►
From YouTube: 2022-08-17 meeting
Description
cncf-opentelemetry meeting-2's Personal Meeting Room
A
A
A
B
To
get
rid
of
these,
this
menu,
like
whatever
this
ribbon
of
options
like
because
it
was
hiding
my
browser,
tab
titles,
so.
B
B
So
yeah
sijo
is
over
today
he's
on
a
vacation
this
week.
B
And
yeah
he
asked
me
to
host
the
meeting.
I
have
only
put
one
agenda,
but
so,
like
I'm
sure
I
think
we
have
a
lot
more
to
discuss.
A
C
B
Okay,
I
guess
we
can
start
now
so
yeah
one
one
of
the
things
that
I
wanted
to
talk
about
was
just
the
release:
Alpha
2
release,
which
would
include
the
new
Prometheus
exporter
packages.
The
alpha
was
one
version.
We
omitted
those
packages
and
we
didn't
release
them,
but
it'll
be
good
to
just
get
some
usage
and
get
them
tested
with
the
Alpha
2
version.
So
I'll
be
doing
that,
if
not
today,
maybe
like
in
the
coming
few
days
by
the
by
the
end
of
this
week
for
sure
foreign.
A
Actually,
on
the
note
of
the
release,
so
this
will
be
the
first
release
where
we've
well
we've
upgraded
diagnostic
source
to
V7
right
so
I
mean.
Maybe
we
already
have
changelogs
that
indicate
that,
but
this
may
be
the
beginning
of
where
users
either
see
that
warning
or
error
whatever
it
is
now
and
I'm
just
wondering
if
maybe
we.
A
B
A
B
B
Yeah
then
I
mean
like
we
should
definitely
come
up
with
like
a
good
text
in
the
change
law
which
will
inform
everyone
about
this
change.
So
you're
saying
you
can
like
draft
up
some
text
for
that
or.
A
B
I
just
saw
I
think
it's
in
API.
We
have
done
that
in
the
past,
like
I
just
saw
a
change
log
where
it
says
updated
system
diagnostic
source
to
version
600.
That's
in
the
API
change
talk.
A
Okay,
yes,
there,
it
is
sure
I
will
do
something
like
that,
but
I'll
be
a
little
more
verbose
trying
to
describe
kind
of
what
people
should
expect
in
different
scenarios,
specifically
like
that.net5
net
core
app
3.1
kind
of
scenario.
B
B
B
C
Haven't
released
Event,
Source
or
Siri
log
yeah
well,.
B
Either
yes,
so
do
we
want
to
like
release
the?
Are
we
waiting
on
something
or
like?
Should
we
be
doing
it
as
part
of
the
Alpha
2
release?
Please.
B
B
Okay,
yeah
I
mean
I,
think
I'll
ask
him,
maybe
he's
still
responsive
because
he
told
me
to
do
a
release
which
includes
the
Prometheus
packages,
but
then
yeah
I
think
I
forgot
to
ask
him
about
this.
His
reasoning
behind
holding
back
on
the
other
two
packages.
B
A
B
Okay,
so
yes
there's
another
issue:
that's
added
here,
I.
A
Just
added
that,
because
looking
at
those
packages
reminded
me
of
this
issue,
it's
it's
kind
of
long
and
I,
don't
know,
might
be
a
little
convoluted,
but
at
some
point
and
I,
don't
think
it's
a
super
urgent
thing,
but
I
think
it.
A
Some
of
the
questions
kind
of
thoughts
that
I've
shared
in
this
issue,
I
think,
would
be
good
ones
if
we
think
there's
anything
that
we
should
do
here
would
be
good
ones
to
resolve
prior
to
our
next
stable
release,
but
maybe
I'll
just
try
to
quickly
summarize
my
thoughts
here.
A
So
I
was
just
kind
of
taking
the
inventory
of
the
various
packages
that
we
have
across
both
the
core
repository
and
also
the
contribute
Repository
and
just
one
kind
of
outlining
those
observations
in
this
issue
and
then
specifically
for
like
the
the
packages
where
we
have
a
DOT
extension.
So
this
this
last
category
that
I
have
here
I've
shared
some
proposals
for
how
we
might
maybe
rename
some
packages.
A
None
of
these
are
strong
opinions,
but
I
just
wanted
to
get
folks
to
kind
of
think
about
naming
because
you
know
it's
always
hard
and
but
I
also
think
it's
important,
particularly
as
like
end
users
are
trying
to
orient
themselves
to
kind
of
what
things
are
and
what
they
should
expect
to
find
within
the
packages
that
we
publish
like,
for
example,
I'll
just
toss
out
one
one
concrete
example
like
the
open,
Telemetry
extensions
Docker
package
that
we
have
in
the
contrib
repository
right
now,
at
least
you
know
it's
the
only
thing.
A
You
know
there
may
be
one
argument
that
could
be
that,
oh
you
know
it
might
have
other
docker-like
functionality
in
the
future,
and
you
know
naming
it
generically
leaves
that
door
open,
but
you
know
I
think
the
other
Theory
folks
follow
is
that
you
know
packages
should
be
very
specific
about
what
they
do
so
like
for
the
docker
one.
One
of
the
one
proposal
that
I
toss
out
is
maybe
renaming
that
you
know
some
to
something
like
opentelementary.resourcetectors
dot,
docker
being
very
specific,
like
I
can
look
at
that
package?
A
Name
and
I
can
know
just
by
its
name
what
I'm
going
to
find
within
that
package
and,
in
any
case,
I
share
some
thoughts
about
some
of
the
other
things
I.
C
A
The
extensions
high
level
my
point
here
is
that
the
all
the
packages
where
we
have
the
name
extensions
in
it
it's
it's
kind
of
there's
a
variety
of
different
things
that
we
mean
by
that.
A
There's
this
notion
of
like
these
resource
detectors,
which
are
extensions
to
the
SDK,
and
then
you
know,
there's
the
serial
log
and
Event
Source
things
that
we
were
just
talking
about
and
though
those
are
not
really
extensions
to
well
I
mean
I,
guess
you
could
describe
them
as
an
extensions
to
the
SDK,
but,
and
we
also
have
another
notion
of
of
a
shim
that
is
in
one
of
our
other
packages.
I,
think
the
open
tracing
package
and
I'm
kind
of
posing
like
hey.
A
You
know
this
is
kind
of
like
a
shim
between
two
technologies
and
maybe
rather
than
extensions
in
this
case.
Maybe
shims
is
a
is
a
better
term
again,
not
strong
opinions
here.
Just
you
know
interested
in
people
kind
of
mulling
this
over
and
and
thinking
you
know,
should
we
get
a
little
more
aggressive
with
our
kind
of
what
we
mean
by
these.
C
A
That
we
have,
in
our
package,
names
I'm
going
to
stop
there
I'm
just
curious,
like
kind
of
people's
thoughts.
There's
a
lot
to
think
about
here.
B
Yeah
I
think
like
there
are
like
a
few
naming
of
few
names,
which
are
very
like
subjective,
open
to
interpretation
for
their
meaning,
but
so
that
it
makes
sense,
I
think
like
to
be
very
clear
about
what,
when
can
you
attach
a
certain
keyword
in
the
package
name
but
yeah?
B
So
I
have
just
just
have
a
question
so
like
most
of
these
anyway
have
not
had
a
stable
release.
So
even
if
we
end
up
deciding
to
change
some
any
of
their
names,
that's
is
that,
like,
okay
with
us,
because
like
we
were
thinking
of
reducing
breaking
changes
for
people
using
extensions
hosting
packages.
Now,
if
we
just
change
the
name
of
the
package
entirely,
then
that's
again,
some
changes
for
them.
A
Yeah
I
think
that's
definitely
something
to
consider
I
mean
yeah.
Changing
the
package
name
is
is
going
to
be
breaking
for
a
lot
of
these.
A
A
Before
November,
definitely
before
we
released
stable
versions
of
these
things,
you
mentioned
the
extensions
hosting
package.
I
I
actually
think
that
one
is
a
good
name,
I
mean
because
it
very
much
follows
what
you
know
like
the
Microsoft
extensions,
okay
kind
of
model,
and
that
is
what
is
what
it's
extending,
but
some
of
the
other
ones
that
are
newer
and
I,
don't
think
are
nearly
as
widely
used.
A
B
Yeah,
so
black
I
definitely
agree
with
you
here,
like
even
I
feel
this
is
more
of
a
shim
like
shim
would
probably
be
a
more
appropriate
name
than
extensions
for,
like
cellulog
and
even
source.
B
B
Yeah,
we
haven't
released
these
two
so
like.
If
we
want
to
change
the
names,
we
can
change
that
before
we
do
before
we
even
do
an
alpha
release,
but
like
yeah,
we
should
discuss
about
the
other
ones
as
well
like
what
extensions
would
mean.
A
Yeah
and
folks
can
continue
to
mold
these
over
and
I
mean
I
specifically
on
the
you
know,
the
serial
log
and
the
Event
Source
One
again
I
I,
wouldn't
be
super
concerned
about
releasing
them,
as
is,
and
then
renaming
it
to
shims.
But
you
know
if
other
people
are
concerned
about
that.
A
B
B
B
Yeah
so
I
think
we
should
maybe
like
start
discussing
and
then
maybe
we
can
document
like
when
to
use
extensions
and
to
use
instrumentation
I
thought
instrumentation
was
clear
at
like
and
yeah,
but
I
think
there
was
some
I
remember
like
in
some
PR.
We
had
some
discussion
that
even
counters
instrumentation
does
not
sound
right,
because
it's
not
actually
but
then
anyway,
that's
a
contract
package,
so
at
least
for
the
main
repo.
We
can
be
very
particular
about
what
these
words
mean.
A
Actually
yeah
the
event
counters.
One
I
think
I
noted
that
two
under
shims
scroll
down
I,
forget
I,
think
I
noted
it.
There.
A
Yeah
yeah
yeah,
so
that
very
much
to
me
feels
like
a
shim
as
well
rather
than
instrumentation.
B
It's
also
doing
like
some
sort
of
instrumentation
right
like
is
it
like,
like
how
runtime
instrumentation
is
emitting
these
metrics
using
those
eight,
the
runtime
apis
I
think
this
one
also
and
its
Matrix
using
so
this
one
lets.
You
do
multiple
things
as
I.
Remember
like
you,
can
create
an
event
counter
provided
counter
and
then
use
that
to
instrument
which
will
basically
be
shimmed
into
an
Hotel
instrument,
and
then
there's
also
like
these
other
metrics,
like
runtime
kind
of
metrics,
which
we've
encounters
M
emit,
which
it's
like
straight
away
emitting
so
yeah.
A
To
me,
the
word
instrumentation
describes
the
API
that
was
used
to
actually
instrument
code
and
so
like
to
your
point
with
the
runtime
stuff
yeah,
the
the
the
the
runtime
is
instrumented
with
the
event
counter
API
to
emit
its
metrics,
then
the
open
Telemetry
component,
basically
just
is
a
shim
which
transforms
That,
Into,
You,
know
open
telemetries
convention
so
that
that's
the
way
that
I
think
about
it
in
the
sense
of
the
event
counters
package,
the
open,
Telemetry,
instrumentation
event
counters
package.
A
It's
it's
I'm,
not
using
that
package
as
an
API
to
instrument
my
code
I'm
using
that
package
to
basically
work
as
a
shim
right,
because
I've
got
code,
maybe
somewhere
I've
instrumented,
my
own
code
with
the
event
counter
API,
but
now
I
want
to
be
able
to
pipe
that
through
into
the
world
of
open
telemetry,
and
that's
what
the
that
that's.
What
the
event
counters
package
here
enables
right,
at
least
to
my
understanding.
I
have
to
say:
I
haven't
actually
used
the
package.
B
Yeah
I
haven't
used
it
either
but
like
like,
as
far
as
I,
remember,
I
think.
Let
me
just
open
the
readme
folder
and
then
gone
trip.
B
B
Like
some
metrics,
which
are
coming
from
the
system
like
CPU
usage
and
stuff,
which,
where
like
no
explicit
user
involvement,
has
we
are
not
explicitly
adding
any
instrument
but
I
think
they
also
in
their
code.
They
also
have
something
where
they
convert
or
like.
If,
if
you
create
a
event
counter
provided
counter,
then
and
start
emitting
values
using
that
counter,
we
they
convert
it
into
an
Hotel
instrument.
B
So
it
does
that
shim
kind
of
work
as
well,
and
it
also
does
this
work
where
some
default,
even
counters,
which
are
emitting
these
kind
of
metrics
like
workings
at
CPU
usage,
GC
Heap
size,
it
emits
them
as
Hotel
metrics.
So
I
I'm,
not
sure
if
this
part
is
considered
as
assuming
or
is
this
just
instrumentation
that.
A
A
C
A
Right
like
these,
this
console
up,
but
I
I,
think
I'm,
following
this
correctly
so
like
it
says,
export
CPU,
Dash,
usage,
I'm,
assuming
that
that's
literally
the
event
counter
name,
that's
being
used
in
that.
That
name
is
not
like
an
open
Telemetry.
You
know
sanctioned
whatever
attribute
name
or
metric
name.
A
So
it's
literally
just
piping
that
through
and
I
think
what
this
component
does
is.
Just
it
simply
sets
up
the
it
like,
provides
a
handy
dandy
way
to
set
up
a
listener
for
an
arbitrary
set
of
event,
counters.
B
A
I've
I
think
it
would
make
sense
for
things
that,
like
you
know,
the
group
of
us
mainly
maintain
I
mean
I
guess
who
who,
whom
who's
the
maintainer
that
package
this.
A
I
think
the
distinction
and
I
kind
of
allude
to
it
in
in
the
issue
that
I
open
there
is
like
if
it's
a
vendor-specific
technology,
you
know
like
we
have
a
number
of
AWS
extensions
within
the
contribute.
Repository
I
have
zero
opinion
about
stuff,
like
that,
like
that
is,
should
totally
be
up
to
AWS
to
decide
you
know
naming
and
how
they're
like
basically
branding
their
packages.
Right,
though
I
think
it's.
A
You
know
it's
on
the
opencumber.net
community,
though,
to
maybe
help
guide
those
types
of
things
and
then
on
like
Technologies,
like
this
I
think
I
think
it's.
B
Okay,
yeah,
which
are
more
like
just.net,
specific,
which
any.net,
which
just
applies
to
dot
it's
not
window
specific.
Then
we
should
enforce
the
naming.
That's
is
that
what
you're
saying
like
since
event
counters
is
just
a
very
generic
thing
which
is
coming
from.net.
It's
not
like
yeah,
it's
not
coming
it's
not
vendor
specific.
So
we
should
be
once
enforcing
the
proper
naming
here.
A
Yeah
I
think
so
and
like
put
another
way
like
like
look
at
the
Hang
Fire
instrumentation
fragrance,
for
example,
you
know
like
what,
if
somebody
came
in
here
and
decided,
you
know:
I'm
gonna,
I'm
gonna
write
instrumentation
for
Hang
Fire,
but
I'm
going
to
rename
I'm
going
to
name
this
package
like
open,
Telemetry,
Dot,
Hang,
Fire
I,
think
we
would
come
in
and
like
highly
recommend
or
suggest
like
no,
you
should
have.
You
should
have
instrumentation
in
in
the
name
to
align
with
the
rest
of
the
things
you
know.
A
So
I
don't
think
it's
like
the
total
Wild
West.
In
that
sense,.
B
Cool
yeah
I
think
maybe
we
can
discuss
more
on
like
how
what
these
definitions
you
can
come
to
some
kind
of
agreement
on
what
these
terms
would
mean
and
then
start
enforcing
it.
B
B
Okay,
Michael
I
had
I
just
want
to
ask
you
a
few
things
about
the
pr
where
you
are
moving
things
away
from
hosting
package
and
adding
them
to
the
SDK
itself.
B
Yeah
actually
before
that,
I
also
wanted
to
talk
about
this
one
PR
good,
that
I
opened
that
PR
page.
So
there
was
this
issue
that
was
created
where
yeah
so
logs
could
be
exported
by
other
console
exporter.
Even
after
Locker
Factory
is
disposed
so
yeah
disposing,
a
log
of
actually
did
not
affect
the
console
exported
continue
to
print
out
logs.
B
So
here
in
the
Repro
is
using
here
on
setup
logging
exits
log
factories
disposed.
This
will
still
be
recorded,
but
then
I
think
cjo
had
commented
saying
that
this
is.
This
exists
outside
of
hotel
as
well
like
even
just
the
built-in
console
like
exporter
or
like
whatever
you
want
to
call
it
provider.
B
So
some
options
were
that
were
that
he
suggested
was
modify
console
exporter
to
not
export,
but
this
will
only
fix
console
exporter.
The
other
one
is
open,
modify
Hotel
logger
to
not
invoke
processor
pipeline
if
disposed,
if
the
factories
disposed
so
I
think
if
we
are
fixing,
then
maybe
this
is
a
better
thing
to
do.
The
second
one
where
we
just
modify
open,
Telemetry
logger
to
not
invoke
the
processor
pipeline,
but
I
also
saw
Michael's
comment,
I
think
on
this
PR
or
the
issue
itself.
Where.
B
So
yeah
this
would,
if
fixing
this,
would
cause
our
logo
provider
to
behave
differently
than
the
runtime
ones.
B
So,
like
yeah
I,
wanted
to
ask
like
your
opinion
and
like
people's
opinion
on
like
this
issue,
like
is
this
something
that
we
should
even
care
about
like?
Do?
We
even
have
to
fix
this.
B
B
But
foreign
also
I,
don't
know
like
if
having
a
different
Behavior
than
the
runtime
ones.
Is
that
something
which
we
don't
prefer
like.
B
C
When
I,
even
like
the
fundamental
logic
here,
is
okay,
they've
called
disposed.
So
we
shouldn't
write
to
the
console
but
I
question.
Even
that
notion
like
dispose,
doesn't
convey
any
sort
of
like
meaning
on,
like
you
shouldn't
execute
any
logic
after
dispose
is
called
all
dispose,
does
is
give
you
a
chance
to
release
managed
resources
right,
so
we're
doing
that.
We're
not
violating
like
any
disposal
of
memory,
we're
just
continuing
to
write
out
logs
as
they're
emitted,
which
I
don't
really
have
an
issue
with
I.
C
A
That's
an
interesting
point:
I
I,
okay,
first
first
I'm
gonna,
preface
this
with
I,
have
not
thought
about
this
deeply.
This
is
the
first
time
looking
at
this
issue,
but
so
so
I'm
I'm
inclined
to
agree
with
Mike
about
the
you
know,
there's
no
harm
in
continuing,
though
cars
you
bring
up
a
good
I,
think
another
line
of
conversation
which
I've
definitely
encountered
before
with
not
just
open
Telemetry
but
like
like
the
New
Relic
agent
products.
A
We've
definitely
had
customers
in
the
past.
Being
like
hey,
you
know:
we've
we've
spun
up
an
environment,
your
your
agents
attached
to
our
application,
it's
doing
a
bunch
of
stuff,
but
now
it's
overhead
is
is
too
high
and
we
can't
restart
our
application.
We
want
some
way
to
be
able
to
at
runtime
disable
some
of
the
Telemetry
that's
being
produced.
A
You
know-
and
maybe
in
this
case
logs
you
know
so
I
can
see
that
as
being
an
interesting
feature
of
of
the
SDK
to
pursue
I,
don't
know
that
I
would
solve
that
feature
through.
Like
a
you
know,
some
means
of
disposing.
It
would
probably
be
more
of
like
an
explicit,
like
you
know,
switch
that
you
could
like
set
to
disable
or
like
a
config
file,
that
you
would
change
and
then
it
would
be
pulled
or
something
by
the
SDK.
B
Mm-Hmm,
but
that
dispose
can
also
act
as
a
switch
I'm,
not
sure
like
if
you
like,
I
I,
don't
know
how
the
logger
gets
reference
of
the
locker,
Factory
and
stuff
like
I,
don't
know
the
implementation
details
of
it
in
my
step
but
like
if
all
we
need
is
like
a
Boolean,
then
if
something
is
disposed,
that
can
act
as
it,
but
yeah
I
mean
it's
not
super
explicit
that
is
true
like
like
disposing
a
factory
will
stop
the
logger
from
logging,
so
so
yeah
like
what
I
think
this
Pi
is
open
and
I
just
basically
also
want
to
ask
like,
even
if
we
fix
this
is
just
fixing
console
exporter,
something
that
we
even
want
to
do
like
I
mean.
C
If
the
logger
Factory
is
disposed,
it
should
perhaps
not
pass
the
logs
to
the
providers
which
would
fix
everyone,
so
they
wouldn't
execute
our
pipeline.
It
wouldn't
execute
any
of
the
other
providers.
Maybe
that's
more
appropriate
I,
don't
have
an
issue
with
I,
think
I
haven't
looked,
but
I
think
the
pr
got
updated.
So
it
just
logs
out
a
message
doesn't
seem
like
a
big
deal
to
let
that
go
in.
B
B
B
B
Yeah
I
yeah
I
mean
there's
no
other
agenda.
I
wanted
to
ask
a
few
questions
about
this,
so
here,
like
I,
think
this
might
have
already
been
discussed.
But
I
was
like
looking
at
the
piano
again
yesterday
and
today.
So
we
don't
want
to
create
an
extension
method
because
we
don't
want
to
create
any
extension
methods
to
service
collection,
apply
to
service
provider.
C
B
C
So
right
there
see
those
three
different
ways
to
start
a
host.
That's
the
basic
problem
is
I.
Couldn't
find
a
good
thing
to
attach
the
extension
against
which
would
work
for
all
the
different
ways
that
you
can
create
a
host
the
the
one
I
went
with
the
I
host
Builder
seems
the
most
appropriate
to
me,
but
it's
not
used
by
the
asp.net
core
builders.
C
Higher
level
is
how
do
we
auto
start
our
providers,
so
we
talked
last
week
about
introducing
a
used
quote
method,
because
that
feels
standard
right.
You
see
a
lot
of
code
that,
like
there's
one
line
where
you
add
your
services
and
there's
another
line
where
you
use
your
services,
so
we're
like.
Okay,
that
sounds
good.
Let's
try
that
then
I
went
in
to
try
it
and
I
was
like.
Oh,
no
there's
not
a
good
way
to
do
this
use
extension
because,
usually
typically,
the
use
is
attached
to
the
application
Builder
and
it's
configuring.
C
B
Can
we
like
I
think
like
what
about
just
the
service
collection,
just
the
services
dot,
add
Singleton
and
all
that
like?
If
we
have,
we
can
have
something
there
as
well,
but
the
user
has
to
ensure
that
that's.
The
last
thing
which
is
run
before
is
that
is
that
something
we
don't
want
to
do
like?
We
don't
want
to
enforce
like
the
so.
What
I'm
saying
is
like
we
could
have
had
a
extension
method
for
against
this
right.
B
C
The
main
problem
is
the
best
mechanism
to
auto
start.
Something
is
to
use
the
interface
I
hosted
service,
that's
part
of
Microsoft
extensions.hosting.obstractions,
so
that
is
really
at
after
this
PR.
The
only
thing
that's
going
to
live
in
the
hosting
assembly
package
thingy
is
that
I
hosted
service,
so
we
need
some
way
to
inject
it
into
the
service
collection.
C
That
used
to
be
the
ad
open.
Telemetry
tracing
extension
did
that,
but
now
we've
moved
that
extension
into
the
SDK,
which
is
missing
that
reference.
So
it's
kind
of
created
this
problem.
Where
now
we
don't
have
a
place
to
inject
that
thing,
but
even
injecting
that
thing
it
only
works
if
you're
using
some
kind
of
host
Builder.
So
even
the
way
we
had
it
before
it
actually
won't
start
in
all
cases
like
if
you
make
your
own
service
collection,
add
your
own
services
to
it
and
then
call
Dot
build.
C
So
you
make
your
own
service
provider
so
you're
using
dependency
injection
but
you're
not
using
the
generic
host.
We
won't
start
in
that
case
because
only
the
generic
host
understands
the
high
hosted
service,
so
it
was
sort
of
not
perfect
even
before,
but
it
probably
works
for,
like
99
of
what
people
do.
B
Okay,
also
like
so
what
other
benefit
is
coming
out
of
using
the
I
hosted
service
like
if,
instead,
we
just
build
it
here
when
we
are
like
configuring,
these
service
collection,
the
service
collection,
we
like
build
our
providers
and
then
it's
running
you.
C
C
C
We
could
do
it
with
reflection
like
in
the
SDK
ad,
open,
Telemetry
tracing
method.
We
could
reflectively
go
and
say:
okay
is
the
hosting
assembly
available.
If
it
is,
we
could
reflectively
dig
out,
I
hosted
service,
create
an
instance.
Dynamically
add
that
in
that
would
work
it's
just
kind
of
a
step
backwards.
As
far
as
we're
trying
to
remove
reflection,
we're
trying
to
remove
Dynamic
code
to
enable
things
like
ahead
of
time,
trimming
support
of
like
unity
and
xamarin
and
mono.
B
C
B
B
B
A
I
I
still
have
yet
to
study
the
actual
examples
super
closely.
But
if
I
recall
the
those
two
methods,
well,
okay,
so
backing
up
so
like
we
all
know
like
from
asp.net
core
3.1
through
like
five
and
then
six
was
kind
of
a
substantial
change
to
kind
of
the
patterns
for
booting
up
a
asp.net
core
app.
A
These
two
methods
weren't,
they
weren't
their
purpose
to
kind
of
serve
those
two
patterns
like
setting
stuff
up
just
within
program.
Cs
versus
you
know
the
old
like
program,
CS,
Dot
and
startup.cs
am
I
making
any
sense.
Yeah.
A
Yeah,
so
with
this
one,
this
one
is
an
example
of
like
a
modern.net6
and
using
that
use
open,
Telemetry
tracing,
but
then,
if
we
search
this
PR
for
a
usage
of
the
other
one,
it's
probably
using
the
the
other
pattern
yeah.
So
like
this
microservices
example,
it's
got
a
startup.cs
file,
yeah.
C
A
C
B
A
B
That
does
not
okay,
so
this
is
against
the
I
host
interface,
so
this
is
I
host
yeah.
This
is
that.
C
C
That
app,
that
VAR,
app
I
think
is
a
web
application
instance.
So
you
could
down
here,
say:
I,
think
dot,
host
dot
use
open
telemetry,
so
you
could
use
the
other
one
here.
You
would
just
need
two
lines.
B
C
B
Yeah
but
then
I
think
we
just
would
need
more
examples.
I
feel
like
like,
if,
like
like
how
Alan
mentioned
someone
is
upgrading
from
I.
Don't
know
what
was
it
net
five
to
net
six
where
they
had
startup.cs?
They
didn't
change
the
like
if
they
always
know
that
you
have
to
end
up
calling
this.
If
we
make
this
a
standard
thing
to
like
start,
the
providers,
then
I
guess.
C
I
mean
let
me
double
check
that
it
works
and
then
I'll
I'll
update
it
for
that.
B
Yeah
now,
like
I,
mean
I
think
this
definitely
sounds
much
better
host
dot
use
open,
telemetry
as
opposed
to
service
provider,
dot,
use,
open,
telemetry.
C
So
what
you,
if
we
put
that
extension
on
the
service
provider,
what
you
have
to
write
here
is
like
app.services.use,
open
Telemetry,
so
it
doesn't
really
help
you
too
much.
You
could
also
where,
where
it's
injecting
the
I
host
into
this
method,
if
you
just
injected
the
Tracer
provider,
it
would
accomplish
the
same
thing
like
the.
If
you
go
look
at
the
source
to
that
use
open,
Telemetry
extension,
it
doesn't
really
do
anything
other
than
just
request.
The
Tracer
provider
out.
A
C
I'm
not
opposed
to
that
I
have
seen
in
newer.net
core
you'll
notice,
there's
a
lot
less
start:
Stop
open,
close
methods,
that's
all
just
kind
of
gone
to
the
Sea
tour
and
the
dispose.
So
you
we
kind
of
broke
with
that
pattern
that
was
kind
of
sprinkled
around
framework.
Some
things
use
start
stop
some
news
begin
in
some
use
open,
close,
it's
sort
of
gone
away
in
the
newer
apis.
It's
just
always
disposed.
Does
the
stop
or
close
or
whatever
needs
to
happen,
but
I'm
not
opposed
to
that?
Actually.
B
A
C
B
Yeah
so
maybe
like
instead
of
I,
had
open
Telemetry
tracing,
it
could
be
configure
open
today
and
finally
use
it
after
you've
configured
it.
B
So
just
one
more
question,
Michael
I
think
so
like
if
we
always
call
this
in
the
host.
That
means
the
Services
by
this
point.
If,
by
this
point
the
services
are
already
added
to
the
collection
right,
all
these
houses
are
registered
yeah.
B
So
then
do
we
still
need
a
hosted
service
to
okay,
I.
Think
I'm
missing
some
point
here
like
but
yeah
I'll
just
ask
again.
So
if
all
the
services
are
registered
at
this
point
before
we
like
enter
this
function,
so
do
we
still
need
a
I
hosted
Service
registration
for
open
Telemetry
to
start?
Could
we
not
just
like
build
in
this,
and
so.
B
B
B
Okay,
so
for
this
one,
do
you
require
the
hosting
package,
extensions
or
hosting
package.
B
C
C
A
A
B
Cool
then
yeah
I
think
we
are
yeah
before
we
go
over.
So
all
right,
then
I
have
a
last
video
about
the
packages
before
releasing
in
yeah.
That's
it.