►
From YouTube: 2021-03-09 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).
C
A
C
B
Yeah,
let's
get
started,
I'm
sharing
my
screen.
Yes,
okay,.
B
So
bringing
back
the
topic
from
maybe
three
weeks
back,
so
we've
been
asking
the
maintenance
and
approvers
from
the
main
repo
to
explicitly
acknowledge
that
they
want
to
be
a
approver,
slash
maintainer
for
the
control
repo,
and
I
got
like
positive
responses
from
few
of
them.
So
next
action
is
on
me.
I
will
actually
submit
a
pr
and
remove
everyone
who
is
not
who
has
not
explicitly
acknowledged
it.
I
think
paulo
you
mentioned
you
want
to
stick
around,
but
I
I
didn't
see
you
here.
So
if
you
want
you
can
like
respond
here.
E
I
I
okay,
I
I
I
respond
there.
I.
B
Vaguely
remember
like
you
mentioned,
you
want
to
continue,
but
I
I
didn't
see
you
playing
here
so
yeah.
So
that's
the
first
update
from
me.
I
will
do
the
actual
changes
and
there
are
few
security.
I
mean
there
are
groups
which
control
these
users,
so
I
can
make
sure
people
are
in
the
right
group.
Interestingly,
for
rayleigh,
he
was
listed
in
the
readme
file,
but
he
was
not
part
of
the
rule,
but
so
I
don't
really
need
to
do
anything
to
remove
him
except
updating
the
document.
B
Okay,
just
to
go
back
to
agenda
items,
there
are
like
couple
of
pr's,
which
we
said
like
last
week
that
will
be
merging
last
week
itself,
but
we
do
have
some
open
questing.
So
let's
take
a
look
at
them,
so
this
is
about
instrumentation.
So
recently
we
changed
the
way
our
open
elementary
tracer
provider
works.
Now
it
understands
legacy
activities
natively
without
the
need
of
using
an
adapter,
so
that
should
unblock
a
lot
of
not
a
lot
like
all
of
the
other
instrumentations
in
the
contrib
repo.
B
However,
while
working
on
that,
we
still
found
like
like
another
issue
which
is
about,
I
mean
the
easiest
way
to
put
that
is,
should
instrumentations
take
a
dependency
on
api
or
sdk.
That's
what
the
issue
is
about,
and
the
underlying
course
is
about
the
surprise.
So
I
can
give
a
quick
summary
of
what
that
issue
is.
So,
in
general,
we've
been
recommending
all
instrumentations
to
just
depend
on
the
api
alone,
because
instrumentation
is
supposed
to
like
produce
instrumentation
or
describe
the
telemetry,
which
only
requires
the
api.
B
However,
some
of
the
instrumentations
offer
a
functionality
called
suppress
downstream
instrumentation.
One
good
example
is
the
grpc
one
which
has
a
feature
by
which
it
will
suppress
the
downstream
instrumentation,
which
is
it
suppresses
the
instrumentation
from
the
underlying
http
client.
Now
that
feature
relies
on
the
usage
of
sdk's
surplus
instrumentation,
and
that
means
in
order
to
offer
such
a
feature,
such
instrumentation
will
be
forced
to
take
a
dependency
on
the
open,
telemetry
spk.
B
B
However,
that's
probably
the
easiest
thing
we
can
do,
and
the
thing
I
have
mentioned
like
what
are
the
downside
of
that.
So
if
an
instrumentation
is
depending
on
this
sdk,
then
it
won't
work
if
a
user
is
using
an
alternate
implementation
of
the
sdk.
B
The
other
alternate
was
to
move
the
the
suppress
functionality
into
the
api
itself,
so
it
could
be
like
part
of
the
open,
elementary
api,
but
then
again
it
brings
the
same
question
like
what,
if,
like
someone
else
like
some
company,
wants
to
have
an
alternate
implementation
of
the
sdk
they
may
or
may
not
implement
that
functionality,
because
that
functionality
is
not
specked
out.
It's
not
part
of
the
official
sdk
spec,
so
it
will
be
a
like
thing
which
is
applicable
to
this
repo.
B
One
of
the
alternate
approach
which
I
haven't
quite
finished
doing
is
to
use
a
different
slot.
I
mean
the
suppress
instrumentation
is
built
on
top
of
context
api,
which
is
purely
an
api
concept,
not
requiring
an
sdk.
So
here
what
one
of
the
proposals
will
create
a
different
slot
with
a
different
name.
This
and
all
the
instrumentations
from
this
repo,
which
will
be
sharing
the
diagnostic
source.
B
B
I
I
want
to
hear
like
if
anyone
has
any
concerns
about
using
this,
because
this
would
mean
like
you,
don't
need
to
take
a
dependency
on
the
sdk,
because
this
feature
is
part
of
api
itself,
so
instrumentations
can
depend
on
just
the
api
and,
of
course
like
this
is
currently
using
a
particular
like
name
for
this
law.
So
any
instrumentation
from
this
repo
can
leverage
it.
B
B
Sorry
did
someone
speak
up.
F
A
quick
question:
does
the
does
the
net
contrib
or
whatever
employ
dependency
injection
for
any
of
their
work
for
the
instrumentation
and
stuff.
B
No
like,
when
you
say
like,
does
it
deploy
dependency
injection?
My
definite
answer
is
no,
because
there
is
no
dependency
on
microsoft.
Extension
stock
dependency
injection.
If
that's
what
you
are
referring
to.
F
B
Yeah,
that's
in
a
different
context.
Yeah.
I
think
the
issue
which
you
were
referring
to
was
how
to
obtain
a
non-default
tracer
provider.
F
B
Yeah,
it's
a
different
reason
why
someone
wants
to
take
a
dependency
on
sdk,
but
in
this
case
it's
specific
to
the
feature
called
suppress
downstream
instrumentation,
which
is
required
by
like
something
like
elasticsearch,
is
one
example,
because
it's
built
on
top
of
http,
so
elasticsearch
would
produce
its
own
scan.
Http
would
produce
its
own
span
and
users
want
to
like
suppress
the
downstream
one
to
avoid
like
a
lot
of
duplicated
information.
So
that's
a
context
for
this
particular
issue.
F
Yeah
I
I
was
just.
I
was
just
wondering:
if
the
trace
api
was
fully
supporting,
you
know
some
way
of
not
requiring
sdk.
Then
then
I
guess
I
have
to
learn
why
the
instrumentation
would
need
to
take
a
dependency.
If
it
all
you
know
if
the
api
doesn't
require
it
so
yeah
anyway,
I
have
to
learn
more
about
it.
So.
B
I
again
share
offline,
like
if
you
want
like
why
this
particular
issue,
but
I
want
to
see
like
if
anyone
else
has.
I
I'm
just
testing
this
approach
with
eric
has
like
tried
out
in
this
yeah.
I
mean
he
has
a
proposal
it
requires.
I
mean
it
does
not
require
the
instrumentation
to
depend
on
the
sdk,
so
it
seems
like
a
neat
option
and
it
doesn't
also
force
us
to
expose
these
methods
as
public.
B
B
Or
like,
if
you
haven't
got
time
to
like
look
at
the
big
history,
you
can
like
respond
in
the
pr
itself.
But
if
I
don't
see
any
issues,
I
I'm
just
validating
with
unit
test
because
it
doesn't
look
like
we
had
unit
tests
coverage
which
tested
this.
So
I
added
like
some
of
some
more
unit
tests,
I'm
still
adding
more
to
make
sure
that
we
have
any
easy
way
to
validate
that
this
approach
works.
B
So
if
no
other
comments
I'll
move
to
the
next
topic,
but
I
just
want
to
see
if
anyone
else
have
any
other
comments
about
this.
A
I
just
had
a
question
because
I
know
that
there's
been
some
discussion
about
this
context,
api
being
a
part
of
the
spec,
and
I
wasn't
sure
where
that
conversation,
if
that
conversation
is,
is
happening
today,.
B
B
Yeah
recollect
seeing
an
issue
about
this
yeah
I
mean
there
was
some
talk
about
it,
but
I
had
to
like
do
some
history
digging
because
there
were
some
links
to
issues,
slash
discussions
about
this
yeah
I
mean
I,
I
have
to
do
some
searching
yeah,
but
if,
if
it's
something
which
should
come
in
the
spec,
then
we
should
be
like
able
to
implement
that
right
in
the
api
itself.
So
that
should
not
be
an
issue.
B
The
current
issue
is
like:
if
we
just
do
something
like
very
specific
youtube.net
and
we
just
expose
it
in
apa,
then
it
it
even
if
we
expose
it
in
the
api
that
doesn't
mean
like
every
http
implementation
would
actually
follow
it,
because
it
will
be
just
like
our
own
invention,
rather
than
coming
from
a
specification.
G
B
I
think
we
can
try
to
explain
that
in
the
context
of
grpc
instrumentation,
which
would
do
the
suppressing.
So,
let's.
B
So,
michael
is
your
question
about
like:
why
are
we
like
even
doing
this?
Is
that.
B
Yeah,
this
is
exactly
the
same
problem
which
I
was
trying
to
find
answer
like.
Since
previously
the
start
and
stop
of
activity
was
not
triggered
by
the
tracer
provider
sdk,
it
was
like
the
activity
source
adapter,
but
now
that
tracer
provider
sdk
is
doing
the
or
responding
to
the
start
and
stop
of
the
legacy
activity.
B
I
I
have
a
feeling
that
we
might
be
able
to
solve
it,
but
like
eric
had
a
unit
test
which
proves
other
way,
so
that
means
I
have
to
do
like
one
more
search
and
see.
Why
would
we
need
to
do
this?
Despite
the
fact
that
tracer
is
doing
it?
Natively.
B
G
That
yeah
I
mean
I
just
did
a
basic
test
where
I
launched
like
one
of
the
example
projects
with
breakpoints
on
the
increment
and
did
decrement
it
looked
to
me
like
they
were
always
just
called
twice.
So
it
was
very
cursory
analysis,
but
I
was
like.
Oh,
we
just
don't
need
this
at
all,
but
yeah.
B
I
was
just
doing
a
quick
check
on
grpc,
so
I
just
started
a
unit
test
to
confirm
that,
like
it
would
work,
I
mean
I
I'm
only
like
part
one
into
that,
because
I
was
looking
at
the
unit
test
for
grpc,
and
this
is
a
unit
test
which
checks
whether
the
surprise
feature
is
working
or
not,
but
it
it
was
just
checking
whether
the
actual
activity
has
display
name
or
not,
and
display
name
is
something
which
is
set
in
the
onstart
callback,
but
the
status
the
grpc
status
is
something
which
was
set
in
the
on
end
callback,
and
since
we
are
not
validating
anything
in
onand,
I
don't
think
like
the
test
coverage
was
sufficient.
B
So
this
is
just
step
one.
I
have
test
added
now
now
I
can
go
ahead
and
play
with
it
little
more
and
see
whether
we
might
be
able
to
get
away
with
that
or
not,
and
I
I'll
just
admit
that
it
is
a
bit
tricky
for
me
to
understand
like
I
know
that
there
was
like
this
increment
if
triggered
and
decrement
was
not
there
originally,
and
it
was
added
in
response
to
the
suppress
instrumentation
feature
in
the
grpc.
B
B
Yeah,
it
would
be
great
if
you
like
completely
get
rid
of
it,
but
based
on
what
eric
says
in
his
unit
test
in
the
elasticsearch,
it
doesn't
seem
to
work.
So
I
want
to
like
spend
some
more
time
to
understand
this
and
see
if
we
have
any
better
way
of
doing
this,
but
if
we
indeed
confirm
that
we
need
it,
then
this
approach,
where
I
think
it's
yeah
having
this
diagnostic
listener,
do
its
own
context,
increment
and
decrement
that
might
work.
G
Yeah,
I
think
that's
that's
reasonable
to
me,
because
at
least
it
localizes
the
change
to
the
diagnostic
source
listener.
It
doesn't
leak
kind
of
the
implementation
that
the
tracer
listener
is
doing.
If
that
makes
sense,
I'm
okay
with
this-
it's
not
the
cleanest,
but
I
think
if
we
have
to
do
something,
this
is
reasonable.
So.
B
First
thing
is
like
we
confirm
whether
this
is
indeed
required
if
it
is
required
yeah.
This
seems
like
the
like
better
among
the
alternate
options.
B
B
It's
still
not
done
like
I'm
still
working
on
it,
so
hopefully
we'll
have
something
before
the
next
week
six.
So
this
is
probably
the
last
thing,
as
I
mentioned
like
once.
We
unblock
this
thing
and
we'll
be
like
basically
unlocking
the
elasticsearch
instrumentation,
which
means
we
can
just
replicate
that
into
the
remaining
instrumentations
and
have
a
some
version
of
the
instrumentation
shift.
B
It's
long
time
since
we
shipped
something
so
we'll
be
able
to
do
another
release
which
of
those
instrumentations
which
depend
on
the
like
1.0
version
of
the
apa
or
sdk.
B
That
also
brings
me
to
the
next
topic,
which
is
there
is
a
like
overall
tracing
roadmap,
which
is
being
published
in
the
maintenance
meeting.
It
should
be
like
coming
out
as
a
blog
post
or
something,
but
there
are
like
few
themes,
which
is
being
focused
on
as
a
overall
theme
for
the
open
elementary,
so
every
languages
are
expected
to
contribute
to
this
effort.
B
Oh
come
on,
it's
still
loading
yeah.
So
it's
like
the
immediate
thing
like
except
matrix.
The
key
item
is
about
like
instrumentation,
so
there
are.
There
is
a
proposal
to
make
instrumentations
stable
before
april
or
me
yeah,
which
means
like
we'll
be
able
to
release
one
dot
of
the
all
the
instrumentations
by
that
time.
So
we'll
have
like
some
more
time
to
fix
it,
probably
like
one
more
month,
but
that's
just
one
thing
like
supporting
more
instrumentation
and
making
it
easy
for
other
people
to
write
instrumentation
as
well.
B
So
that's
one
thing
and
of
course
there
are
other
things
like
ease
of
on-boarding,
improved
self-diagnostics.
B
The
ross
asks
in
the
maintenance
meeting
if
anyone
wants
to
like
join
this
effort,
at
least
in
their
own
language,
feel
free
to
reach
out
to
ted,
who
is
doing
the
overall
driving
this
project,
but
you
don't
necessarily
have
to
reach
out
anyone.
You
can
just
do
changes
in
the
document
itself
if
you
want
there
are
like,
as
part
of
this
like
one
of
the
thing
which
we
want
to
address
is
the
onboarding
experience
for
asp.net
core.
B
There
is
a
project
which
we
have
specifically
for
sp
net
core,
which
is
a
microsoft,
extensions
hosting
package.
Sorry,
the
open
elementary
extension-
it's
somewhat
broken
now
so
like
like
improving
this-
would
be
in
alignment
with
the
overall
tracing
roadmap,
so
we'll
be
trying
to
fix
the
issues.
I
think
it's
not
here
like
it.
It's
thrown
out
in
several
issues
how
to
make
it
easy
for
configuring,
something
using
delay
model.
B
So
those
are
like
overall
themes
and
if
anyone
is
like
trying
to
contribute,
this
is
one
area
to
contribute
right
away,
instrumentation
and
ease
of
onboarding,
which
involves
the
extensions
of
the
extensions
hosting.
This
is
like
all
I
say,
like
matrix.
Matrix,
is
still
like
going
on
parallelly
victor
he's
already
in
this
group,
so
he's
doing
a
lot
of
prototype
with
dot
netron
painting
and
hopefully,
we'll
have
like
some
some
build,
some
build
from
the
dot
net
itself
to
start
building
on
top
by
next
month.
H
To
a
lot
of
the
things
that
we're
doing
here,
but
for
the
convenience
api,
I
understand
that
that's
to
make
onboarding
simpler,
how
does
that
fit
into
like
the
dot
net,
sdk
strategy
or
just
road
map
in
general,.
B
Yeah
so
one
of
the
issue,
which
let
me
open
an
issue
so
or
let's
do
like
one
example
of,
for
example
like
if
you
just
take
the
instrumentation
for
this
http,
I
am
fairly
confident
that
this
is
one
of
the
easiest
thing
to
onboard,
because
all
you
need
is
add
one
package
and
add
like
one
liner.
B
This,
like
rest
of
the
thing,
should
already
be
there,
so
it's
just
a
matter
of
like
one
line,
so
this
onboarding
is
very
easy,
but
then
the
question
is:
how
do
you
configure
anything
more
or
how
do
you
change
your
exporters
settings
at
that
point?
Things
gets
really
complicated.
Some
some
issues
are
open
here
about
using
the
let's
see
if
we
can
find
that.
B
Yeah,
probably
these
issues,
so
if
you
are
in
like
sp
net
core
world
the
way
you
configure
things
are
something
like
this
like
it's.
These
are
the
things
which
want
to
explore
and
improve,
because
even
though
we
support
like.net
framework
majority
of
the
customers
are
going
to
be
in
the
asp.net
core
world,
so
someone
wants
to
like
configure
like
something
like
jaeger
for
extends.
B
They
should
be
able
to
write
code
like
this,
which
we
don't
currently
support,
because
in
sp
network
there
is
a
like
easy
way
to
configure
pretty
much
everything,
but
we
don't
support
it.
We
need
to
explore
like
how
to
make
this
support,
because
we
tried
this
earlier,
but
this
requires
access
to
microsoft,
ba
and
microsoft
extensions.com.
B
So
a
lot
of
dependencies
will
be
required,
so
we
want
to
limit
it
to
only
the
extension
start
hosting
project,
not
the
sdk,
so
the
sdk
would
continue
to
work
without
that,
so
the
convenience
api
is
only
for
those
people
who
are
using
the
extensions,
slash
hosting
project,
which
is
majority
of
the
customers
using
asp.net
core
or
any
dotnet
core
worker
applications
as
well.
B
So
those
are
the
things
which
I
I
have
in
mind
to
improve
the
onboarding
experience
and
we
do
have
like
this
was
asked
like
several
times
in
the
past
as
well.
B
B
D
Okay,
so
so
I
I,
I
think
one
thing
we
might
be
able
to
do
like
probably
have
a
funnel
for
people
to
create
issue
if
they
feel
anything
that
that
can
be
made
simpler,
and
I
know
that
is
working
with
matt
the
pm
trying
to
collect
the
feedback,
so
they
all
have
some
form.
D
B
D
Us
exactly
yeah,
so
so
ted
is
working
on
the
like
just
to
kick
this
off,
but
he's
running
out
of
bandwidth.
So
so
matt
is
going
to
help
out
here
and
you
can
expect
that
there
are
some
pms
driving
this
work
just
to
like
broadcast,
probably
a
like
blog
post
on
the
medium
like
the
open,
telemetry
medium
tell
people
hey
like
we
think,
with
the
1.0
release
on
many
like
four
or
five
sdks.
We
think
we're
at
the
stage
where
we
want
to
get
user
feedback
broadly
and
go
and
read.
D
D
Something
like
issue
template,
but
we
need
to
work
with
and
math
to
make
sure
it's
consistent
across
all
the
other
reports.
Yeah.
I
G
D
Yeah,
I
agree
so
so
siri.
The
suggestion
is,
keep
this
in
mind
and
we
don't
need
to
create
that
issue
template
because
we
want
that
to
be
consistent
across
and
do
the
announcement
later.
We
might
need
to
think
about
like
because
this
this
thing,
like
a
one-time
thing,
so
one
big
announcement,
you
can
expect
a
lot
of
people
come
and
try,
but
later,
if
people
still
have
the
similar
thing,
when
they
read
the
document,
they
don't
know
how
to
follow
or
it's
hard.
D
B
Yeah
we
could
just
use
the
existing
combination
of,
like
enhancement,
tag
followed
by
an
additional
label
like
ease
of
onboarding.
That
can
be
the
hair
like
forever
like
any
day.
People
can
create
issues
in
that
thing
yeah
we
do
have
like
I
mean,
apart
from
all
these
research,
we
already
have
like
the
issue
which
I
mentioned
like
these
are
open
for
a
long
time
I
did
try
to
make
it
and
raj
is
also
trying
to
make
it
work.
It
occurred
yeah.
B
I
think
this
is
one
of
the
issue
like
we
want
to
do
that,
but
we
cannot
take
a
dependency
on
like
all
these
extra
packages
from
microsoft,
so
we
want
to
like
restrict
it
to
the
extensions
hosting
package
so
that
that
would
definitely
be
a
big
improvement
on
the
overall
onboarding
inside
again.
D
If
you
look
at
the
current
document,
it's
like
like
most
likely
we're
talking
about
hey,
like
we
assume
you
already
know
a
lot
of
about
open
time,
so
go
and
read
this
spec
or
go
and
understand
this
concept,
but
we
don't
have
a
good
end-to-end
scenario
like
imagine
if
the
user,
even
if
you
look
at
the
guy,
get
started
with
like
five
minutes
after
five
minutes,
what
he
got
is
a
console
log
and
what
people
most
most
of
the
folks.
What
they're
really
looking
for
is.
D
I
want
to
write
some
code
and
see
something
zero
like
give
me
a
jager
or
zip
king,
something
I
know
aloha
folks
like
michael.
He
has
his
personal
blog
doing
this,
but
we
might
need
to
consider
that,
like
here,
just
people
people
when
they
read
this
page,
we
give
them
something
very
simple.
After
five
minutes
they
can
see
something
viral
they
can
play
with
that
and
they
understand
the
value
of
doing
this.
That.
B
D
G
D
B
End,
so
can
you
take
a
look
at
this
issue?
Is
it
like
I'm
just
presenting
it?
This
is
something
which
sergey
wrote
like
long
back.
B
Is
this
like
the
kind
of
thing
which
might
help
like
this
is
like
it's
called
like
beyond
getting
started,
so
you
got
the
five
minutes
experience
you
saw
something
in
console,
which
means
you
did
something,
but
then
like
you
need
to
do
like
advanced
steps,
so
you
need
to
put
things
into
zipkin
or
more
useful
production
kind
of
thing
rather
than
console
and
then
like
adding
more
things
to
configure
sdk
to
achieve
different
things
and
so
on,
and
in
fact
this
talks
about,
like
all
the
advanced
concepts
in
a
single
entry,
it's
fairly
big
detailed.
D
G
D
They
have
to
discover
something
else,
then
they
probably
discover.
Oh,
I
want
to
build
the
extension
or
I
want
to
get
jager.
So
tell
me
like
if
you
read
this
document
after
you
finish.
Oh
congratulations.
You
can
see
this
cool
stuff
from
the
console
and
what
should
I
do
next?
I'm
live
okay,
yeah
and
then
the
answer
is:
go
and
like
like
scroll
up
the
docs
folder
and
go
to
the
examples.
Folder
there's
a
full
like.
D
Like
how
can
I
discover
that,
currently
it's
hard
right,
we're
already
too
familiar
like
everyone
inside
this
sick
meeting,
is
too
familiar
with
our
stuff,
but
I
suspect
most
of
the
new
user.
Would
we
need
help?
That's
why
I
see
people
ask
questions
from
from
historically
like
the
gator
channel
and
now
the
slack
channel.
B
Yeah,
I
was
wondering
like
how
like
did
python
or
any
other
language
have
like
code
this
like
any
better,
because
I
I
know
that
python
has
getting
started.
I
don't
know
whether
they
have
okay.
I
think
they
have
slightly
better
because
it's
a
continuous
document
after
they're
done
with
like
initial
hello
world,
they
just
continue
to
yeah
yeah,
okay,.
D
B
So,
let's
I'll
create
an
issue
and
maybe
like
start
working
on
beyond
getting
started
so
that
like
that
would
be
the
starting
point,
of
course
like.
If
other
ideas
are
there,
we
can
take
it,
but
basically
like
once
you're
done
with
this
document
at
the
end,
like
you,
you
you're
done
like
what's
next
and
then
take
you
to
the
next
level
yeah
exactly.
D
So
when
I
was
preparing
those
documents
I
feel
like,
I
I'm
probably
trying
to
repeat
the
mistake
that
I
I
made
like
10
like
20
years
ago.
God
knows
you
know,
I'm
fc,
like
I'm
saying.
If
you
use
video
studio,
use
the
wizard
create
a
hollow
word.
It's
awesome
like
five
minutes
and
after
that
you're
screwed.
You
have
to
look
at
all
the
message
voting
and
understand
the
132
apis.
J
Yeah,
I
I
come
in
and
out
and
go
I'm
still
here.
So
I
I
can
echo
a
lot
of
what
you
guys
are
saying,
I'm
on
an
enterprise
team
for
my
company
and
we
have
to.
We
have
the
fortunate
and
unfortunate
business
of
writing
a
lot
of
documentation
internally
and
a
lot
of
the
gripes
we
get
is
generally
just
the
the
bread
there's
like
two
ways.
You
want
to
look
at
it,
the
tutorial
view
and
the
reference
view.
J
Sometimes
you
just
get
the
point
where
you
know
enough,
but
you
want
to
know
like
the
dot-net
docs
like
there
was
a
c-sharp
library
do
and
you
want
to
go
right
there
and
figure
that
out,
but
and
then
have
the
tutorial.
Docs
stand
on
the
you
know,
shoulders
of
giant
so
to
speak
so
that
it
points
to
like.
J
If
you
want
to
learn
more
about
that
specific
thing
go
there,
and
so
I
I
guess
all
I'm
really
amounting
to
say
is:
if
that's
something
I
could
even
I
could
see
my
team
help
out
with
or
something
fast
and
I
won't
commit
to
anything
specifically,
but
that
that's
something
that
I
do
spend
a
lot
of
time
looking
at
and
if
I
were
to
start,
I
I've
had
gaps.
I'll
be
honest.
J
I've
had
gaps
between
like
six
months
when
I
started
a
long
time
ago
till
now,
and
now
I'm
kind
of
needing
to
catch
up,
and
so,
if
you
guys
want
to
get
some
feedback
on
where
to
go,
if
this
makes
sense,
because
a
lot
of
my
developers
don't
have
a
clue
much
about
open,
telemetry,
much
less,
even
how
that's
going
to
benefit
them
from
like
our
production
stage.
So,
like
I
see
alan
west
on
here,
but
we
have
like
new
relic
like
I.
J
I
don't
really
understand
what
our,
how
our
instrumentation's
going
to
help
in
the
long
run
like
like
there's.
So
many
there's
like
a
lot
of
pieces
here
and
there
from
zero
to
hero
kind
of
thing
start
with.
Oh.
J
Good
in
my
console
like
like
riley,
oh
riley,
right
yeah,
you
like
what
you
said,
then
what
then,
what
do
I
do
now?
So
it's
like
eventually,
once
you
learn
the
sdk,
then
you
want
to
go
the
next
level
and
then
you
talk
about
vendors,
discoverability
or
whatever
you
know
open
source
discoverability.
What
tools
can
you
use
to
comprise
of
a
full
offering
to
production.
B
I
won't
like
fill
the
whole
thing,
I'll
just
create
a
template
and
see
feedback
from
others,
whether
we
are
in
the
right
track
and
then
take
it
from
there
like
if
it
is
just
as
simple
as
like
having
a
continuity
after
you
done
with
your
get
started
in
five
minutes,
then
let's
learn
something
more
advanced
and
like
progressively
increase
the
concepts
and
increase
the
depth,
and
eventually
you
should
be
able
to
write
like
end-to-end
yeah.
I
like
it.
D
B
Yeah
I
mean,
I
think,
like
one
of
the
thing
which
we
I
personally
like
is
the
asp.net
course
document,
because
they
have
like
tutorials
for
like
creating
all
these
concerts,
but
throughout
the
document
they
refer
to
like
like,
if
you
are
requiring
any
in-depth
tutorial
on
anything
in
specific,
they
will
ask
you
to
like
what
exactly
is
dependency
injection,
so
it
will
take
you
to
the
actual
dependency
injection
talk.
This
talks
about
all
the
theory
and
all
the
functionalities
are
available
in
the
dependent,
but
you
don't
need
to
know
this
to
get
started
so
tutorials.
B
They
conveniently
avoid
this
detail
like
pretty
much
everything
like
options:
logging
routing
they
just
do
like
end-to-end
thing
and
linking
from
the
tutorials
like
getting
started
equivalent
to
these
advanced
concepts.
I
really
like
that.
I'm
a
big
fan
of
this
kind
of
thing.
So,
if
you
want
to
keep
it
simple,
you
can
always
like
you
don't
need
to
learn
like
any
advanced
thing.
You
can
just
do
like
in
five
minutes,
and
there
is
enough
to
keep
your
update
on
like
you
can,
like
click
certain
concepts,
and
it
will
take
you
to
the
next
level.
B
B
The
same
topic.
On
the
I
mean
and
the
same
topic
of
like
improving
onboard.
This
is
like
pure
talks.
I
think,
but
even
after
that
there
was
some
discussion
in
the
past
like
for
a
typical
asp.net
core
user,
we
could
introduce
potentially
like
meta
package
kind
of
thing.
Like
you
just
install
that
package,
it
will
bring
all
the
other
packages,
so
you
don't
have
to
go
and
individually
install
each
and
every
package
and
there
will
be
like
a
one
liner
instead
of
creating.
B
I
think
this
example
should
make
it
very
clear.
So
this
example
you
can
see
like
it's
fairly
big.
Like
you
add
all
these
things,
or
maybe
like
helper
methods,
which
is
also
in
the
theme
open
elementary
as
a
whole,
is
going
you'd
simply
have
a
single
method
like
services,
dot,
ad
open
telemetry.
B
It
will
take
care
of
like
adding
all
these
like
well-known
instrumentations
and,
of
course
it
can
be
like
a
convenience
api.
So
you
don't
need
to
go
and
do
it,
because
this
requires
like
installing
at
least
five
packages
like
the
sdk
two
or
three
instrumentations
and
one
or
more
exporters.
B
J
J
J
B
J
B
Yeah,
so
this
was
probably
because,
like
what
is
what
is
after
that,
like
you
got
started
in
five
minutes,
you
got
things.
What
is
next,
so
that's
where
I
was
kind
of
saying
like
you
will
continue
this
dope,
which
says
okay
beyond
getting
started
or
something
which
would
take
you
to
the
next
level
of
topics
which
I
am,
I
think
like
very
similar
to
what
this
blog
was
all
about.
B
The
first
one
was
very
easy.
You
get
something
despairing
concerning
same
with
what
python
is
also
doing.
It's
very
easy.
Like
you
just
get
started
with,
then
you
emit
something
to
console
your
five
minutes
is
done.
Then
you
go
and
explain
more
advanced
topics,
and
you
only
refer
to
like
spec,
where
you
think,
like
a
more
detailed
explanation
of
a
concept,
is
required.
Gotcha.
J
B
And
like
if
you
are,
if
you
have,
if
you
are
not
like
super
familiar
with
this
project
already,
then
you
are
the
best
candidate,
because
people
who
are
already
familiar
it
may
not
make
sense
for
them
to
like
review
the
dope,
because
they
most
likely
skim
through
the
details,
because
they
already
know
these
things.
A
B
F
Well,
nothing
really
to
share
per
se
immediately.
All
I
what
we're
working
on
or
what
I've
been
working
on
pushing
on
is
really
just
to
contribute
to
the
two
cigs,
the
api
sdk
sigs
for
metric
and
the
data
model
sigs
for
metric
and
in
those
two
sigs
they've
been
discussing.
You
know
key
issues
like
you
know,
they're
trying
to
figure
out
histogram
they're,
trying
to
I'm
trying
to
push
on
what
is
an
identity.
F
There's
some
questions.
That's
going
to
be
pending
regarding
how
we
pass
label
sets
how
we
use
label
sets.
There's
questions
about.
You
know,
data
types,
whether
we
pass
ins
or
doubles
so
stuff
like
that
is,
is
you
know
going
on
separately
to
that
as
c
joe
alluded
to
earlier?
F
There's
some
work
being
done
with
the
net
runtime
team
to
just
kind
of
start,
prototyping
or
really
not
even
prototyping,
but
just
trying
out
different
conceptual
concepts
in
terms
of
what
the
net
runtime
api
might
look
like
for
metrics
and
those
from
those
experiments
we've
been,
you
know,
surfacing
lots
of
questions
and
so
forth
with
no
answer
for
those
that
are
interested.
We're
welcome
to
share
the
the
the
you
know.
A
link
to
you
know
that
experiment
if
people
want
to
take
a
look
and
so
forth.
B
Far,
okay
yeah!
So
if
anyone
else
is
like
interested
in
like
working
closely
for
the
matrix
prototype,
please
reach
out
to
me
or
victor
in
jitter.
Oh
sorry,
not
jitter,
slack.
F
There
is
one
one
upcoming
item
that
the
the
data
model
sig
is
working
on.
There
will
likely
be
a
change
in
the
proto
in
the
in
the
you
know,
protocol
message
or
otlp,
so
those
that
are
involved
with
doing
exporters
may
have
you
know
some
changes
coming
down
the
pipe
once
those
things
are,
you
know
have
deemed
it.
You
know
good
enough.
F
B
F
Yeah,
sorry
c
joe
actually,
I've
heard
and
and
this
is
for
the
sake
here
to
confirm,
but
I
have
heard
that
the
there
are
production
level
code
currently
using.
You
know
metrics
otlp,
but
not
necessarily.net.
So
I
don't
know
of
anybody.
There
shouldn't
be
anybody
taking
the
net
metrics,
but
in
case
there
are.
B
Yeah,
we
don't
have
the
like
the
current
exporter
in.net,
it
does
not
support
exporting
metrics.
It
only
looks
at
traces
yeah.
We
only
have
the
prometheus
one,
which
is
the
only
one
which
people
were
using
at
least
based
on
my
knowledge
unless
they
have
written
their
own
exporter.
This
is
only
one
which
we
ever
had.
D
I
put
links
in
the
in
the
chat.
People
are
interested
in
the
metrics.
The
number
one
thing
is
to
understand
the
overall
open
climate,
matrix
goal
and
the
scope
and
the
timeline,
so
you
can
find
that
from
the
first
link.
Cj
it'll
be
great
if
you
could
put
the
link
in
the
meeting
notes
as
well.
The
second
one
is
what
the
metrics
folks
agreed
on
the
the
initial
like
prototype
scenario,
so
for
prototype.
We
focus
on
the
scenario.
If
there's
additional
need
that
we
we
don't
have
in
the
scenario.
D
That
means
it's
something
not
super
critical
and
we'll
put
that
later.
So
currently,
I
know
there
are
several
language
stakes
like
on
java
and
dotnet,
and
on
python
are
doing
the
prototype
and
we
list
all
the
findings.
The
goal
is
just
to
have
a
list
of
boolean
questions
like
do
we
do
this
or
not,
for
example,
do
we
want
all
the
metrics
name
to
be
case
sensitive
or
not?
Do
we
allow
any
type
of
the
names
and
do
we
treat
unit
as
a
must
have
for
the
api?
D
Do
we
like
support
both
double
and
the
end
or
everything
should
be
double,
so
I
expect
we'll
have
like
about
20
of
those
questions
and
then
we'll
make
decision
on
which
one
we
use,
and
some
of
the
things
might
be
interesting
to
you
guys,
or
it
might
actually
be
a
blogger
if
you
try
to
adopt
that.
So,
if
you
have
that
concern,
please
join
this
meeting
and,
like
the
matrix
meeting,
I
find
from
the
open,
telemetry
community
repo.
I
Okay,
yeah
good.
B
So
the
only
thing
like
we
mentioned
like
we
will
have
produced
by
like
mid
here
for
like
induces
to
try
out
but
like
since
most
people
who
are
joining
the
cigar
like
people
who
are
trying
to
contribute.
Then
you
have
an
opportunity
like
right
now,
because
victory
is
like
doing
prototypes
daily
and
working
very
closely
with
dotnet
runtime.
So
if
you
have
like
some
interest
or
some
feedback
or
anything,
that
now
is
the
right
time
to
engage.
F
I
don't
know
if
this
is
the
right
time
to
discuss
or
whether
it's
the
right
forum
per
se,
but
it
goes
back
to
your
first
question
regarding
depend
taking
dependencies
on
the
sdk,
so
I've
been
kind
of
tracking
the
an
issue
that
I
think
and
I
could
be
wrong,
but
I
think
the
spec,
as
is
specked
out
today
for
tracing
and
thusly,
also
impacting
metric,
is
that
currently
the
tracing
spec
in
hotel
doesn't
really
account.
For
you
know,
how
did
I
put
okay
to
get
more
specific?
You
know
you
need.
F
You
need
the
sdk
to
get
a
trace
provider.
It's
a
custom,
one
right.
B
F
Right,
yes
right,
and
so
so
I
I
ping
some
other
like
the
java
folks,
they
depend
on,
they
use
dependency
injection,
and
thus
my
question
earlier
is:
does
net
have
any
example
of
people
using
dependency
injection
to
apply
the
trace
provider
in
a
library
you
know
without
taking
a
dependency
on
sdk.
So
so
there's
a
lot
more
questions
involved
around
that.
But
suffice
it
to
say.
D
I
I
think
victor
asked
several
questions
or
I
misunderstood.
I
think
the
first
question
is
hey.
We
have
trace
provider
and
we
have
the
trust
provider
implementation
inside
the
open,
telemetry
sdk.
So
for
people
who
like
to
instrument
their
code,
can
they
just
take
dependency
on
api?
My
answer
is
yes.
If
you
try
today
open
telemetry
for
instrumentation,
you
take
dependency
on
api.
Nothing
with
sdk.
Now
sdk
is
helping
you
to
do
the
configuration
and
we
didn't
achieve
that.
D
We
failed
completely,
and
the
second
question
I
heard
from
victor
is
hey:
is
there
any
donut
dependency
injection
for
such
like
provider
thing?
I
think
ilogger
is
a
good
example
and
I
believe,
a
lot
of
folks
in
the
donut
runtime
team.
They
they
hate
that.
That's
why
they're
saying
for
metrics.
They
don't
want
to
use
the
di
again.
So
I'm
I'm
a
little
bit
confused
like
why
we
want
to
explore
the
di
here.
F
F
Right
and
and
and
then
to
link
that
to
the
first
question
that
you,
you
replied
or
that
you
you
know
refer
to,
is
if
we
don't
use
di,
is
there
a
way
today
in
the
in.net
implementing
the
hotel
trace
spec?
Is
there
a
way
to
get
something
other
than
the
default
provider.
B
Yeah,
I
think
what
victory's
asking
is
like
I'm
a
library
like
I
want
to
make
sure
the
traces
which
I
emit
from
my
library
goes
to
like
pipeline
a
not
pipeline
b
or
something
else.
It
is
possible
in
dotnet,
because
internet
we
recommend
people
to
use
the
diagnostic
source
api
directly,
so
they'll
just
create
their
activity
source
call
it
like
some
name.
Let's
say
I
am
library
a
and
then
another
library
could
create
another
activity,
source
called
library
b
and
in
the
final
application
we
could
configure
enough
tracer.
B
You
can
potentially
create
like
two
tracer
providers
and
have
the
first
one,
listen
to
library
a
and
have
the
second
provider
listen
to
library
b,
so
you
will
achieve
you
provide
an
example
for
that
for
oh
yeah.
It
should
be
like
very
straightforward.
I
can
write
it.
I
have
a
small
issue
in
my
hand,
so
I
am
working
one-handed.
So
it's
a
bit
slow,
but
I
pretty
sure,
like
we
have
examples
which
show
this.
So
let's
take
I'll
just
take
this
example.
B
B
Yeah,
so
you
have
to
use
some
imagination
right
now,
because
I
am
not
typing,
but
this
is
creating
one
provider,
which
is
only
going
to
look
for
traces
from
this
source
like
http
client
test,
and
let's
assume
that
this
code,
which
I
highlighted,
is
from
a
library,
it
creates
an
activity
source
with
that
name.
So
basically,
what
this
means
is
all
traces
from
this
library
go
through
this
pipeline.
B
Now,
if
I
want
like
another
library
to
create
like
http,
client
test2
I'll
just
like
do
that
in
that
library,
so
this
code
will
be
like
duplicated
in
that
code,
they'll
be
just
using
a
different
name
and
in
my
main
application
I'll
create
another
provider.
Instance,
let's
call
it
open
elementary
2
and
replace
the
source
name
with
http
client,
test2,
so
internet
you
can
kind
of
achieve.
Not
kind
of.
B
You
can
really
achieve
the
scenario
where
traces
from
this
different
libraries
go
into
different
providers,
but
the
general
question
is,
if
I
were
to
use
the
api
like
the
open,
elementary.net
api,
not
the
diagnostic
source,
api
you'd
only
be
able
to
get
a.
You
cannot
create
activity
source
which
is
equivalent
of
tracer
like
like
using
the
new
api.
B
You
have
to
get
the
tracer
from
a
provider
and
then
that
provider
is
a
question
is
like
how
do
we
get
like
provider,
one
or
provider
two,
because
the
only
option
is
to
get
provider
dot,
global
or
default?
B
So
I
don't
know
the
answer
like
how
other
languages
solve
it.
Yeah.
If
you
have
va,
then
the
let
the
library
take
a
dependency
on
ta
and
have
the
application
injected
for
that
library.
So,
if,
if
you
are
library
a
when
you
ask
da
hey
give
me
a
tracer
provider,
you'll
get
provider
for
a
and
if
you're
b
you
get
a
different
instance
of
the
provider,
but
that's
very
specific
to
how
that
language
implements.
But
in
general
I
don't
have
a
good
answer
to
the
original
program.
B
If
you
are
not
using
the
dot
nets
api,
where
you
can
create
activity
source
out
of
thin
air,
then
I
wouldn't
really
know
how
to
achieve
this
particular
scenario
which
would
just
describe
it.
Probably
it's
the
same
with
like
python,
at
least
based
on
the
example
which
I
tried.
B
Yeah,
it's
I
mean
for
me,
it
is
unsolved.
I
would,
I
think,
one
of
the
reason
it's
undefined
in
the
spec
is
it
maybe
left
us
a
language
specific
thing
like
go
figure
out
in
the
language,
whether
you
want
to
use
da
or
some
other
mechanism,
but
yeah
I
mean
I
agree
that
it's
underspecified,
I
don't
really
know
whether
it
will
be
ever
be
specified
in
the
spec
or
maybe
it
will
always
be
a
language.
B
B
We
just
put
this
into
this
project,
because
this
one
has
a
dependency
on
the
extension
start
hosting
abstractions
and
like
this
is
a
place
where
we
would
intend
to
add
all
the
data
related
stuff
like
if
you
want
to
configure
something
using
the
asp.net
core
da
approach,
we'll
restrict
it
only
to
this
project,
but
not
the
core
sdk.
J
I
think
those
are
pretty
clear
like
if
I
was
a
developer,
come
into
the
documentation
based
on
the
opinions
that
you
just
stated
either
di
or
no
no
di,
just
kind
of
highlighting
that
in
the
documentation.
What
what
are
you
here
for
what
you
know?
What
are
you
here
for?
Do
you
want
to
use
it
here
or
here
you
know,
you
know
just
some
examples.
I
was
thinking,
though,
because,
like
me,
I'm
biggest
thing
about
us
in
our
company
is
that
we
try
to
like
build
in
a
way
that
we're
not
invasive
to
our
consumers.
J
B
Awesome
yeah,
I
mean
it's
a
good
topic
to
discuss,
but
I
don't
have
the
solutions.
Oh
yeah
spec
would
eventually
even
either
eventually
specified
or
will
just
remain
silent
and
it
will
be
left
up
to
the
individual
languages
to
solve
it
in
their
own
way.
Yeah.
F
So
sijo,
you
know
to
the
question
about
you
know
after
the
five
minute.
What's
next,
I
think
this
topic,
since
you
already
have
a
solution
for
net
specific.
I
think
that
would
be
a
wonderful
thing
to
document.
B
Oh
okay,
okay
yeah!
I
should
I'll
put
the
template
and
I'll
definitely
include
this
part,
but
it's
not
yet
clear
to
me
like.
Is
it
like
a
common
scenario
for
the
libraries
to
send
their
laws
or
traces
to
a
different
pipeline?
B
F
B
J
B
By
the
way,
last
thing
is
like
I
have
put
like
milestones
like
in
case.
Anyone
wants
to
take
a
look
at
the
milestone.
We
don't
know
when
one
dot
one
stable
will
come.
B
So
I
put
like
beta1
and
beta2,
which
should
come
out
like
sometime
next
week,
the
first
version
and
the
one
in
april
I
think
most
likely
will
align
with
instrumentations
being
marked
as
stable,
that's
likely
when
we
will
do
1.1,
because
without
instrumentations
I
I
don't
yeah,
we
did
add
like
some
features,
but
probably
not
worth
releasing
like
1.1
without
install.
It
should
be
like
ready.
If
the
tracing
roadmap
is
to
be
believed,
then
it
should
be
ready
like
sometime
in
end
of
april.
B
So
in
another
two
months
we
should
have
1.1
along
with
all
the
instrumentations
released
us
to
be
decided
like
I
don't
know
whether
it
will
be
1.0
or
1.1
directly
yeah.
Okay,
thanks.
Everyone
like
see
you
again
next
week,
feel
free
to
reach
out
to
us
in
slack.
If
you
have
any
questions,
we'll
see
you
next
week,.