correlation
context
is.
I
think
this
is
just
what
I
should
be
looking
for.
Let
me
see.
A
So
I
mean
that's
kind
of
a
good
case
of
because
it's
also
supposed
to
be
that
the
correlation
context
of
a
child.
So,
if
you
add
it,
add
the
user
id
to
a
correlation
context
in
a
child
when
it
goes
back
to
the
parent,
it's
supposed
to
go
back
to
the
previous
context,
and
it
would
lose
that
it's
kind
of
a
good
example
of
life
shouldn't
lose
that
and
if
it
kept
it,
then
it
would
be
able
to
add
it
to
this
band
when
it
ends.
B
A
There
yeah
it
would
be
yeah
an
issue
for
that
on
the
specification.
Repo
would
be
great
mentioning
specifically
the
honeycomb
feature.
A
I'd
like
to
do
it
that
they're
all
in
the
same
repo
with
the
sdk
and
api,
so
that
you
can
just
build
once
okay
and
allies
once
I
think
it
would
make
it
a
lot
simpler.
I
think
that's
something
that.
E
A
Yeah
right
now
it's
a
major
pain,
updating
and
forget
we're
not
updating
because
something
changes,
but
it
doesn't
get
incorporated
into
the
other
libraries
and
if
they
were
all
in
the
same
repo
yeah,
it
would
yeah
literally
I'd
much
prefer
that.
E
I
believe
is
we're
not
setting
a
span
attribute
of
error,
true,
and
it
seems
like,
even
though
it's
not
part
of
the
spec,
whatever
that,
like
every
implementation
and
vendor
like
relies
on
error,
true
to
marx
fans
as
aired
they
don't
and
then,
because
the
exceptions
are
supposed
to
go
into
events
and
right,
I
can't
say
that,
like
lightstep
is,
is
actually
using
that
now
I
did
notice
yesterday.
That
inception
did
show
up
as
a
log
on
a
span.
So
that
was
good,
but
it
won't.
E
E
Because
the
spec
says
you
have
to
put
the
exception
stuff
into
an
event
right,
zipkin
and
jager
aren't
supporting
that
at
all.
Right
now
like
as
far
as
I
know
like,
because
they
don't
have
the
concept
of
I
think
of
like
logs
necessarily
like
on
spans
or
something
or
maybe
the
exporters
isn't
working.
But.
A
E
A
E
Of
error,
true,
like
nobody's
handling
like
the
span
status,
thing
right
now,
so,
okay,
so
maybe
it's
one
of
those
things
I
don't
want
to
like
rely
on
long
term
but
like
in
the
interim,
like,
I
think
everybody
would
expect.
Like.
Oh
yeah,
when
this
thing
exports
hey
it's
going
to
show
up
as
being
red
but
sounds
like
something
that
would
be.
A
Yeah
that
would
work
since
it
doesn't
sound
like
something
that
will
stay
around
forever,
because
it's,
I
hope,
not,
I
hope
everybody
just
likes
it
worked
out.
Yeah
I
mean
they're
supposed
to
be
using
the
I
mean
really
the
their
exporter,
and
the
collector
should
be
doing
this.
So
maybe
I
should
open
an
issue
there
about
this
and
we
can
discuss
with
lightstep
and
people
why
they're
not
looking
yeah
I'll
set
up
lights
up
too
I'll
hold
up.
Okay,.
A
C
A
C
E
E
E
B
B
I'm
using
it
there
is,
there
is,
has
there
still
isn't
a
package,
but
I
just
copy
paste
it
to
my
code
base
and
add
that
disclaimer
to
use
the
package
and
once
we
once
you
release
and
there's
some
modification
we
did
so
maybe
we
can
talk
about
that
if
you
want
to
talk
about
now,
I'm
free,
but
if
you
want
to
talk
about
this
later
over
is
like
I
can
talk
about
what
I
change
it
change
it.
I
think
yeah
yeah,
it's
changed.
B
B
So
brian,
I
didn't
hear
you,
you
said
that
you
go
want
to
go
over
a
slack
or
or
here
no.
B
B
Okay,
so
there
is
the
telemetry
events
for
a
router
patch
and
endpoints
patch,
those
I'm
using,
but
I'm
using
in
hand
by
hand
with
the
phoenix
instrumenter
library
that
we
created.
I
think
that
we
created,
then
I
don't
quite
remember
I
mean
there's
it's
not
the
package,
so
I
think
we
did
in
this
phoenix
instrument
instrumentor
one.
We
also
instrument
the
controller
render
call.
B
There
is
a
controller
call
events,
but
I
think
that
the
end
points
patch
and
holders
batch
is
almost
the
same.
So
we
just
instrument
the
affinity
controller
render
because
it
comes
with
more
information
about
the
view
being
rendered
the
template,
the
format
and
you
can
get
things
like
routes
info,
but
I
think
that
in
the
end
point
especially
us
can
have
that,
and
there
is
another
thing
that
we
use
together
with
those
two,
your
package,
the
phoenix
renter
and
doing
names.
B
Third,
one,
that's
a
plug,
because
the
thanks
stuff,
it
does
not
fire
if
there
is
a
plug
before
the
endpoint
call.
That
just
holds
the
the
call
the
request.
For
example,
if
you
have
a
pipeline
with
authorization
and
authentication
and
that
returns
a
unauthorized,
your
open
intelligence
won't
fire
anything
any
kind
of
spam.
B
So
we
use
the
plug
to
cover
that
case.
So
right
now
we
are
using
three
things
to
instrument
phoenix.
True,
true
only
sorry,
sometimes
I
try
to
say
a
word
that
I
don't
know
how
to
pronounce
it
to
instrument
things
as
a
whole.
It's
the
plug,
it's
the
fenix,
instrumentate
instrumenter
and
then
your
package,
it's
the
three
together.
E
C
E
So
the
last
one
you
mentioned
like
there's
nothing
that
we
can
do
as
far
as
like
quote-unquote
phoenix
is
concerned,
because
it's
more
that's
more
of
like
it's
happening
at
a
point
that
phoenix
doesn't.
E
So
of
the
issues
is
that
they
don't,
and
I
I
understand
the
reasoning
why,
but
they
don't
want
jose,
didn't
want
to
fire
plum
tree
events
until
a
you've
gone
through
static
assets
and
b
that
a
route
has
actually
been
matched
because
it
could
cause
a
lot
of
noise
and
there's
also
like
not
a
lot
of
ways
to
filter
things
right
now,
so
I
can
continue
to
stand
behind
his
position
on
that
and
not
moving
it
higher
up
and
the
way
that
we've
kind
of
looked
at
addressing
that
too,
is
we
want
to
handle
that
more
in
the
actual
handler.
E
E
E
What
all
we
would
do
is
just
the
same
way
that,
like
the
way
it
is
right
now
is
like
the
router
or
that
sorry,
the
endpoint
start
event
happens,
and
at
that
point
like
we
don't
have
router
information
and
plug
so
like
in
the
plug
router.
You
do
like
when
you
hit
past
that
point,
plug
router
has
all
that
information
hydrated,
but
because
phoenix
does
things
differently,
the
routing
piece
happens
at
a
different
spot.
E
So
what
I
was
trying
to
go
for
was
like
each
phoenix
event
that
happens
in
the
in
their
chain
of
events
is
just
adding
more
information
to
to
the
span
and
then
it
in
the
case
of
like
the
name
like
so
for
the
name
like
the
only
thing
I
can
do
for
the
beginning
is
just
to
say
like
it's,
you
know
like
an
http
and
then
whatever
the
method
is
and
then
when
we
hit
an
actual
route.
E
I
re
I'm
currently
replacing
the
name
with
the
tokenized
version
of
the
route
and
I'm
probably
actually
going
to
change
that
at
some
point
soon
to
still
keep
the
http
and
the
verb
and
then
adding
the
tokenized
route
just
append
that
onto
the
end
of
it,
because
it's
making
our
rats
a
little
our
graphs
a
little
weird.
So
we
can
add
the
controller
stuff.
E
E
E
E
E
The
big
thing
is
like
extracting
the
trace
context
from
the
headers
and
then
for
it
to
continue
forward
like
you're
gonna
have
to
manually
add
it
to
the
socket
state
regardless,
so
that
on
the
next
channel
event,
when
you
do
a
handle,
you're
gonna
have
to
like
take
that
off
the
socket,
and
so
I
think,
there's
some
convenience
helpers
that
I
can
make
there.
E
But
I
don't
think
that,
like
wrapping
into
any
of
the
events
is
actually
gonna
happen,
I
think
you're
gonna
have
to
like,
in
your
socket
and
in
your
channels,
you're
just
gonna
have
to
go
in
and
like
write,
write,
something
either
way
so
so
yeah.
I
don't
know
that.
There's
a
lot
more
to
go
on
it.
If
you
want
to
do
you
want
to
open
like
an
issue
or
pr
just
adding
a
handle,
a
telemetry
handler
for
the
controller
stuff,
the
control
review
stuff
that
you
want,
like
which
attributes
you
want
to
add.
B
E
We
removed
it
in
like
one
when
we
switched
the
telemetry.
The
phoenix
instrumentation
hooks
were
deleted.
E
B
Oh
god,
we
used
that.
I
think
that
let
me
see
if
there
is
a
telemetry
event
for
the
render,
at
least
because
I
don't
remember
there
have
been
any
kind
of
telemetry
events
for
a
render
before
I
created
that.
E
On
endpoint
stop,
we
do
have
whatever
information
was
available
and
I
believe
on
router's
start.
It
will
tell
you
at
least
the
controller
and
the
action.
E
B
Yeah,
I
don't
know
if
at
least
I'm
not
finding
the
telemetry
events
on
the
documentation
that
talks
about
templates
and
things
like
that,
but
if
they
remove
it
from
the
this
meter,
maybe
they
have
that
as
a
metadata
on
the
telemetry.
It's.
E
E
They
all
use
the
exact
same
code.
So
literally,
all
that
happens
is
like
when
you,
when
it
does
like
logger
info
like
on
the
request
or
something
it's
just
taking
like
all.
If
you
look
in
the
code
like
all
the
same
metadata
is
right
there
and
then
it
just
says
like
log
this
and
then
like
emit
this
as
a
telemetry
of
it.
But
it's
all
the
same
stuff
and
ecto
does
the
same
thing.
So.
E
I
just
don't
know
where
it
is
because
I
only
do
apis
so
like
there's,
not
a
lot
of
like
views
that
get
rendered
for
me,
but
we
can
totally
add
that
as
attributes
there's
some
configurable
things
that
the
big
thing
for
that
library
that's
gonna
need,
I
honestly
is
it's.
Gonna
need
an
initialization.
E
Allow
users
to
pass
parameters
on
initialization
as
to
what
other
attributes
they
want,
because
I'm
not
going
to
add
like
every
attribute
just
randomly,
because
we
can.
E
E
B
Me
too
I'll
create
the
issue
later
today,
so
we
can
talk
on
github
about
this.