►
From YouTube: 2020-12-03 Governance Committee private 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).
D
It
lets
you
use
a
canon,
eos
slr,
except
I'm
crashing
at
my
parent's
place
right
now
and
it's
packed
in
a
bag
somewhere.
So
I'm
just
using
my
shirt
when
you,
when
you
select
it
as
a
camera,
if
there's
no
camera
actually
connected,
it
just
shows
that
bitmap,
oh.
B
C
Yeah,
hey
ben
thanks,
yeah,
I
so
I
work
for
the
linux
foundation
ted
and
I
were
chatting
I
guess
a
few
weeks
ago
about
the
about
conformance
programs
and
I
think
some
of
the
stuff
that
you're
looking
at
so
yeah
happy
to
thanks
for
letting
me
join
and
happy
to
chat
about
some
of
the
the
details
around
that
yeah
sounds
great.
A
We
should
hopefully
have
the
rest
of
the
crew
trickling
in
over
the
next
couple
of
minutes.
Yeah,
there's
yuri.
E
A
Last
big
push-
I
haven't
put
it
on
the
agenda
here,
but
just
an
fyi
trying
to
get
this
versioning
stuff
over
the
finish
line.
That's
a
big
push,
so
we're
gonna
push
hard
to
get
that
all
resolved
next
week,
but
it
definitely
feels
like
a
squeeze.
A
We
have
some
external
groups
that
were
are
really
pressuring
us
to
1.0
or
stabilized
tracing
this
month,
like
you
know
spring,
for
example,
wants
to
release
with
with
this
included,
but
that
just
feels
unrealistic
at
this
point
so,
but
we're
trying
to
at
least
get
everything
in
and
done,
and
hopefully
uh.net
at
least
could
make
an
announcement
by
end
of
year.
A
So
we
should
be
able
to
make
some
kind
of
1.0
announcement
by
the
end
of
the
year,
or
rather
I
should
let
me
be
more
firm
about
that.
We
are
going
to
make
a
1.0
announcement
by
the
end
of
the
year,
but
we
may
not
have
much
in
the
way
of
implementations
there.
Just
looking
at
the
clock
right
now.
A
So
that's
my
update
on
that
front.
You'll
see
an
otep
go
up
about
this
tomorrow.
A
And
maybe
we
should
just
get
rolling
here.
This
meeting
is
recorded,
so
if
people
missed
steve's
presentation
on
this,
they
can
just
watch
the
video.
B
Do
you
mind
just
like
I
mean
sort
of
almost
just
for
posterity,
but
can
you
set
some
context
and
I
know
it
was
over
email
but
for
people
watching
the
recording
like
yes
here?
What
are
we
trying
to
accomplish.
A
Yeah,
absolutely
okay,
so
what
we
are
interested
in
is
when
people
use
open
telemetry.
A
We
want
that
to
mean
something
and
when
people
integrate
with
open
telemetry,
we
want
that
to
mean
something
specific
and
we
want
to
be
able
to
define
a
lot
of
these
things
in
ways
that
are
concrete
and
ideally
legal
like
how
one
can
or
cannot
use
open,
telemetry
or
use
the
term
open
telemetry.
A
And
there
are
some
the
usual
restrictions
apply
to
the
open
telemetry
term.
You
know
as
a
general
trademark,
but
the
cncf
has
found
ways
to
make
additional
certification
trademarks
that
that
allow
people
to
be
able
to
actually
further
integrate
the
work,
they're
doing
and
their
own
brand
into
the
open
telemetry
brand,
and
we're
able
to
put
some
more
stringent
requirements
on
on
that.
So
that
seems
like
a
way
to
help
us
with
things
like
open,
telemetry
distributions,
further
defining
how
those
work.
A
So
that's,
ultimately
the
goal
to
find
a
way
to
to
have
open
telemetry
distributions
that
can
call
themselves
open
telemetry
and
have
that
mean
something
concrete.
C
Yeah
thanks
ted,
and
I
can
I
can
just
to
add
a
little
bit
to
that
so
yeah.
So
I
work
at
the
lf
and
work
with
cncf
in
some
of
the
scenes
of
projects.
One
of
the
one
of
the
things
that
we
we
look
at
from
the
trademark
site
is
conformance
programs,
like
ted,
was
talking
about
and
part
of
the
the
background,
for
it
is
really
you
know
with
with
open
telemetry
or
any
of
our
other
projects,
with
it
being
an
open
source
project.
C
C
However,
they
want,
but
part
of
the
part
of
the
idea
is
that
at
the
same
time,
there's
benefits
to
having
those
different
modified,
downstream
versions
still
be
compatible
with
each
other
and
have
have
it
so
the
technical
community
can
define
what
compatibility
means
and
so
the
way
that
I
think,
as
ted
was
describing
the
way
that
we've
found
this
to
be
effective
and
still
consistent
with
open
source
principles
has
been
to
build
a
trade
trademark
conformance
program
around
that.
So
the
idea
is
really
we
create
a
trademark
that
represents.
That
is
a
conformance
mark.
C
So
it's-
and
there
are,
there-
are
some
nuances
in
different.
The
the
term
certification
mark
does
have
a
pretty
specific
meaning
in
the
trademark
world,
and
it's
actually
different
from
what
we're
doing
here.
So
we
tend
to
call
these
conformance
programs
or
conformance
marks.
We
still
end
up
using
the
certified
terminology.
Sometimes
so,
for
instance,
certified
kubernetes
is
an
example
of
this,
and
it's
not
strictly
speaking,
it's
not
a
certification
mark,
but
the
the
nuances
there
are
a
little
kind
of
a
little
arcane.
So
we
don't.
C
We
don't
need
to
get
into
that,
but
the
idea
is
to
really
create
a
brand
like
that,
so
for
certified,
kubernetes,
there's
that
term
and
there's
a
certified
kubernetes
badge
that
incorporates
the
the
kubernetes
logo
and
then
the
there's
there's
kind
of
two
aspects
to
it.
One
is
there's
a
set
of
terms
and
conditions
and
a
participation,
form
and
kind
of
the
operational
legal
side
of
it.
That
are
that's
something
that
we've
developed.
C
We've
got
a
template
package
for
this,
so
it's
something
that
we
can
roll
out
fairly
easily,
but
it's
basically
the
kind
of
the
the
legal
padding
around
saying.
Okay,
here,
you
will
get
a
trademark
license
to
use
the
sort
of
the
you
know,
the
conformance
mark
in
exchange
for
demonstrating
through
testing
or
through
other
mechanisms,
demonstrating
that
your
implementation
abides
by
the
technical
definitions
of
being
compatible.
C
So
it's
that
kind
of
give
one
for
the
other.
It's
if
you,
if
you
as
a
vendor
or
community
distribution
or
whatever,
if
you
demonstrate
that
your
your
solution
meets
whatever
the
technical
definition
of
compatibility
is,
then
you
get
to
use
the
trademark
license.
So
that's
all
kind
of
on
the
that's
the
pieces
of
that.
C
The
kind
of
terms
and
conditions
of
that
and
that
sort
of
thing-
that's
all
stuff
that
I
can
provide
after
we
kind
of
talk
through
some
of
the
specifics
of
what
you're
picturing
the
program
would
look
like
the
other
side
of
it,
then,
is
the
technical,
the
kind
of
technical
definition
of
compatibility,
and
that's
something
that
I
think
would
be
entirely
left
to
to
you
all
into
the
open
telemetry
community
to
define
what
compatibility
means
so
what
it
means
for
it
to
be
conformant.
C
What
kind
of
category
of
things
you're
looking
at
so
we've
had
some
projects,
doing
conformance
programs
where
the
thing
being
tested
as
conformant
is
a
distribution
of
the
project.
Software
we've
had
other
projects
where
it's
they've
used
conformance
programs
for
something
more
like
plug-ins,
so
demonstrating
that
your
plug-in
or
your
kind
of
added
component
can
speak
with,
can
communicate
with
the
apis
appropriately
or
can
do
whatever
whatever
is
necessary
for
being
conformant.
So
I
think
all
of
this
is
is
something
that
kind
of
we
would
l.
C
C
The
other
part,
I
would
say
just
from
from
a
governance
side,
so
the
in
cncf,
the
the
kind
of
oversight
of
conformance
programs,
is
driven
through
the
governing
board,
so
getting
so
kind
of
the
formal
okay
to
launch
it
and
to
go
forward
with
it
with
it
is
something
that
would
be
driven
through
the
governing
board.
That's
something
that
I
can
work
with
them
and
can
work
with
the
cncf
staff
when
we're
moving.
That
way,
I
think
it's
kind
of
just
one
of
flag.
C
That
is
being
there
are
these,
because
it's
a
it's
kind
of
a
licensing
of
trademark
assets
that
are
that
are
overseen
by
scenes
by
the
cncf
governing
board,
that's
kind
of
where
the
the
governance
for
it
drives
through.
So
so
I
can
help
work
with
with
both
sides
of
that,
but
just
want
to
kind
of
give
that,
as
the
context
is
that
all
kind
of
the
overview
does
that
all
make
sense.
Do
you
have
questions
on
any
of
that.
A
I
think
we'll
have
questions
about
how
to
apply
it
to
open
telemetry,
and
I
think
one
thing
we
have
to
do
is
we.
We
don't
actually
have
like
a
strict
idea
of
what
we're
looking
for
right
now.
So
we're
gonna
have
to
have
some
debates
internally,
but
I
am
curious
about
like
I've.
Looked
at
the
kubernetes
example.
Are
there
other
examples
in
the
cncf
or
otherwise
that
are
like
good
things
for
us
to
look
at.
C
Yeah,
so
in
in
cncf,
I
think
kubernetes
is
the
only
one
that
is
kind
of
a
formal
conformance
program
like
that.
We've
got
a
handful
of
other
ones
that
we've
had
for
other
projects,
and
I
can
se
I
can
add,
links
to
the
the
meeting
minutes
afterwards.
You
can
send
it
out
if
you
want
to
see,
but
a
few
of
them
for
open
mainframe
project.
There's
a
project
called
zoe,
which
is
has
done
a
conformance
program
like
very
similar
to
this.
C
Each
area,
which
is
a
project
in
lfai
and
data,
has
done
something
similar,
but
they've
been
a
little
slower
on
the
uptake
of
of
companies
and
solutions
participating
in
it.
There's
several
in
the
networking
space
that
have
done
something
similar.
One
thing:
I
guess
one
thing
I
would
say
about
this:
that's
kind
of
you
know,
I
think
the
the
starting
point
is
just
is
really
thinking
about
what
the
thing
is
that
you
want
to
that.
C
The
conformance
would
apply
to
so
I
think
ted
you
had
mentioned
potentially
distributions
earlier,
I'm
I'll
I'll
say
I'm
not
at
all
familiar
with
the
open
telemetry.
You
know
code
or
project
itself.
So
I
I
from
the
functionality
side,
I'm
not
going
to
have
a
lot
to
weigh
in
on
there,
but
if
you
know,
I
think
the
first
step
would
be
figuring
out.
What
the
excuse
me,
what
the
thing
is
to
be
to
be
marked
as
conformant.
C
The
second
step
is
going
to
be
focusing
on
what
the
testing
process
or
the
evaluation
process
looks
like,
and
that's
where
we've
had
a
pretty
wide
range
of
everything,
ranging
from
just
kind
of
check
the
box
self-assertions.
So
somebody
is
just
asserting
you
know,
literally
checking
a
bounce
box
and
asserting
that,
yes,
our
solution
meets
the
definition
kind
of
ranging
from
that
to
something
that
is
run
this
set
of
tests.
C
You
run
this
self-test
suite
provide
the
testing
results
to
us,
which
is
what
certified
kubernetes
looks
more
like
kind
of
ranging
up
to
something
that
can
be
as
extensive
as
having
third-party
testing
labs.
Third
parties
that
go
and
actually
test
it
themselves
to
demonstrate,
there's
a
pretty
wide
spectrum
of
what
what
the
options
can
look
like
there,
and
it's
really
driven
by
you
know.
E
Yeah
yeah,
it's
kind
of
interesting
that
you
talk
about
the
kind
of
plug-in
the
distribution
that
implements
an
api.
We
kind
of
have
some
other
elements
of
each
of
these
because
of
the
complexity
of
a
project.
Okay,.
C
Okay,
yeah
and
the
and
the
conformance
program
can
kind
of
operate
as
an
umbrella.
That
way
it
can
be
kind
of
one
structure
in
the
way
that
testing
yeah
the
testing
and
the
process
works.
But
can
you
have
certainly
have
different
meanings
for
distributions
or
plug-ins
or
whatever
the
different
aspects
of
it
are.
A
Yeah,
I
think
we
we
have
two
major
aspects
that
we
want
to
look
at.
One
is
this
standalone
service
called
a
collector.
We
have
to
decide
what
conformance
means
for
that,
but
to
use
kubernetes
as
an
example.
They
have
a
set
of
api
tests
as
long
as
your
thing,
your
binary,
whatever
it
is,
is
compliant
with
those
api
tests.
Then
it's
a
collector.
So
that's
one
definition
that
says
people
can
go,
write
a
collector
in
rust
and
as
long
as
they
implement
these
things
that
counts.
A
I
don't
know
if
we
want
to
say
that
we
may
want
to
say
like
no.
A
collector
is
something
built
through
a
certain
pipeline
and
is
a
certain
set
of
you
know
core
modules
and
plug-ins,
or
something
like
that.
I
we
haven't
had
any
of
these
discussions.
Yet
as
a
group
gotcha.
C
And
just
just
to
ask
a
bit
on
that
and
again
apologies
because
I
really
don't
know
the
the
background
here,
but
so
for
for
something
like
that,
then
somebody,
if
you
were
going
for
the
first
option
there,
where
it's
really
just
it
meets
the
kind
of
meets
the
it's
compliant
with
the
api
test.
So
the
idea
would
be
with
that.
It
sounds
like
would
be.
Somebody
might
not
start
from
open
telemetry
code
at
all.
They
may.
C
A
Would
be
one
example
of
gotcha,
which
I
think
is
the
approach
kubernetes
takes
if
I'm
correct,
if
it
quacks
like
kubernetes,
then
it's
kubernetes.
C
A
Yeah
the
place
where
it's
a
little
more
confusing
is
we
have
distros
of
per
language,
so
there's
something
you
install
in
each
language
and
most
of
the
configuration
that
you
have
to
do
is
kind
of
vendor
or
back-end
specific.
A
So
you
have
to
do
a
lot
of
configuration
to
make
open
telemetry
work
because
it's
kind
of
a
tool
kit,
but
if
you
want
to
connect
it
to
lightstep
or
jager
or
splunk
or
whatever,
almost
all
of
that
configuration
goes
away
and
if
there
are
plugins
that
that
people
need
like
a
set
of
plugins
people
need
to
talk
to
a
specific
backend,
then
those
have
to
be
added
as
well,
and
so
we
were
thinking
that
people
can
bundle
all
that
stuff
up.
A
However,
they
want
and
produce
a
thing,
that's
their
open,
telemetry
distro
for
java,
but
it
seems
like
we
do
need
to
have
some
rules
about
that,
because
we
don't
want
people
to
say
just
anything
they
make
is
is
open,
telemetry
distro
for
java,
but
that
part
is
a
little
bit
trickier
in
my
mind.
I
actually
haven't
thought
clearly
about
how
we
would
go
about
defining
that,
because
it's
not
like
a
service
like
the
collector,
where
you
can
just
say
if
it
has
api
compatibility,
it
counts.
B
To
me
that
if
we
thought
about
this
from
the
standpoint
of
what
we're
trying
to
guard
against,
like
you
know,
this
is
a
lot
of
work
and
process
that
we'll
be
creating,
and
I
agree,
the
collector
is
the
easier
case
and
that
the
the
runtime
pieces,
whatever
you
want
to
call
that
apis,
sdks
etc
are
harder,
seems
like
what
we
want
to
guard
against
is
a
situation
where
any
provider
vendor
otherwise
could
you
know
essentially
hoodwink
their
users
into
installing
an
open,
inflammatory
distro
that
then
locks
them
into
that
into
that
provider.
B
That's
the
thing
we're
trying
to
guard
against,
so
it's
almost
as
if
we
want
to
be
able
to
validate
ideally
programmatically
that
that
you
have
maintained
portability
by
using
that
distribution
and-
and
I
I'm
just
I-
I
correct
me
if
I'm
wrong,
but
I
think
that's
the
thing
that
we're
trying
to
ensure
here
is
that
open,
telemetry,
the
value
proposition
is
you
know
the
there's
a
lot
of
there's
a
lot
of
integrations
out
there
and
you're
not
going
to
get
locked
in
like
that,
and
so
as
long
as
the
distributions
preserve
that
I
I'm
comfortable
with
them
right
like
intuitively.
B
But
we
need
to
make
that
formal,
which
I
think
is
not
actually
that
different
in
some
ways
than
some
of
the
you
know
guarantees
people
want
about
a
kubernetes
distribution
right,
it's
like,
but
but
it's
a
different
integration
surface.
But
do
people
agree
that
that's
the
thing
we're
trying
to
guard
against,
because
I
think
that
would
make
it
easier
to
define
like
what
the
what
compliance
actually
means.
A
A
This
is
why
I
think
some
people
have
cold
feet
about
just
going
out
and
saying,
like
a
collector,
is
this
set
of
apis
because
we
were
kind
of
like
the
community
like
if
you
have
a
cool
thing,
that
you
contribute
it
back
to
the
collector?
If
it's,
if
it's
not
just
a
plug-in,
if
it's
something
meaningful
and
we
will
dpl
so
yeah,
we
could
just
gpl
the
thing
we
want
to.
A
So
without
using
this
sort
of,
I
was
trying
not
to
say
like
without
using
gpl.
I
was
curious.
Steve
are
there
approaches
people
use
just
to
to
encourage
gently,
encourage
the
community
to
contribute
back
upstream
instead
of
forking.
C
So
I
think
so
a
couple
a
couple
thoughts,
I
think,
within
the
scope
of
the
of
something
like
a
conformance
program.
One
way
to
tackle
that
is
to
is
kind
of
again
tied
to
the
way
that
it
is
so
anything
that
anything,
that's
trademark,
related
we're
thinking
about
what
is
kind
of
what
does
this
brand
or
logo?
C
Or
what
does
this
conformance
mark
mean
in
the
mind
of
the
end
users
kind
of
whoever
the
end
users
are
for
this
for
the
particular
thing,
and
so
something
like
something
I
think
bent
what
you
were
saying
before
about
the
the
idea
of
wanting
of
is
the
purpose
of
this.
To
do
something
like
avoid
vendor
lock-in.
I
think
part
yeah
that
part
of
that
would
be
driven
from
the
idea
of
this.
C
This
conformance
mark
would
mean,
if
you're,
if
you're,
developing
a
conformance
program
where
the
goal
is
to
to
have
compatibility
across
different
implementations,
something
then
yeah,
then
that
marks
signals
to
the
end
users
that
you
will
not
be
locked
in,
because
you
can
expect
that
if
you're
going
from
one
conform,
one
thing
that's
marked
conformant
to
another.
One
it'll
behave
in
this
defined
way,
so
I
think
that's
so
but
ted.
I
think
in
terms
of
contributions
coming
back
upstream.
C
I
think
one
way
within
that
in
a
in
a
within
the
conformance
program
would
be
just
different.
Signaling
of
you
know,
sorry,
with
the
conformance
program,
there's
kind
of
one
thing
we've
seen
projects
do
is
have
like
base
level
conformance,
but
other
profiles.
On
top
of
that,
so
the
idea
is
kind
of
the
base
layer
is
anything
being
called.
Conformant
meets
these
minimal
requirements,
but
then
additional
profiles
are
kind
of
optional,
additional
things.
On
top
of
that,
additional
criteria
that
are
met
and
typically
we've
seen
those
be
more
functional
criteria.
C
So
something
like
something
like
additional
security.
You
know
some
you,
you
meet
these
additional
tests
and
your
you
can
add
a
like
security
profile
marking
to
it
and
that's
something
that
would
show
up
on
the
badge
you
could.
I
could
see
something
where
you
could
also
have
like
you
know,
an
open
marking
or
something
where
it's
just
kind
of
a
one
of
the
defined
profiles
is
you
have
released
it
as
open
source
or
you've,
released
it
as
apache
or
even
you've,
contributed
it
back.
You
know
into
the
upstream
open,
telemetry,
so
yeah.
A
Yeah
yeah,
I'm
wondering
if
I
would
prefer
like
the
gpl,
is
the
stick
and
we
would
prefer
to
use
a
carrot
and
the
way
I
see
carrots
working
is
people
the
the
companies
that
have
the
people
power
to
to
really
fork
and
go
hard
with
one
of
these,
I
suspect,
with
the
same
companies
that
would
really
really
like
to
have
these
certifications,
and
so,
to
my
mind,
the
way
we
can
encourage
a
contribution
to
kind
of
stay
in
core
and
have
like
a
center
of
gravity.
A
Stay
in
core
would
be
to
have
these
certifications
help.
You
know
if
you
could
go,
make
your
own
cool
thing,
but
then
you
lose
the
certification
and
you
want
the
certification,
and
so
that
encourages
you
to
to
contribute
to
the
upstream
implementation,
specifically
just
to
finish
that's
a
case
where
say
api
tests
actually
wouldn't
do
that
job.
My
mind,
yeah.
C
So
I
I
guess
one
one
question
I
do
have
on
that
ted
is,
if
you
are,
if
the
goal
is
to
have
them,
contribute
it
back
up
into
open
telemetry
into
the
the
upstream,
then
kind
of
you
can
almost
set
the
conformance
program
aside,
because
really,
if
they're
contributing
it
back
up
into
into
open
telemetry,
then
they
can.
Then
you
know
that
is
open,
telemetry,
I'm,
you
know
being
a
little
tautological
here,
but
it's
like.
C
If
it
you
know,
you
can
accurately
use
the
open,
telemetry
brand
and
name
to
describe
the
thing
that
is
the
upstream
open
telemetry.
So
if
somebody
is
using
literally
that
kind
of
unmodified,
then
you
know
that
itself
can
be
kind
of
a
you
know:
an
impetus
to
want
to
contribute
back
up
into
the
into
the
upstream
yeah,
but
I
think
the
I
I
think.
As
far
as
with
with
what
you're
talking
about
about
api
tests,
I
do
take
your
point.
C
If
the,
if
the,
if
compatibility,
is
just
defined
through
api
tests-
and
you
don't
care
about
what's
inside
then
yeah,
the
kind
of
default
view
of
that
would
probably
be.
Are
you
yo?
Does
that
mean
that
kind
of
the
open
telemetry
code
isn't
relevant
at
all
at
all,
as
long
as
you're
just
meeting
the
apis
and
yeah?
I
guess
one
thing
you
could
consider
with
the
api
test
and
I'd
want
to
I'd,
want
to
think
about
this,
a
bit
more
and
probably
talk
with
some
folks
here.
C
But
I
guess
you
could
consider
saying
as
a
work.
Are
you
for
something
like
that?
Are
you
going
to
have
as
a
requirement
that
to
participate
in
the
program?
You
also
have
to
release
it
as
open
source
or
release
it
under
apache
2,
or
something
like
that.
So
not
just
not
just
an
additional
optional
profile,
but
literally
that
as
a
requirement
of
the
project
of
the
program.
I'd
want
to
think
about
that
a
bit
more
right.
A
C
F
Ben
I
wanted
to
go
back
into
your
question
before,
like
what
kind
of
things
we
want
to
protect
ourselves
from
so
I
mean
we
have
a
leverage
over
vendors
who
want
to
be
part
of
it,
but
our
mission
is
to
make
cloud
native
software
instrumented,
so
we
may
want
to
have
certification
for
instrumentation
for
instrumentation
sites
where
remain
like.
F
They
may
not
want
this
certification
per
se
like
we
don't
have
that
much
of
the
leverage,
but
we
still
can
I
mean
if
all
the
instrumentation
size
will
be
compatible
with
opencl
m32,
it
might
be
even
bigger
wins
and
one
we
under
distributed
some
sdk.
That
is
not
fully
compatible.
B
Hi
boris,
I
was,
I
was
wondering
about
it's
kind
of
along
those
lines,
but
I
think
the
instrumentation
is
actually
obviously
valuable,
but
the
things
that
I
expect
a
lot
of
these
distributions
will
do
that
are
particularly
valuable,
is
reducing
the
need
to
do
any
recompilation.
B
I
think
it's
just
going
to
become
less
and
less
effort
to
actually
integrate
like,
for
instance,
if
I'm,
if
I'm
some,
you
know,
if
I'm
a
vendor,
it's
going
to
be
easier
for
me
to
distribute
my
open
telemetry
stuff.
If
our
customers
don't
need
to
make
any
code
changes
whatsoever
to
use
me
as
a
vendor-
and
I
think
that's
probably
the
main
draw
and
the
instrumentation
coverage
isn't
it's
sort
of
neither
here
nor
there
on
that,
and
I
think
that's
probably
what
the
brunt
of
the
distributions
will
actually
be.
B
B
So
that
would
be
one
mechanism
if,
if,
if
you
know
compliant
providers
or
were
required
to
maintain
both
of
those
options,
sergey
that
doesn't
really
speak
to
your
your
point.
But
it's
kind
of
vaguely
related.
A
Just
to
follow
up
on
sergey's
point,
I
think
what
sergey
you're
saying
is
we
we
want
native
instrumentation.
We
want
to
encourage
rails
to
just
bake,
open
telemetry
into
it,
and
possibly
we
could
also
offer
a
certification
to
those
instrumentation
or
sorry
to
those
framework
maintainers
to
be
able
to
say
this
is
an
open,
telemetry
thing
like
a
badge,
essentially
that
you
can
put
that's
right.
F
Yeah
and
ideally,
if
it's
sdk
from
a
company
like
like
full
cloud,
sdk
that
this
sdk
wouldn't
take
depends
on
their
like
sdk
to
operate
on
a
cloud
not
a
like
sdk
for
opencmg,
then
this
sdk
might
be
instrumented
like
rest
api
sdk.
Then
this
sdk
will
be
instrumented
as
open
telemetry.
Api
natively,
rather
than
taking,
depends
on
x
cloud
like
opencdm
right.
B
A
You
just
baked
in
some
other
thing.
I
think
all
of
this
is
is
interesting.
A
One
approach
I've
been
thinking
about
just
to
pitch
it
here
is
for
for
things
like
the
collector
and
even
these
distros
we
could
have
like
a
build
pipeline
for
these.
The
thing
that
comes
to
my
mind
is
jquery.
I
don't
know
if
people
use
that,
but
it
had
a
huge
array
of
plugins
and
you
didn't
want
them
all
all
the
time,
and
so
what
you
did
was
you
selected
which
plugins
you
wanted,
and
then
it
built
you
a
binary
for
that.
A
Essentially,
so
one
approach
is
like
official
open
telemetry
things
are
only
things
that
come
out
of
a
certain
pipeline
like
that,
so
you
have
to
like
get
it
into
this
pipeline
and
it
has
to
be
built
and
then
hosted
in
like
the
open
telemetry
registry
somewhere.
A
Some
approach
like
that
would
put
work
and
resource
consumption
and
things
like
that
on
the
project.
But
that's
like
a
different
approach
from
just
saying
you
have
to
be
api
compliant
right.
That
gets
that's
one
way
to
do
what
ben
is
talking
about,
where
you
have
to
be
able
to
build
this
thing
by
adding
plug-ins
to
the
original
thing.
It
can't
be
some
other
implementation.
C
And-
and
I
think
going
going
kind
of
off
of
that-
just
one
one
additional
thought-
I
think
just
in
thinking
about
something
like
a
conformance
program
and
how
downstream
you
know,
organizations
or
communities
or
whatever
would
use
would
be
using
the
open
telemetry
mark.
I
think
it's
important
to
keep
in
mind.
You
want
to
keep
a
distinction
between
what
is
open,
telemetry
itself.
So
what
is
the
upstream
project?
C
What
is
code
coming
from
the
upstream
project
versus
what
is
somebody
else's
distribution
or
implementation,
or
whatever
of
it,
so
just
kind
of
keeping
that
idea
that
the
conformance
program
is
for
the
things
that
are
others,
you
know
other
downstream
uses
or
compatible
or
integration
of
versus
open
telemetry
itself
means
the
project.
It
means
the
stuff
from
the
project
yeah
right.
A
A
We
could
go
with
this,
but
it
does
seem
like
we
do
need
to
like
do
something
around
these
terms
to
to
nail
them
down
before
it
gets
too
late,
and
you
know
stuff
starts
popping
up
calling
itself
open,
telemetry
distro
with
now
the
extended
open,
telemetry
api
that
you
can
only
use
with
this
distro
or
some.
You
know
vendor
lock
anything
like
that.
G
Let
me
say:
compatibility:
are
we
talking
about
api
compatibility
or,
like
I
wouldn't
expect
like
a
vendor
sdk
to
be
replaced
like
if
it
exports
to
a
vendor,
you
would
not
expect
it
to
be
replaced
or
to
export
to
jager.
For
example,
right.
That's
not
the
requirement
there.
A
Yeah
yeah,
I
don't
know
what
one
example
is.
You
could
just
say.
Look
anything
that
can
bind
to
the
open
telemetry
api
is
is
a
distro.
I
don't
know
that
we
want
to
say
that
right.
That's
just
like
something!
That's
like
api
compliant.
You
know
the
question
I
have.
I
think
it's
actually
a
little
more
vague
with
with
the
sdk
and
the
stuff
that
gets
embedded
in
the
client,
but
so
go
ahead.
H
A
Yes,
that
that
is
like
number
one,
no,
no!
I
don't.
I
think
I
presume
we
can
enforce
that
through
a
compliance
program.
Saying
like
your
thing,
can't
provide
an
api
extension.
C
Let
me
let
me
take
a
look
at
that
and
dig
into
that
a
little
bit
the
the
background
of
all
of
this
and
I'm
not
going
to
get
into
the
details.
But
the
background
of
all
of
this
is
also
at
the
same
antitrust
considerations,
and
so
I'm
not
going
to
get
into
the
details
there,
but
just
making
sure
that
the
in
putting
together
a
conformance
program
that
we're
doing
it
in
a
way
that
isn't
overly
restricting
what
companies
can
do
or
what
what
companies
or
communities
or
anybody
can
do
with
with
the
code.
C
G
But
for
the
record,
I
I
wouldn't
necessarily
agree
that,
like
no
api
extension
is
so,
there
might
be
like
different
levels
in
the
certification
or
something.
But
I
totally
see
the
need
for
api
extensions
by
vendors,
like
even
in
jaeger.
We
had
certain
functions
which
were
not
an
open
tracing
because
they
simply
provided
an
additional
functionality
that
there
is
no
way
to
deal
with
the
open
tracing
api
right
and
so
like
saying
that
you
can't
do
that.
It's
kind
of.
A
Yeah,
I
mean
I'm
thinking
about
the
instrumentation
api
specifically,
but
I
agree
it
if
we
get
overly
restricted
it
sucks.
But
to
me
I
feel
like
like
we
have
to
put
those
things
into
the
api
like
we
have
to
add
that
kind
of
stuff
and
make
the
api
usable
for
everybody.
A
Otherwise
it's
hard
to
define
what
if
people
go
around
saying
this
is
open,
telemetry,
plus
plus,
and
you
need
to
use
the
open,
telemetry
plus
plus
package
to
use
it,
and
everyone
starts
depending
on
that
thing.
Instead
of
the
regular
open,
telemetry
api
or
just
some
high
level
thing
that
wraps
up
some
api
calls,
but
ultimately
just
devolves
down
into
api
calls
anything
that
goes
around
the
api
and
talks
directly
to
a
plugin
seems
like
it's
bad
news.
If
it's
in
the
instrumentation
api.
A
A
But
I
feel
like
we're
losing
our
audience
here.
So
what
should
we
do
his
next
step?
Steve
just
so
we're
not
just
sitting
here
spitfalling
with
wasting
your
time.
C
No,
no,
it's
all
good,
I
so
I
do
think
so.
I
think
I've
got
a
couple
questions.
I'm
gonna
go
back
and
look
at
just
from
my
side
about
what
is
doable,
but
I
think
I
think
those
are
a
little
kind
of
parallel
to
figuring
out
what
I
think
that
core
question
of
what
are
the
initial
starting
points
for
technically
what
you
want,
the
conformance
to
look
like
what
are
the
conformant
things?
What
are
the
ways
in
which
you
would
test
them
in
which
people
would
demonstrate
that
they
are
conformant?
C
And
you
know
it
sounds
like
there's
still
a
lot
of
conversation
to
be
had
about
that.
So
I
don't
know.
If
that's
something
I
know
in
in
other
projects,
I've
seen
them
spin
out
a
working
group.
That's
focused
on
conformance.
I
don't
know.
If
that's
something
that
you
know
you
would
be
looking
to
do
here
or
if
you'd
look
to
continue
doing
it
through
governance
through
these
sort
of
meetings.
But
it
sounds
like
that
kind
of
continued
discussions
around
that
from
the
technical
side
is
kind
of
the
starting
point.
A
Yeah
yeah,
I
do,
I
suspect
we
want
to
open
this
discussion
up
and
have
it
not
just
be
the
gc
deciding
it
like.
Ultimately,
we'll
have
to
go
through
some
public
process,
but
I
think
we
want
to
have
our
ducks
in
a
row
and
maybe,
as
a
gc,
have
something
of
a
framework
of
a
plan
to
put
together
when
we
kick
that
off.
So
it's
not
just
blank
slate,
so
I
think
that's
like
our
next
goal
is
to
it
sounds
like
for
things
like
api
compliance.
That
seems
very
clear-cut.
A
It
seems
like
the
part
that
we
need
to
understand
more
about
is
if
we
want
to
have
compliance,
mean
plugins
that
compile
like,
like
you,
have
to
use
this
particular
implementation,
and
your
thing
must
be
an
open
source
plugin
that
is
added
to
this
implementation.
I.
C
Think
it's
particularly
the
latter
that
I'm
thinking
about
the
the
particularly
saying
you
have
to
you
have
to
release
your
thing
as
open
source
or
you
may
not
add
any
extensions
to
it.
I
think
those
are
the
two
things
that
I
want
to
do
I'm
going
to
do
some
thinking
about
and
come
back
on
that
yeah.
A
Great
that
that
would
be
super
helpful
cool.
Okay.
I
think
I
think
we're
good
awesome
really
appreciate
it.