►
From YouTube: 2023-01-24 meeting
Description
cncf-opentelemetry@cncf.io's Personal Meeting Room
A
I
heard
you
were
Stateside
recently.
A
Did
you
graduate
from
it
or
are
you
just
started
no.
A
B
A
Hey
Alex,
hey
Kamal,
how's
it
going.
C
A
Yeah,
we'll
probably
just
wait
a
few
more
minutes
just
to
see
if
people
keep
trickling
in
I.
Think
if
you
haven't
yet
already
there's
an
attached
meeting
notes
and
you
can
add
yourself
to
the
attendees
if
you
haven't
already
and
then,
if
you
have
anything,
you
want
to
talk
about,
add
it
to
the
agenda.
D
A
So
Eden
I
know
that
dinesha
said
last
time
that
he
was
going
to
reach
out
to
you
about
potentially
doing
a
code
review
in
this
meeting.
Are
you
up
for
it
or
is
that
not
something
we
should
be
adding
to
the
agenda.
B
Like
yeah
yeah
I
think
we
can
do
it
I
think
like
like.
Maybe
it
will
be
more
productive
to
like
to
work
a
couple
of
points
here
that
I
faced
when
like
like
the
bad
news,
is
that
I
still
don't
know?
What's
the
reason
for
the
all
the
issues
at
the
the
next
place,
but
I
discovered
a
lot
of
other
interesting
things
that
are
also
worth
noticing.
So
maybe,
if
we
will
go
through
them,
it
would
be
like
more
productive
than
like
just
go
line
by
line.
B
Something
like
that
and
like.
If
we
have
additional
questions
like
I'll
just
do
the
workflow
yeah.
A
Sure,
okay,
so
yeah,
that
sounds
good.
We
can
jump
in
I,
don't
know
Eden.
If
you
wanted
to
share
I
could
share
my
screen
as
well.
Yeah.
D
B
Okay,
so
it's
the
that's
the
issues
of
the
national
account,
so
I'm
still
not
really
sure
what
is
coding
but
I,
while
trying
to
reproduce
I
I
faced
the
couple's
problems
that
some
of
them
I
knew
about,
but
I
think
it
will
be
interesting
to
share
those
with
the
group.
So,
like
the
first
issue,
slash
something
that
is
not
implemented
yet
is
this
we
have
this
offset
results
the
Json
file,
which
I
don't
know.
B
If
you
remember
this
is
basically
this
is
our
way
to
support
multiple
versions
of
app
installs
libraries
that
we
are
instrumenting.
So
basically,
when
we
write
instrumentation,
we
usually
want
to
access.
Let's
say
we
want
to
to
instrument
the
handle
stream
function,
so
we
want
to
access
like
the
tab
or
URL
or
some
or
some
field
on
the
server
transport
struct,
and
because,
like
this
truck
is,
is
part
of
the
grpc
package.
It
might
change
over
over
different
versions,
so
we
need
to
to
have
a
stable
way
to
to
access
variables.
B
Probability
DF.
That
is
like
agnostic
to
the
to
the
version
that
the
target
process
is
using.
So,
for
for
being
able
to
do
that,
we
have
this
file,
which
is
basically
a
translation.
You
can
see
that
it's
basically
invest
on
this
other
project
that
I'm
not
sure
if,
if
it's
part
of
the
vehicle,
yet,
if
not
I
will
definitely
add
it.
D
B
Yeah
somebody
's,
maybe
just
stop
directory
for
now
or
like
that-
will
have
also
this
code
included.
But
basically,
what's
this
simple
program
do
and
like
this
also
can
be
improved
by
a
lot?
Is
that
you
specify
different
things
that
you
want
to
track
basically
struck
and
Fields?
So
let's
say
you
want
to
show
you
we
want
to
access
from
ebpf,
let's
say
to
access
the
the
transport
stream
yeah,
okay,
and
we
want
to
to
access
this.
B
B
I
think
the
semantic
convention
says
like
span
that
created
for
HTTP
travels
need
to
contain
this
method
field
with
attribute
of
I.
Don't
remember
the
name
but
probably
http.net,
something
like
that.
So
we
want
to
access
it
from
pdpf
and
to
stay
agnostic
to
the
library
version.
We
need
to
track
it
so
the
ways
it
happens.
We
have
this
offset
tracker
project
I.
Basically,
just
I
have
to
point
out
what
is
the
full
package
name
that
I
want
that
this
truck
is
included
in
and
like
what
is
the
net
full
name
of
destructed?
B
What
is
the
field
that
I'm
interested
in
and
once
a
I
will
run?
This
just
go
run
main.go.
It
will
generate
this
of
the
offset
some
opposite
results,
not
in
j45
and
like
as
you
can
see.
There
are
a
lot
of
duplications
here.
I
think
we
can
that's
a
different
discussion,
but
we
can
probably
think
of
a
better
phone
and
it
represent
some
of
this
opposite
like
that's.
B
Not
there
very
efficient
way
to
represent
them,
maybe
I,
don't
know
something
with
intervals
or
like
version
range
or
something
like
that,
but
the
problem
with
this
whole
mechanism,
which
is
which
works
pretty
well
by
the
way
if
the
I
tested
it
on
multiple,
like
libraries
that
actually
changed
the
way
that
the
things
that
gopc
changed
like
the
the.
B
Struck,
they
moved
the
position
of
them
and
it
actually
worked
the
the
right
offset
and
this
whole
didn't
work,
but
the
problem
with
that
is
that
it
have
to
be
updated
every
time,
a
new
version
of
of
any
of
the
open
source,
libraries
that
were
on
instrumenting
or
even
the
standard
libraries,
because
they're
also
doing
it
for
Netflix
HTTP,
and
it's
changing
so
right
now.
It's
something
that
I
just
run
periodically
manually
and
just
update
the
office,
the
dropbox
at
the
Json
file,
but
we
can
definitely
build
some
automation
around
it.
B
Pigment
even
like
something
basic
is
just
a
crown
job
that
updated
like
every
day
or
every
week
and
automatically
open
support
request.
Without
so
we.
So
when
a
new
version
of
grpc
of
go
comes
out,
we
will
automatically
be
able
to
support
it
and,
like
I,
think
the
next
version
of
it
will
be
to
actually
be
able
to
download
this
file
from
the
network
or
maybe
think
about
a
mechanism
that
will
be
able
to
to
support
new
versions
of
Open
Source
libraries
without
having
to
add
the
user
to
update
the
instrumentation
application.
B
This
notation
process,
I
hope
you
guys
are
following
like
no
one
really
said
anything
but.
E
B
Yeah
sure
so
there
is
there's
this
thing
in
in
binaries
in
binary
files
called
dwarf,
dwa.
B
Dwy,
which
the
go
compilers
add
to
the
binary
by
default,
is
basically
not
that
yeah
this
one,
which
is
basically
mostly
debugging
information
and
like
this,
this
format
is
something
that
you
can
read
from
a
compiled
binary
and
basically
understand
where
the
exact
offset
of
the
of
the
different
structs
are
located.
B
But
the
problem
with
that
is
that,
like
many
grow,
applications
choose
to
to
remove
it
because
it
it
reduced
the
binary
Side
by
a
lot.
You
probably
know
this
by
combining
with
your
build
the
Shelby
Flags
I,
don't
remember
the
executive,
the
S
Dash
W,
or
something
like
that.
It's
like
it's
pretty
common
practice
to
to
reduce
the
goal.
B
The
go
binary
size.
So
we
don't
want
this
to
be
a
requirement
for
the
applications,
because
again
many
applications
does
not
have
this
dual
information.
So
what's
this
whole
offset
tracker
project
do
is
basically
and
show
you.
It
forgot,
forgo
standard
library
and
go
in
general.
You
just
download
the
latest
Go
version
from
there
is
where
they
upload
any
new
Go
version.
So
we
basically
no
no
I,
don't
know,
go
1.19.2.
B
Yeah
so
I
have
this
like
dummy
applications.
It
doesn't
do
anything
so
compiles
it
with
this
relevant
Go
version
and,
of
course,
with
dwarf
symbols,
and
it
is
all
those
debug
symbols
and
then
it
passes
the
the
output
binary
and
like
know
how
to
pass
this
dwarf
information
and
and
basically
stay
within
this
Json
file
and
the
cool
thing
about
it
is
that,
like
this,
actually
applies
to
also
to
binaries
without
the
Dual
information
so
like.
This
is
how
we
walk
around
these
requirements
to
include
debug
symbols.
B
Mysql
driver
or
something
like
this
or
some
offensive
library
that
you
want
to
instrument.
It
will
do
the
same.
It
will
just
replace
this
has
picked
the
results
of
this
mod
file.
Here
yeah,
it
will
just
like
detect,
which
version
of
the
library
I
want
to
do.
I
want
to
find
the
offsets
it
will
replace
those
placeholders.
B
We
will
compile
a
go
application
with
dwarf
symbols
for
this
specific
Library
past
the
dwarf
symbols
and
outputs
into
the
Json
files
so
like
to
slum
up
all
this
and
I
could
think
that
I
discovered
during
trying
to
produce
the
back
of
the
natural
powder.
Is
that,
like,
like
we
missed
a
couple
of
versions
and
like
the
latest
goal
version
is
still
not
updated
here.
B
So
I
simply
ran
this
this
program
again
and
got
all
the
latest
updates
and
like,
as
you
can
see,
we
can
probably
guess
what
the
opposite
will
be
because
they
are
not
changing
by
much.
You
know
stay
the
same
and
yeah
so
like
that's
one
issue
and
I.
Think
it's
like
like
by
trying
to
debug
this
specific
thing,
like
you
learned
about
how
this
entire
offset
thing
work.
B
Does
you
have
someone,
has
more
questions?
Can
I
go
to
the
next
one.
F
D
F
B
Yeah
I
think
it
does
not
require
the
gold
pistilan
thing
and
but
I'm
not
really
sure
I
have
to
check
I.
Think
I
think,
like
this
time
mechanism
also
work
for
this,
like
I,
definitely
tested
it
with
binaries.
That
does
not
that
are
compiled
with
these
subject
lines
plugs
for
removing
the
Dual
information.
B
I
do
know
that,
like
there
is
additional
things
that
you
can
do
to
go
binary
through
business
size,
even
even
more,
if
you
know
that
you
can
do
like
some
creeping
of
all
the
strings
and
stuff
like
that
that
basically
even
like,
if
you
have
a
panic,
so
you
want
to
even
have
been
able
to
see
the
stock
price
and
the
different
functioners,
because
the
strip
but
I
think
that
this
is
less
common
than
just
like
compiling
without
debug
symbols.
B
F
We
are
like
maybe
just
give
two
bit
of
a
context.
We
are
also
building
an
evpf
space
profiler.
So
that's
why
I
asked
that
question.
I
know
that
there
there
are
information
about
like
strike
unwinding
in
there
and
go
run
time
itself.
It's
using
that.
So,
as
you
told
like,
whenever
a
panic
happens
like
that's,
why
I
asked
like
four
normal
operations,
like
you,
wouldn't
actually
remove
that
section?
Maybe
you
can
remove
the
divorce,
but
not
the
PC
Line
tab.
F
That's
why
I
asked
maybe
it
could
could
have
been
used
if
you
ever
tried
that
or
not
it's
a
large
section
there,
there
are
different
informations
in
there.
So
I
also
don't
know
if
you
have
the
necessary
offset
information
in
that
yeah.
F
B
Yeah
great
so
I'll
take
this
as
an
action.
I
can
tell
you
it's
not
better,
but
I
think
I
looked
on
it,
but.
B
Okay,
so
another
thing
that
I
discovered
is
like
and
I
think
this
is
very
much
demonstrated
why
it's
so
important
to
choose
a
stable
function
when
we
instrument,
so
just
as
a
reminder
whenever
we
write
a
new,
instrumentation
and
I.
Think
that's
like
something
that
most
of
God
instrumentation
in
the
other
in
other
programming
languages,
also
that
you
have
to
to
attach
yourself
of
the
hook
into
some
functions
and
like
like
either
do
some
modification
for
context,
propagation
or
create
fans.
B
You
have
to
do
some
sort
of
something
before
before
the
function
execute
and
the
function
ends.
So
you,
you
would
usually
attach
yourself
into
some
function.
For
this
example
like
the
chosen
function
is
under
screen,
but
there
is
also,
let
me
show
you.
B
Since
I
looked
at
it
right,
so
basically,
when
you,
when
you
define
an
instrumentation,
you
have
to
basically
specify
a
select
on
the
right
of
all
the
relevant
functions
that
you
want
to
to
attach
yourself
to
and
like
the
things
that
this
instrumentation
will
only
be
loaded
into
the
camera
kernel
and
be
executed
only
if
the
target
binary
contains
all
this
function,
if
some,
if
some
of
the
function
does
not
is,
are
not
located
in
the
binary,
this
instrumentation
will
not
be
loaded
and,
like
it's
very,
it's
obviously
important
to
choose
some
sort
of
table
functions
that
does
not
change
very
frequently
between
between
different
versions
of
the
instrument
library,
because,
unlike
offsets
and
trucks
and
Fields,
we
do
not
have
a
way
to
track
changed
functions,
because
if
the
code
change,
we
can,
we
probably
need
to
rewrite
our
instrumentation
in
some
way
or
find
a
better
function.
B
So
so,
for
this
example,
we
actually
use
this
decode
header
function
to
to
actually
extract
the
different
grpc
headers
of
the
incoming
request
and
basically,
in
order
to
do
context
propagation.
B
We
need
to
read
if
there
is
a
base
audio
standard
in
those
headers
and
like
this
function,
looked
like
a
really
good
place
to
do
this,
because
it
just
gets
an
argument
with
all
the
relevant
headers
already
passed
already
ready
to
to
use,
but
unfortunately,
also
something
that
I
discovered
during
trying
to
look
into
this
certainty
issue
is
that
in
this
recent
version
of
grpc,
the
only
reason
they
actually
removed
this
function
and
they're
not
using
it
anymore.
B
So,
luckily,
for
us,
the
tricks
is
pretty
simple,
because
we
can
just
call
the
function
above
it
in
the
stock,
the
operate
headers,
which
get
basically
the
same
arguments
again.
We
are
interested
in
this
Frame
thing.
The
frame
strap
and
this
one
did
not
change
throughout
any
Library
version
of
gopc
from
what
I
saw
so
like
one
of
the.
B
So
if
we
are
using
like
and
what
would
be
the
behavior
of,
let's
say
we
have
a
process
running
with
grpc
latest
version
and
of
course
it
will
not
include
this
decode
header
function.
So
basically,
this
is
the
entire
instrumentation
will
not
be
loaded
and
will
not
get
automatically
instrumentation
for
grpc
servers.
B
So
in
this
case
the
the
fix
was
basically
just
to
Target
a
different
function
and
more
stable
functions
operate
headers
instead
of
decode
headers,
but
yeah
I
think
this
is
something
that
is
always
going
to
be
tricky
and
I
feel
like
this
is
the
challenge
for
most
automatic
instrumentation
as
well
like
how
do?
How
do
you
choose
a
stable
point
in
the
code
to
basically
to
trigger
the
adpr
function
or
the
automatic
instrumentation
in
other
languages?.
B
A
Is
kind
of
a
complicated
question
right
because
this
is
going
to
be
a
large
amount
of
our
long-term
support
needs
to
solve
this
right
because,
especially
since
these
are
not
exported
functions,
there's
no
requirement
that
the
actual
instrumentation.
That's
writing
this
like
if
they
write
with
the
standards
of
go
like
there's,
nothing
for
them
to.
A
You
know,
force
them
to
keep
this,
but
then
the
question
becomes
like
you
know
what,
if
this
instrumentation
this
evpf
Auto
instrumentation
project
was
used
to
input
is
right,
the
prior
version
of
grpc
and
now
we're
building
a
new
one
for
for
the
new
grpc
like.
Is
there
a
backwards
compatible
support
here?
Like
does
this
project
still
support
the
old
version
of
grpc
like
if
we
add
change
this,
it
obviously
isn't
going
to
so
like
that
story
is
going
to
be
really
complex.
A
I
think
in
the
Long
Haul
is
how
we're
able
to
provide
support
and
I
think
if
that
gets
it.
A
broader
issue
that
I'll
probably
take
an
action
item
on,
is
that
we
need
to
have
way
clear
documentation
as
to
like
compatibility
of
support.
You
know,
like
this
version
of
go
supports.
This
version
of
you
know
this
library
and
all
these
other
things
so
have
you
had
thoughts
on
the
backwards
compatibility
question,
though,.
B
Yeah
so,
first
of
all
agree
to
to
both
both
of
the
ideas.
I
think,
that's
that
documentation
and
like
a
better
way
to
to
build
this
compatibility
is
definitely
important.
In
this
case.
We
still,
we
don't
break
the
Backward
Compatible
of
grpc
because
operate
headers
data
stand
and
these
consistent
since
the
beginning
of
geography.
B
But
yes,
it's
always
going
to
be
a
challenge
and
I
think
like
one
way
to
maybe
like
we
might
face
an
issue
where,
like
we
can't
keep
this
backward
compatibility
so
because
we
know
the
version
of
the
job
is
the
the
geography
version
of
the
target
process
is
using
the
runtime.
B
We
can
maybe
actually
like
make
this
documentation
mechanism
more
sophisticated
by
saying
that
this
specific
automatic
instrumentation
is
relevant
only
starting
from
this
version
until
that
version
and
like
like
having
different
types
of
instrumentation
or
Logics
trigger
them,
but
according
to
the
to
the
version,
the
library
that
is
being
used-
and
hopefully
you
will
be
able
to
always
get
the
steering
logic
applied
to
any
like
a
person.
But
if
not
I
think
there
is
also
a
way
for
us
to
solve
it.
A
Okay,
I
got
you
yeah
I.
Think
I'll
I'll
also
create
a
ticket
to
kind
of
like
I,
think
in
a
policy
or
get
some
sort
of
strategy
documented
in
this
project
to
say,
like
you
know,
we're
going
to
try
to
do
it
this
way,
or
maybe
we
need
to
look
into
certain
things,
but
yeah,
okay,
I
think
you're
right,
like
I,
think
we
can
get
a
little
more
sophisticated
here.
A
The
the
key
thing,
though,
is
like,
as
you
add,
that
sophistication.
Does
that
mean
that
you
have
like
you
know,
20
different
versions
of
the
same
code
that
you
have
to
maintain
like,
ideally
not
but
yeah.
So
hopefully
we
can
be
smart
about.
B
That
yeah,
hopefully
we
could
find
that
a
good
function,
that's
the
consistent
throughout
the
versions
and
that
it
doesn't
really
carry
the
function
body
change
too
much.
We
just
need
the
the
function
itself,
like
the
name
of
the
function
and
the
and
the
different
arguments
as
long
as
they
stay
the
same.
It's
fine,
the
logic
itself
doesn't
really
matter
because
we
are
not
dependent
on
the
logic
inside
of
the
function.
B
We
just
like
need
to
be
called
whenever
this
function
starts
and
whenever
this
function
ends
and
do
the
instrumentation
stuff,
okay,
yeah
sure.
So
another
thing
that
I
noticed
is
this
function
so
like
there's,
this
concept
of
VPN
helpers
and
the
guys
from
polyphoon
is
probably
now.
This
is
like
all
those
function,
hyper
functions
that
you
can
use
when
you
are
writing.
B
B
It
nowadays
is
by
using
this
okay
DPS
a
time
gets
boots
and
asked
which
basically
just
give
you
the
number
of
I,
think
nanoseconds
that
lapsed
since
the
since
the
last
reboot,
and
then
we
were
able
to
calculate
the
difference,
but
unfortunately,
this
function
is
also
something
that
is
only
available
on
recent
Channel
version.
I
think
version
five,
something
but
bottom
line.
We
want
to
support
like
other
version
of
the
channels,
then
what
this
function
is
available
for
well
again,
also
a
pretty
easy
change
for
now.
B
B
Version
so
like
another
thing
that
we
need
to
be
to
have
in
mind
is
like,
like
and
in
general,
we
probably
need
to
I,
don't
know,
have
some
tests
of
a
way
to
to
make
sure
we
are
not
breaking
current
compatibility.
B
Yeah
because
I
know,
because
the
first
thing
that
I
I
tried
to
reproduce
the
issue.
I
was
using
the
wrong
calendar
version
and
then
brought
that.
But
after
after
I
saw
it,
I
I
switched
to
your
kernel
version
like
the
problem
still
exist,
and
it's
still
not
really
sure
and
what's
the
reason.
But
but
it's
a
good
exercise
I'm
covering
a
lot
of
interesting
things
and
I
think
that
it
helps
us
understand
what
what
works
well
and
what
needs
to
be
improved.
B
But
yeah
I
will
like
I,
don't
have
any
any
clues
right
now
for
what
is
the
reason
behind
the
issue,
but
I
still
understand.
You'll
have
some
some
idea.
A
A
You
might
actually
be
better
off
to
take
the
issue
for
the
wpfk
time
get
transfer,
but
one
of
the
things
that
I
do
think
I
would
recommend,
is
just
setting
a
minimal,
supported
kernel
version
and
saying
that
anything
below
that
like
we're,
not
gonna
I
mean
like
maybe
eventually
we
support
it,
but
I
would
probably
say
like
getting
to
a
stable
release
with
this
project
with
whatever
minimum
is
and
then
saying
like
in
the
future
and
in
the
future,
like
start
to
add,
like
backwards,
compatible
support.
A
If
it's
possible,
like
you,
may
find
that
that's
not
possible,
because
otherwise
you're
just
going
to
be
chasing
your
tail
right,
like
you're,
always
going
to
be
trying
to
get
older
and
a
little
older,
yeah.
B
Yeah,
definitely,
and
if
we
connect
also
something
to
think
about
this,
like
how
we
make
sure
we
are
not
breaking
the
discount
compatibility
thing,
because
it's
really
easy
to
just
use
the
Mbps
helper
function
is
not
supported
in
the
incredible
in
all
channels
and
like
like
the
compatibility
requirements
so
yeah.
A
Yeah,
okay,
I'll
I
can
I'm
gonna
sign
this
new
issue
for.
A
F
B
I
don't
know,
that's
a
good
question.
F
B
Yeah
I
can
tell
you
that
there
are
some
like
this
whole
way
of
measuring.
Time
is
a
little
bit
buggy
and
maybe
maybe
it's
related
to
this
off
CPU
type
thing
but
like
if
I'm
trying
to
basically
create
a
distributed
Trace
with
weird
from
like
Java
applications
that
are
not
instrumented
and
python
applications
that
are
important
and
go
applications
that
is
auto
instrumented,
like
this
different
experimentation.
B
Libraries
measure
time
differently,
and
so
like
you
sometimes
the
stands
with
like
is
not
with
an
offset
in
the
like
in
the
final
phrase
so
like
this
is
something
you
still
need
to
be
explored,
but
like
this,
like,
maybe
also
okay,.
F
About
the
kernel
stuff,
we
recently
implemented
this
in
our
like
agent.
We
are
using
the
BPF
and
the
BPF
supports
ETF.
Maybe
you
know
about
that.
F
D
F
B
The
stadium
one
I
think
it's
also.
Basically,
it's
also.
B
F
B
B
Yeah
definitely
like
assume
like
in
a
general
like
you're
from
Polar
signals
right,
so
you
probably
have
a
lot
of
Philippines
knowledge
to
to
share
with
us.
So
yeah,
definitely
like.
F
Happy
to
help
like
we
are
doing
a
lot
of
stuff
around
go
around
time
and
like
actually
trying
to
profile,
ebpf
and
I
think
the
profile
a
profiler,
and
this
instrumentation
library
has
a
lot
of
intersection
points.
So
we
would
love
to
help
great.
B
B
Okay,
I
think
that's
for
like
debugging
session
that
I
try
to
do
on
the
issue.
Hopefully
it
was
helpful
if
you
have
additional
questions
before
I
move
to
the
next
thing,
be
free
and
like
this
next
one
is
just
an
update
of
what
I'm
currently
working
on.
So
there
is
this
launcher
thing:
I,
don't
know
if
you
guys
remember
that,
basically
we
need
some
paste
in
memory
to
to
write.
I,
don't
know,
places
that
generated
places.
B
I
didn't
send
ID
that
the
go
Application
can
access
to
when
when
we're
doing
context
propagation.
So
like
the
way
we
do
it
right
now
is
that,
like
the
the
co,
application
actually
have
to
be
a
started
from
and
different
applications
launcher
walk
around,
but
I
think
I
found
a
much
better
way
to
do
that,
but
I'm
not
really
sure
they're,
even
working
so
hopefully
by
the
next
meeting,
I'll
be
happy
to
actually
show
your
PLC.
B
How
we
do
that,
but
yeah
that
that's
what
I'm
currently
focused
on
is
like,
basically
just
not
using
this
launcher
thing
and
which
requires
users
to
obviously
launch
the
programs
with
different
programs.
I.
Think
that,
like
not
the
comfortable
things
to
do
yeah,
that's
all
from
my
end.
A
Awesome
thanks
a
bunch
even
that
was
a
a
great
little
insight,
I
think
into
the
offsets.
I
think
that
was
I
think
a
point
of
confusion
in
the
last
meeting,
so
yeah.
D
A
Really
helpful
Dinesh,
you
are
also
very
interested
in
that
is
that
you
have
any
other
questions
on
that.
One.
E
E
Okay,
I
think
offset
tracker
I.
Think
let's
create
an
issue
or
if
you
can
just
add
it
as
a
tools
under
Tools
I,
think
it
should
be
good.
That's
what
we
follow
in
most
other
reports
as
well.
A
So
Dinesh
I
know
Eden's
got
a
few
action
items
on
here.
Would
you
be
up
for
taking?
That
is
just
creating
that
issue
and
then
maybe
making
a
PR
to
migrate?
That.
E
A
E
A
Yeah,
okay,
cool!
Do
we
have
an
issue
tracking
that
eval
migration,
or
even
do
you
just
plan
to
open
a
PR
for
it.
A
Okay,
so
back
to
the
agenda,
there's
like
the
last
thing
is
updating
on
reviews
for
this
other
PR
yeah.
D
D
C
Well,
I
I,
don't
want
to
discuss,
but
I
have
one
small
thing.
So
at
least
few
people
told
me
that
when
they
are
trying
to
open
these
links
for
approvers
or
maintainers
in
this
repo
go
instrumentation,
they
are
receiving
404
error.
I,
don't
know
how
does
that
can
happen
because
for
me
it
works,
but
for
some
people
it
doesn't
work
so
I,
don't
know.
Where
is
the
issue
like
links
in
the
readme
files
in
the
in
the
contributing
I
think
file.
D
Contributing
file,
if
you
could,
you
would
get
me
Tyler
or
open
issue.
A
I
sorry
I'm
just
catching
the
tail
in
there.
It
looks
like
I
got
technical
issues
there.
Sorry,
but
I
I
agree
yeah.
We
need
somebody
to
to
send
an
issue.
I
would
also
say
if
you
wanted
to
try
reproducing
some
of
their
behavior
log
out
of
good
Hub
or
use
like
a
private
browser.
A
A
F
A
A
Okay,
nothing
else
is
on
the
agenda.
If
you
have
something
else,
you
wanted
to
bring
up
love
to
hear
from
you.
A
One
of
the
things
I
think
that
would
be
really
helpful,
is
I,
know.
Dinesh
had
looked
at
trying
to
just
use
the
auto
instrumentation
for
the
evpf
I
think
we're
getting
close
to.
Maybe
you
should
be
able
to
retry
that,
but
I
think
that
if
you
also
are
on
the
call-
and
you
have
some
time
to
maybe
go
look
into
that-
it
helped
explore
a
lot
of
bugs
so
I
think
getting
people
to
actually
just
use.
It
is
really
helpful.
A
A
Cool
all
right,
so
we've
got
some
action
items.
Nothing
else
on
the
agenda.
Yeah
I
think
we
could
probably
end
the
meeting
here.
We
have
15
minutes
left
so
yeah
thanks
everyone
for
joining.
Definitely
some
things
to
do.
Definitely
a
lot
of
work
to
go
forward.
I.
Think
it's
really
exciting,
though
so
I'm
pretty
happy
to
see
the
group
yeah.
We
will
see
yeah.
We
will
see
you
all
in
two
weeks,
time
or
via
slack,
but
until
then
yeah
bye.