►
From YouTube: 2021-12-14 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
C
I
I
thought
I
was
all
fine
and
good
until
I
realized
this
morning
that
we
have
our
old
sdk
telemetry
product
has
a
log
for
j2
appender.
C
C
All
right,
let
me
pull
up.
I
just
dumped
brain
dumped
random
stuff
onto
the
agenda.
Since
I
haven't
looked
at
some
of
these
pr's
and
then
I
started
looking
at
some
issues.
So
let's
see
there's
a
couple
important
ones
and
then
several
not
as
important
things.
C
This
so.
C
Just
wanted
to
get.
Did
everybody
see
this
already.
C
D
D
B
C
D
D
D
D
A
They
want
yeah
yeah,
but
in
that
case
and
by
default
we're
adhering
to
the
schema
and
user
has
I
mean
this
is
kind
of
a
stretch,
but
user
has
to
set
the
property.
So
it's
kind
of
like
there's
a
user
input
needed
to
enable
these
attributes.
D
C
C
Open,
telemetry
instrumentation
would
use
the
hotel
schema
url
and
we
could
then
have
additional
things
if
we
wanted,
but
those
would
be,
then
we
could
never
ever
change
them.
D
D
C
C
C
D
B
B
D
B
E
C
C
Search
for
this,
I
was
just
looking
up
what
I
need
to
search
by
experimental
span
attributes,
so
this
is
kind
of
our
our
convention
for
everywhere.
So
I
think,
if
we
drill
into
any
of
these.
B
A
C
B
C
Okay,
so
are
people?
Are
we
okay?
I
mean,
if
I
kind
of
just
keep
poking
back,
just
want
people
to
be
really
clear
that
if
my
current
reading
of
this
pr
we
would
release,
we
would
basically
hide
all
instrumentation
behind
an
experimental
flag
and
only
enable
those
instrumentations,
as
the
semantic
conventions
become
stabilized.
D
Gonna
be
using
these
anyways,
so
where
did
we
get?
I
think
so.
I
think
that
even
that
would
be
okay,
so
vendors
are
free
to
do
whatever
they
want,
but
this
rap
was
being
official,
open,
telemetry
repo.
We
probably
have
to
hide
everything,
but
if
we
do
that,
that
means
that
our
agent
is
unusable,
it
doesn't
do
anything
it
doesn't
produce
any
telemetry
and
that
defeats
the
whole
purpose
of
this.
B
So
we
either
defeat
the
purpose
of
the
agent
or
especially
in
our
distress.
We
turn
that
stuff
on
and
then
I
think
we
defeat
the
whole
purpose
of
this
instrumentation
stability
discussion,
because
when
all
the
vendors
are
turning
it
on,
that's
a
bit
too
much
then
what's
stable
and
nothing
would
be
stable
right.
C
So
what
do
you
think
about
once
once
http
semantic
conventions
are
stable?
Does
it
make
sense
at
that
point
right,
we
could
publish
a
java
agent
that
only
captures
http
server
and
http
client
spans.
It's
not
amazing,
but
it's
not
totally
useless.
B
I
was
sort
of
wondering
what
the
point
is
to
hide
instrumentation,
because
we
are
already
providing
it
like.
It
was
supposed
to
be
stable.
We
weren't
supposed
to
be
priming,
providing
in
the
first
place
like
what.
What
is
the
benefit
to
users?
I
guess
if
we
then
disable
just
some
of
the
instrumentation.
B
B
D
D
D
Eventually,
yes,
but
right
now
we
say
that
everything
is
telemetry
is
experimental
or
unstable.
Not
experimental,
unstable
telemetry
is
unstable
because
specification
is
not
fixed,
so
some
changes
can
be
expected.
We
already
say
that
as
soon
as
we
have
first
table
specification,
for
example
http,
then
we
modify
that
sentence
and
say
everything,
except
for
these.
D
C
Mark
yeah
jack
was
kind
of
getting
at
the
same
question
that
I
mean.
D
B
C
Yeah,
is
there
anything
okay,
this
this
makes
sense
to
me
for
sure.
Is
there
anything
worth
exploring
about
the
schema
url,
because
if
I
recall
the
lack
of
schema
url
was
defined
to
mean
that
it
had
it
had
to
be
stable.
If
we
we
couldn't
break
it.
If
it
didn't,
have
a
schema
url,
because
that
meant
there
would
be
no
possibility
for
automatic
schema
conversions.
C
D
C
C
C
Okay,
I
will
anybody.
C
C
If
we
had
http
and
messaging,
which
are
you
know,
sort
of
at
least
have
people
working
on
them
and
targets,
then
it
becomes.
B
C
C
Try
to
understand,
try
to
understand
what
we
can
do
because
currently
his
he
seems
to
really
like
the
idea
that
we
should
be.
He
thinks
we
should
be
hiding
all
unstable,
auto
instrumentation,
behind
an
experimental
flag.
C
Okay,
sure
yeah
yeah
yeah
just
get
as
much
information
as
you
can
try
to
see
where,
where,
if
you
can
help
bridge
a
path
to
anywhere,
because
right
now,
I'm
I
I've
been
talking
around
in
circles,
I'm
back
to
the
to
say.
Basically,
I
think
I
just
copy
pasted
my
yeah.
I
copy
pasted
this
with
my
first
comment,
I,
like
you,
this
is
what
we're
gonna
do
and
we're
back
here
again.
C
Yeah
and
then,
if
you
ping,
on
slack
with
where
it
ended
up,
I
I
can
follow
up
on
this
issue
tomorrow.
If
there's
anything
that
you
think
is
worth
like
summarizing,
I
will
try.
Okay.
C
Okay!
Enough
of
that,
okay,
lots
of
little
things
anything
that
anybody
else
wanted
to
chat
about.
First.
A
Oh
okay,
not.
C
B
Okay,
I
think
that.
C
Route,
I
think
route.
I
don't
know
why
I
say
route
but
as
anna
are
gonna
rely
like
if
it's
like
route
66
like
highway
route,
like
I
say,
root,
yeah.
But
if
it's
a
routing
problem
or
do
you
say,
rooting.
C
A
A
A
So
I
arrived
at
at
using
something
like
server
spending
which
was
similar,
but
anyway
I
it
seemed
to
me
that
the
pattern
of
storing
something
in
the
context
and
updating
in
in
the
middle
and
then
using
the
information
at
the
end
of
the
instrument
recall.
But
this
this
should
be
the
way
to
go
about
it.
A
So
server
span
naming
looked
like
something
that's
rather
similar
to
the
root,
and
my
first
idea
was
to
just
refactor
it
and
to
remove
any
mentions
of
http
root
from
the
start
of
the
span
and
just
have
everything
set
the
route
via
the
server
spending
class.
So
this
pr
actually
does
that.
I
removed
the
removed
setting
the
spam
name
in
the
server
instrumentation
like
setting
that
root
as
spamming,
and
just
they
relied
on
calling
the
servers
by
naming
class
to
update
the
name
which.
A
Yeah
it's
I
mean
I,
I
agree
with
your
comment.
First,
that
it's
kind
of
maybe
a
bit
more
confusing,
because
previously
you
just
had
started
an
end
and
then
as
well.
You
had
the
correct
or
the
correct
spam
name
at
the
very
beginning,
but
right
now
you
also
have
to
remember
to
call
like
server
spending,
but
it
probably
has
to
be
done
anyway,
if,
if
not
by
several
instrumentation
done
by
something
like
spring,
so.
C
C
B
A
A
When
the
like,
when
we
called
server
span,
naming
we
actually
said
two
things
we
set.
The
root
will
be
the
span
name,
which
will
be
then
factored
to
root
and
we
set
the
source
right,
so
we
would
have
to
coordinate
setting
the
source
and
the
root
at
the
of
http
spam
processing
and
yeah.
Here's
that
hypothetical
scenario
that
I
described,
that
what
did
you
have
filter
that
that
didn't
really
find
its
roots?
Then
it
shouldn't
set
the
source
because
it
said
you
get
as
this
fun
name
and
so
on
and
so
on.
A
I
I
I
don't
know
if
this
actually
happens,
but
this
is
kind
of
a
thing
that.
C
I
was
just
thinking
that
library
ins
like
as
the
instrument
instrumenter
api
for
library
users
or,
for
you
know,
manual
instrumenters,
like
in
that
case,
I
could
see
it
being
more
common
to
know
your
route
that
you
want,
because
you're
likely
instrumenting
down
lower
in
your
code.
C
C
C
C
Maybe
we
just
take,
I
mean,
take
it
all
and
take
this
pr
in
as
it
is,
and
if
we
find
I
mean
we
can
always
change
those
two
instrumentations
later
at
least
we
get
most
of
the
structure
we
want
and
we
can
still
think
on
them,
because
I
don't
think,
there's
it's
not
introducing
any.
We
know
we
need
the
that
api
surface
to
do
it
yeah
in
this
manner,.
A
That's
true
anyway,
and
anyway,
like
the
other,
maybe
a
bit
weak
argument
for
removing
the
root
from
the
start
was
that
if
you
have
two
ways
to
set
the
root,
then
it
starts
being
confusing
because
which
one
should
you
use?
It's
not
so
obvious
that
when,
if
like
in
a
situation
when
you
only
have
one
way
to
do
everything.
C
A
But
then
again,
we
probably
won't
run
into
many
situations
like
that,
like
due
to
that
the
two
instrumentations
that
I
mentioned
and
probably
nothing
more,
oh
well
anyway,
that's
that's
one
thing,
the
other
thing
about
this
pr
that
laurie
mentioned.
We
have
some
instrumentations
that
use
service
pandemic
for
setting
things
that
and
http
root.
B
A
A
A
And
well,
with
with
exception
of
grails
they're,
all
like
not
really
hd
instrumentations
grace
grey.
This
is
the
controller
firmware
controller
framework,
but
it's
groovy
and
it
doesn't
really
give
you
access
to
the
roots,
so
it's
impossible
to
determine
the
route
from
it.
E
A
C
Right
yeah
cause,
I
I
mean
that's,
I
don't
know
nikita
honora
if
you
have
thoughts
on
whether
this
these
the
span
names,
whether
the
http
root
needs
to
be
an
actual
http
root
versus
whatever
random
thing
that
we
capture
like
a
controller
method.
Name,
I
mean
that's
like
a.
It
is
a
root.
It's
just,
maybe
not
an
http
route.
A
Yeah-
and
I
mean
it's-
it's
a
very
useful
attribute
for
quickly
finding
what
what
class,
what
piece
of
code
is
responsible
for
the
span.
So
I
mean
that's
like
my
personal
opinion,
but
as
long
as
it
satisfies
the
condition
that
you
can
quickly
found
the
class
from
it,
then
it's
fine.
A
C
D
Right
after
I
interrupt
you
or
reading
the
spec,
you
asked
what
the
difference
between
span
name
and
root.
If
the
root
doesn't
include
the
application
root,
it
should
be
prepended
to
the
span
name.
D
D
D
A
C
B
D
D
A
D
C
C
Okay,
so
that
that
sounds
like
I,
I
tend
to
yeah,
I
agree
also,
but
so,
if
you
can
find
a
way
to
not
capture,
oh
you
were
saying
to
basically
do
span,
update
name
directly
from
those.
C
A
This
sort
of
I
mean
there's,
there's
still
a
lot
of
uncertainty
around
this
topic,
but
I'll
just
try
to
do
it.
Iteratively
and
we'll
see
where
we'll
end
up.
C
Okay,
yeah,
I
I'm
totally
good
with
you
know,
with
the
first
pass.
B
C
A
C
C
Cool,
so
just
ping
me,
you
said
there
was
something
you
were
gonna.
Do
the
the
grails
jax
ws
stuff
still.
C
A
Okay,
so
there's
another
pr
from
I.
I
was
looking
at
this
instrumentation
api
recently
and
I
tried
to
pick
up
pick
up
stuff
that
shouldn't
really
be
in
the
public
api
and
I
was
trying
to
remove
that
and
clean
it
up
a
little
and
it
seems
like
the
spanky
class
is
probably
not
something
that
we
should
expose,
because
it's
kind
of
an
internal
concept,
but
on
the
other
hand
it
needs
to
be
accessible
from
instrumentos
from
the
server
span
class.
C
But
it's
not
well,
let's
see,
maybe
I
could.
D
C
Which
one
this
one
is
internal,
it's
in
or
oh,
I
don't
know,
I'm
not,
it
will
be
yeah.
A
C
A
C
D
A
Because
those
two
methods
are
the
instance
methods
of
of
the
class,
and
you
still
need
to
expose
the
server
spam
key
instance.
So.
C
A
So
the
spike
is
very
thin
wrapper
over
the
context.
Key
that
the
only
thing
it
achieves
is
not
not
exposing
that
to
a
complex
instance.
D
A
Yeah,
so
the
server
span
is
like
one
specific
use
case
of
spankies.
Oh
it's
like.
C
D
A
Yeah,
there
is
one
more
thing
that
I'm
still
wondering
about
whether
we
actually
should
have
the
server
expand
class
at
all
and
whether
shouldn't
we
switch
to
the
local
root
spam
concept,
because
it's
probably
well,
I'm
thinking,
probably
but
it's
kind
of
what
it's
supposed
to
be
used
for
and
like
the
the
spam
suppression
for
http
servers
and
accessing
the
local
route
are
two
concepts
and
we
just
we
currently
merge
it
into
one
class
and.
C
Yeah,
I
I
think
the
the
local
route
is
very
compelling
use
case
and
versus
all
the
yeah.
That's
really
the
only
compelling
use
case
that
I
couldn't
think
of
yeah.
All
the
rest
are
yeah
internal
span,
suppression,
details.
C
Oh
you're
saying
that
a
completely
oh,
that's
true,
because
we
probably
would
want
a
separate
too
much
so
many
context.
B
C
Like
I,
like
the
local
route,
I
mean
it's
a
it's
a
good
concept.
C
Okay,
yeah,
I
think
we're
good
then
here.
A
A
C
Just
turn
on
this
one:
let
me
just
confirm
with
manila
tomorrow:
okay
and
then
I
will,
I
will
go
ahead
and
merge,
as
is.
C
Well,
our
hour
is
up,
but
it's
there.
Let's
see
thing
important.
C
That
one's
not
important
logging
policy.
B
B
B
B
A
A
D
C
What
I
was
wondering
was
so
if
it's
zero
and
we
pass
in
zero
to
end
here,
won't
we
capture
http
dot
status
code
attribute
as
zero.
C
A
C
A
A
C
Thank
you.
Did
I
not
yes,
okay,
all
right,
I
knew
that
one
was
gonna,
be
easy.
C
This
one,
I
think
it
makes
sense
to
me-
I
don't
know
I
might,
but
we
defer
to
the
sdk
for
start
and
end
sometimes,
but
we
also
use
the
time
extractor
for
start
and
end.
Sometimes
oh
and
metrics,
that
was
the
problem,
was
the
metrics
when
we
defer
span
start
time
and
end
time
to
the
sdk,
then
the
metrics.
B
B
B
C
B
D
C
How
do
you,
the
anchored
clock,
is
it's
anchored
to
the
local
root
span
like
it's
passed
down
in
the
so
it's
in
the
context
it's
passed
down.
A
B
C
All
right,
I
will
update
the
issue
with
just
our
discussion
notes.
Just
so
I
remember
when
I
look
at
that
again
why
it's
not
so
easy.
B
E
A
D
B
D
C
You
know
in
the
race
to
everybody
wants
to
ask
who
somebody
commented.
That
was
so.
The
our
old
uses
system,
error
and
file.
Writer-
oh,
like
I,
never
thought
I'd
say
this
was
a
good
thing
and
then
there's
our
problematic
log4j
appender,
which
I
like,
but
it's
in
a
pet
you're
using
blog
for
j2
and
so
anyway
have.
D
B
D
B
C
More
and
more
yeah
people
who
don't
like
to
say
yes
to
things,
who've
been
burnt
many
times
before
for
saying
yes,.