►
From YouTube: 2022-04-26 meeting
Description
No description was provided for this meeting.
If this is YOUR meeting, an easy way to fix this is to add a description to your video, wherever mtngs.io found it (probably YouTube).
A
A
A
And
we
don't
have
anything
in
the
agenda,
so
if
anyone
has
questions
or
anything,
we
can
go
over
it.
I
think
nothing
happened
since
last
week.
Actually
we
did
merge
the
1.3
beta
1
and
did
the
actual
release
of
you
confirm
if
I
actually
did
that.
So
I
think
I
did
1.3
one
yeah.
A
B
Put
out
one
pr
yesterday
that
maybe
I
wouldn't
mind
talking
about
briefly
and
I
actually
converted
it
to
a
draft,
because
after
a
few
comments
from
riley,
specifically
the
last
comment
so
context.
This
pr
is
one
attempt
to
introduce
the
notion
of
limits
on
just
span
data
so
like
limiting
the
number
of
attributes
limiting
attribute
length
when
they're
strings,
and
this
is
doing
it
via
the
shim
api.
B
About
this,
a
couple
weeks
back
really
brought
up
a
good
point
that
another
way
to
handle
this
may
be
just
to
simply
do
it
in
the
exporter.
So
there's
a
number
of
reasons
why
that
idea,
intrigues
me
the
maybe
one
of
them,
the
main
one
being
that
per
the
spec.
B
C
B
If
an
application
had
multiple
trace
providers,
you
know
you
could
have
a
different
configuration
value
for
these
things
that
wouldn't
be
possible
with
this
pr,
but
it
would
be
if
the
exporter
just
honored
this
stuff-
and
I
just
was
curious
about
I-
I
was
going
to
go
ahead
and
just
put
this
in
draft
and
then
take
a
stab
at
doing
that.
But
I
was
curious
if
any
other
folks
had
thoughts
or
concerns
about
one
approach
over
the
other.
B
A
Downside
with
doing
this
in
exporter
is
like
every
exporter
would
have
to
do
that
if
they
care
about
right.
Limiting
this
thing
most
likely
we'll
only
care
about
otlp
to
begin
with,
but
still
general
thing.
If
sdk
is
not
doing
it,
then
every
exporter
has
to
do
it.
So
that's
that's
a
general
problem
not
specific
to
this
one.
A
I
I
mean
another
thing
was
I
I'm
just
still
going
through
it.
So
is
the
primary
goal
to
prevent
back-ends
from
getting
like
too
many
things
and
getting
throttled.
Is
that
the
primary
idea,
or
is
the
primary
idea
like
to
put
because
we
have
this
dimension-
capping
max
matrix
inside
the
matrix
sdk?
A
A
B
Yeah
so
yeah.
Ideally,
I
think
that
this
would
be,
and
I
opened
up
an
issue
with
the
the.net
runtime
project.
Yeah.
B
I
think
it
solved
there
because
I
think
that
that
would
be
the
place
where
both
of
these
concerns
would
be
solved.
I
think
they're
both
legitimate
concerns
that
can
be
achieved
via
the
same
mechanism.
So
you
know
I'm
coming
at
this
from
the
standpoint
of
the
back
end,
and
you
know
I
work
for
neural.
Like
neuralic
does
limit
attributes,
I
don't
think
neuralix
the
only
back
end
like
this,
but
basically
we'll
drop
data.
B
If
a
span
has
too
many
attributes
or
attributes
that
are
beyond
some
certain
length,
so
we
advise
customers
to
set
these
things
because
they're
configurable
things
in
the
specification,
it's
just
that
the
net
sdk
doesn't
currently
support
it.
So
we
have
java
customers
that
are
successfully
doing
this
and
and
so
on.
B
The
configuration,
I'm
pretty
sure,
is
done
by
the
sdk,
but
it's
it's
honored
by
the
api.
Okay,
so.
B
A
B
Right
and
I
think
that
if,
if
the
net
team
introduced,
you
know,
I
just
proposed
some
like
static
configuration
on
the
activity
source
class.
If
they
did
something
like
that,
then
we
could.
We
could
have
some
similar
extension
methods
off,
like
tracer
provider,
like
the
set
max
streams
and
stuff
that
basically
just
set
that
static
state.
Maybe
okay,.
A
A
Or
the
alternate
option
is
just
do
it
in
the
otlp
exporter
for
now,
which
would
only
be
honored
by
the
otp
exporter,
but
it
won't
prevent
the
original
issue
of
like
too
many
too
much
memory
usage.
B
Right
and
I
think
the
too
much
memory
usage
is
going
to
become
it
may
not
be
there-
maybe
not
be
concrete
examples
where
that's
a
problem
today
I
don't
know,
but
I
think
it
could
become
a
example.
I
think
james
newton
saw
brought
this
up
in
the
new
james
newton.
I
was
calling
him
james
newton
socked.
B
He
he
brought
this
up
in
the
context
of
the
grpc
spec,
which
enables
you
to
add
an
unbounded
number
of
unbounded
number
of
events
to
inactivate.
A
B
Yeah,
I
think
I
think,
we're
again,
probably
not
a
huge
problem
today,
but
I
think
where
people
that
may
start
running
into
a
problem
is
if
they
consume
a
third-party
library
that
has
instrumentation
that
is
prone
to
adding
stuff
something
they
don't
have
control
over.
A
A
A
So
if
we
cannot
really
implement
that
in
the
sdk,
if
people
are
using
the
activity
api
because
we
don't,
you
would
only
know
about
how
many
events
are
added
when
we
run
the
export
process
when
we
on
some
processor
okay,
so
we
won't
be
able
to
do
this
in
the
sdk,
even
if
you
want
to
right
now
right,
okay,
yeah,
but
now
yeah.
No!
No!
A
Now
I
see
the
whole
problem
so
with
apa
we
can
do
it,
but
it
won't
like
really
block
anyone
from
using
the
activity
ap
directly
and
still
having
the
same
issue.
Okay,
yeah.
I
can
see
like
pro
both
approaches
having
their
own
issues
and
what
was
the
reason
why
dot
net
decided
not
to
do
it.
B
I
actually
just
saw
that
this
guy
closed,
so
I
haven't
actually
looked
at
this
super
recently.
A
A
Yeah
I'm
thinking
like
if
we
want
to
do
this
in
sdk.
We
cannot
do
this
without
ap
support
like
we
need
to
have
like
some
callback,
but
that
would
be
like
very
expensive
right.
I
mean
if
you
I
was
imagining
like
some
callback
on
the
set
method
that
would
allow
was
to
subscribe
to
it
and
keep
the
count
and
once
the
count
is
reached
we'll
have
to
like
return
false
or
something
so
that
dotnet
will
start
ignoring
it.
Okay,
yeah.
A
Like
list
like
listing.net
or
dictionary,
it
can
just
keep
adding
you
can
add
like
any
number
of
items
you
do
it
right.
So
that's
probably
what
he's
referring
to
in
general.net
for
any
allocation
any
objects
here,
that's
my
understanding,
but
maybe
yeah
in
general,
don't
limit
the
resources
for
any
allocations
in
any
object.
Yeah.
B
B
So
you
know
the
static
configuration
that
I
proposed
would
affect
all
consumers
of
the
activities
which
I
think
makes
the
solution
less
than
ideal,
but
I
think
that
it
would
become
dramatically
more
complex
if
going
beyond
like
the
static
config,
like
I
thought
about
like
if
if
this
would
be,
if
this
were,
if
this,
if
you
could
specify
this
at
that
activity,
listener.
A
A
Okay,
that's
very
interesting!
Maybe
it's
still
possible
because
we
just
need
to
like
work
through
this
a
little
bit
more
because
let's
say
that
the
user
did
not
open
elementary
tracer
provider
did
not
set
this
limit
and
will
not
register
that
callback.
So
there
should
be
like
no
perfect,
but
if
there
is
a
user
who
is
setting
these
limits,
then
only
we
will
subscribe
to
that
callback
and
then
from
that
callback
we
will.
I
don't
know
what
that
callback
would
look
like,
but
something
like
total
count
seem
till.
A
Then
or
returns
like
true
or
false,
just
thinking
about
cloud
like
if
it
falls,
then
dotnet
will
just
ignore
that
set
tag
so
that
in
the
sdk,
the
callback
we
can
simply
check.
What
is
the
actual
limit
right
now
versus
the
configured
limit
and
based
on
that
return
to
workforce,
and
then
dotnet
can
choose
to
ignore
it.
A
That
way
like
there
is
no
perfect
for
folks
who
don't
use
this
feature
and
the
feature
is
still
implemented
in
the
sdk.
We
are
just
asking
for
a
new
api
to
get
a
call
back
when
and
whether
we
use
it
for
like
dropping
the
tags
for
based
on
the
count
or
something
else,
that's
totally
up
to
the
sdk.
A
A
Return
value
which
indicates:
should
they
store
it
or
should
they
or
you
I
mean
the
reason
why
I
asked
that
is
there
were
like
other
issues
about
the
secrets
or
credentials
being
part
of
the
activity
tags,
or
even
I
logger.
A
So
I
I
saw
like
someone
proposing
like
some.
Maybe
it
was
blanche,
mike
blanche
who
mentioned
like
something
like
this,
because
that
way
like
we
can
prevent
any
secrets
from
being
accidentally
set,
because
whenever
there
is
a
set
tag,
there
is
a
callback
which
gets
the
key
and
value
and
they
can
do
the
validation
right
there,
instead
of
like
doing
it
later
after
the
fact
doing,
in
the
sdk
or
in
the
exporter.
B
Yeah,
that's
interesting.
Actually
I
think
I
I
think
I
might
remember
something
along
these
lines
too.
I
think
it
was
something
about
filtering
like
the
ability
to
to
filter
out
attributes
or
or
filter
out,
maybe
maybe
what
I'm
recalling
is
being
able
to
filter
out
entire
spans
like
at
the
using
a
processor,
but
it
seems
like
you
would
want
to
use
a
sampler
in
that
case,
but.
A
This
would
be
more
like
you
don't
want
to
drop
the
entire
span.
It's
like
we
just
have
like
some
attributes
or
some
events
within
that
which
is
accidentally
containing
something
which
we
didn't
intend
to
take
a
credential
or
something
yeah
yeah.
So
it's
not
like
dropping
the
whole
thing.
It's
just
dropping
one
tag
or
replacing
it
with
something
else.
A
So
this
callback
on
set
tag
would
exactly
allow
that,
because
you'll
get
the
original
set
tag,
arguments
like
the
key
and
value
and
we
can
return
either
return
the
same
essays
or
we
can
return
the
modified
value
after
replacing
the
secret
or
we
can
say,
like
simply
drop.
The
whole
thing
like
by
like
returning
now
to
indicate
that
like
so.
A
If
this
is
apa
which
dot
net
exposes,
then
we
could
the
sdk
could
leverage
it
for
and
like
the
secret
hiding
purpose
and
also
we
can
implement
the
attribute
limit
using
the
same
mechanism,
because
we
simply
check
how
many
attributes
are
added
till
then
and
if
it
is
exceeding
the
limit
we
return.
Also
dotnet
will
just
ignore
it.
A
This
open
elementary
scenario,
so
I
think
such
a
proposal
might
be.
A
More
acceptable,
I
mean
I
I'm
not
saying
like
they
will
accept
it.
I
just
think
that
there
is
more
use
cases,
and
maybe
we
cannot.
I
can
find
that
issue
where
people
are
asking
for
the
credential
detection
things
okay
and
see
if
we
can
like
get
more
use
cases.
That
would
be
a
more
convincing
story
because
then
we'll
solve
like
multiple
things
in
one
shot.
B
Yeah
I
like
that
idea.
I
I
can
I
can,
if
you
find
that
link
that
would
be
yeah.
C
A
A
A
A
People
generally
ask:
how
do
I
accidentally?
How
do
I
prevent
accidental
leaking
of
potential
or
something
sensitive?
Okay,
and
we
have
like
different,
disconnected
approaches
like
we
have
the
general
activity
processor
and
we
have
the
instrumentations
which
have
this
filter
idea
and
which
can
or
like
the
enriched
idea
where
you
take
yeah
exactly
this
like
you
can
overwrite
one.
So
these
are
like
all
disconnected
ideas,
but
if
we
get
some
call
back
like
that,
then
we
could
offer
a
universal
thing
from
the
sdk
itself.
A
So
everyone
can
use
that
as
opposed
to
using
any
like
themes
like
building
their
own
individual
features.
So
maybe
like
we'll
need
to
spend
some
time
just
to
build
a
stronger
case.
Then
we
can
solve
that
if,
if
that
api
is
approved
but
like
with
the
use
cases
being
more
and
more
strong,
I
think
dotnet
would
have
like
more
like
willingness
to
accept
things.
B
A
Sounds
good,
but
for
now
I
I
think
to
immediately
unlock
the
finale
which
you
were
referring
to.
We
could
do
that
will
be
exported,
but
then
the
issue
is,
it
will
be
an
otlp
exporter,
specific
setting
right
like
you
like
something
in
the
other
exporter
options.
B
Yeah,
I
was
thinking
about
that,
so
so
a
few,
I
guess
the
the
couple
of
options
that
I
was
thinking
about,
so
either
a
go
ahead
and
do
this
in
all
the
exporters
for
now,
which
I
think
would
be
you
know,
otp,
jaeger
and
zip
can
and
then
expose
configuration
like
that
set
metric
stream
type
of
stuff,
but
you
know
the
tracer
provided
prior
level.
I
think
that's
maybe
option
one
option.
Two.
All
of
these,
all
of
these
things
should
be
configurable
via
environment.
B
Yeah,
so
we
could
only
expose
it
via
environment
variables
and
we
would
not
need
to
expand
any.
You
know
programmatic
thing,
I
think
that's
less
than
ideal,
but
I
I
think
it
would
it
it
would.
It
would
serve
a
need
for,
I
think,
a
good
number
of
people
and.
A
B
Yeah
correct,
or
I
mean
in
that
case
we
could
do
jager
and
zipkin
as
well.
I
mean
there
would
be
nothing
that
would
prevent
us
from
doing
that,
but
it
would
just
mean
that
we
wouldn't.
A
Have
the
short-term,
which
is
to
prevent
items
being
dropped
by
the
back
end.
C
A
Find
like
any
more
use
cases
like
I
go
through
this
because
debug
is
from
microsoft
and
he
has
this
requirement
and
I
think
he's
doing
like
some
work
around
right
now
internally,
so
maybe
like.
If
you
have
a
more
stronger
case,
we
can
try
to
see
if
how
he's
doing
that
then
see
if
he
can
make
it
a
stronger
case.
Store.Net
okay
looks
like
it
was
assigned
to
someone
who
wanted
to
do
it,
but
then
yeah.
It
never
happened.
Okay,
okay,
so
yeah.
A
Let's
keep
this
draft
and
let's
explore
the
idea
of
doing
that
in
the
exporter
to
begin
with
and
slowly
see,
if
we
can,
if
we
can
build
a
stronger
case
to
have
it
in
the
api
so
that
we
can,
the
sdk
can
implement
something
on
top.
A
Yeah,
nothing
else
from
the
agenda,
so
yeah
I
mean
just
other
updates.
I
was
mostly
trying
to
work
on
the
otp
log
exporter
because
that's
that's
the
only
thing
which
is
like
really
missing
for
us
to
say,
like
logging
is
also
done
apart
from
like
features
of
course,
but
like
in
terms
of
overall
signals.
A
A
The
names
would
change,
of
course,
because
it's
not
stable
yet,
but
at
least
that
would
help
us
like
say
that
people
can
start
using
it
and
that
I
said
I
don't
have
anything
in
the
agenda,
so
we
can
I'll
just
put
some
notes
about
the
spr
which
we
just
discussed,
and
then
we
can
meet
again
next
week.
B
Just
a
minor
thing
about
the
tlb
log
exporter:
maybe
what
do
you
think
about
in
the
next
release,
aligning
it
with
the
versioning
of
the
other
core
components?
C
A
A
A
Not
like
perfect,
but
I
think
it
was
the
easiest
way
to
achieve
this.
So
what
we
did
was
we
had
1.2
rc
4
rc5,
just
like
the
core
sdk,
the
sdk
went
1.2
stable,
so
this
one
also
produced
the
package
1.2.0,
but
we
are
uploading
it
to
manually
two
music.
So
we
skipped
it.
Then
we
still
get
1.3
so
now
there
is
no
stable
version
of
this
thing,
but
it
feels
a
little
bit
difficult
to
go
from
like
release
category
to
beta.
A
You
already
commented
about
that
in
the
issue,
but
this
is
the
easiest
to
implement
like
so
that
we
don't
need
to
like
implement
any
new
thing,
because
the
other
alternate
option
would
be
to
automate
this.
Maybe
that's
a
good
thing,
like
yeah
michael,
had
a
pr
to
do
that.
A
Now
that
I'm
like
little
bit
free
from
all
the
releases,
maybe
I
can
like
figure
out
that
pr
and
see
if
he
can
do
it
for
package
release
all
the
way
till
you
get
it
with
minimal
manual
touch
that
would
be
possible
yeah.
A
B
I
had
proposed
that
as
an
idea
on
a
different
issue,
but
right
now
I
mean
I'm
just
focused
on
the
logs.
I
can
just
because
I
think
it
would
make
yeah.
It
definitely
makes
sense
to
align
that
just
with
the
with
the
standard
like
otlp
exporter,
yeah.
A
This
is
a
these
are
like
spec
mandated
things
so
we'll
stick
with.
We
treated
treat
them
as
core,
except
that
we
want
to
release
it
stable,
so
it
will
always
be
like
aligned
in
terms
of
the
first
three
digits
with
the
core
one.
Okay,.
B
B
A
A
So
I'm
still
expecting
it's
not
going
to
happen
like
even
in
summer,
because
the
original
plan
was
to
have
it
in
like
march
1st.
That
didn't
happen.
So.
A
Yeah
there
was
this
special
sig,
maybe
it's
still
there
yeah
there
was
this
instrumentation
saying,
but
I
don't.
I
don't
think
there
was
any
mention
about
like
actual
progress,
so
yeah,
so
we'll
have
like
plenty
of
time
to
figure
that
out
and
we'll
deal
with
it
when
it
actually
comes.
A
All
right,
I
think,
that's
it,
oh
by
the
way
I
just
have
a
q
question.
So
do
you
know
like
if,
like
anyone
he's
using
the
otlp
log
exporter,
because
when
you
open
the
first
pr
for
otp
log
exporter,
I
saw
like
some
20
plus
people
like
water
comes
out.
A
I
I
just
wanted
to
know
whether,
like
people
are
actually
using
it
all,
because
I
know
that
within
microsoft
we
are
still
I
mean
our
otp
capabilities
are
only
limited
to
prices,
so
there
is
no
logging
or
tlp
receiver
for
us,
it's
still
being
built.
So
I
don't
have
like
anyone
actually
who
can
validate
this
thing.
So
do
you
know
if
new
early
customers
are
using
it
or
planning
to
use
it
once
we
usually
pay
more
usable
one.
B
A
I
mean
I
mean
I
don't
know
whether
you
have
like
ability,
I
mean
any
telemetry
within
neural
link
to
know
how
many
logs
have
this
attribute.
That
will
tell
us
that,
okay,
it's
not
right,
that's
actually.
Actually
we
should
have
the
resource
already
right.
The
resource
should
already
I
mean
I
don't
know
whether
it
accepts
results.
A
A
Yeah,
that
seems
sure
like
brother,
like
because
we
did
not
support
scopes
until
like
last
week,
but
we
haven't
really
shipped
it.
I
just
merged
that
beer
just
to
make
progress,
but
I
I
don't
really
know
whether
that
would
be
acceptable
to
everyone,
because
scopes
can
be
like
very
noisy,
especially
if
you
run
it
in
asp.net
core,
like
spinner
core
itself
at
some
scopes
and
user
can
add
their
own
and
I'm
right
now
putting
everything
into
attribute
to
me.
That's
the
only
place,
so
let
me
quickly
show
that.
A
Yeah
I
saw
that
pr-
I
put
a
screenshot
here
this
one,
so
it
will
be
like
scopes
here:
scope,
one
scope,
two
scope,
three
and
in
a
typical
sp
network
application.
I
think
you
would
see
at
least
two
depth
two
level
with
three
plus
attributes
in
each
step.
So
you
are
essentially
like
adding,
like
maybe
like
six
to
eight
attributes,
even
though
you
probably
only
care
about
the
scope
which
you
added
yourself,
not
the
one
from
framework.
So.
A
Sorry
yeah,
I
was
looking
at
this
showing
this
pr
where
like
when
you
enable
scopes.
This
is
what's
going
to
happen.
Like
I
mean
this
is
a
statement.
You
have
scope
from
two
depths
or
two
nested
scopes
and
yeah.
In
turn,
what
it
becomes
is
scope,
0
attribute.
I
mean
attributes
prefix
with
scope.
0
then
prefix
with
scope
1..
A
So
this
has
essentially
resulted
in
like
1
two
and
the
template.
So
three
attributes
just
by
the
user
thing,
and
assuming
that
this
is
in
sp
net
core,
then
spinner
core
has
its
own
scope.
So
your
scope
would
be
more
like
three
or
four
three
or
four
and
scope
zero
and
one
two
three
would
likely
be
the
one
from
asp.net
core
itself.
A
So
the
question
is
like:
do
people
really
want
to
pay
the
cost,
because
these
are
like
ingestion
like
you,
you
are
transferring
this
over
the
wire
and
you
are
paying
for
some
cost
to
your
vendor,
so
it
wasn't
sure
like
what's
the
best
way
to
solve
it,
so
I
just
did
the
dirty
easy
way,
but
then
I'm
curious
like
if
any
neural
customers
like
started
enabling
scopes
and
they
start
complaining,
then
we
need
to
like
provide
slightly
different
way
to
deal
with
scopes
like
instead
of
like
taking
scopes.
A
As
is
we
should
like
to
use
our
pick.
Okay,
I
want
only
scopes
for
my
immediate
scope
or
my
immediate
end
scopes,
something
like
that
sure,
but
yeah
I
just
want
to
like
hear
from
people
before
we
do
like
any
any
any
any
feature
like
that.
So
that's
why
I
just
wanted
to
electrically
ship
this
thing
like
people
use
it
and
then
complain.
A
Then
we
can
solve
it
a
very
complex
problem,
in
my
opinion,
even
though
it's
very
easy
to
use
scopes,
it's
very
difficult
from
a
logger
provider
standpoint,
because
scopes
can
be
like
arbitrarily
depth
and
it
can
be
anything
it
can
be
key
value
pairs.
It
can
be
strings
so
yeah.
It
could
be
interesting
like
to
hear
from
actual
customers
like
how
do
they
intend
to
use
scopes
and
how
do
they
intend
to?
I
mean
how
do
they
expect
the
scopes
to
be
sent?
A
Do
they
expect
all
the
scopes
to
be
treated
equally
or
do
we
prefer
the
scope
nearer
to
the
actual
dog
statement,
or
vice
versa,
or
do
we
prefer
the
one
from
all
the
way
at
the
beginning,
yeah?
So
azure
is
like
adding
support
you
know,
or
on
the
same
time,
so
we
are
still
learning
like
what's
the
best
way.
So
if
you
hear
from
any
customers,
please
place
an
issue
or
I
think
I'll
create
an
issue
for
getting
some
feedback
on
this
thing.
A
But
if
you
hear
someone
like,
please
encourage
them
to
click
comment.
B
B
Yeah,
the.
A
B
Separately,
you
know
on
that
topic
of
knowing
whether
it's
dot-net
or
not.
I've
had
it
on
my
to-do
list
to
propose
a
spec
pr
to
include
the
user
agent
when
sending
data
from
otlp
and
maybe
any
exporter.
But
you
know
otlp
is
the
one
I'm
most
interested
in
just
some
user
agent
metadata.
Knowing
that
it
was
like.
You
know,
the.net
sdk
versus
java.
A
B
B
We
I
I
definitely
have
some
people
at
new
relic
that
are
very
interested
in
those
numbers
and
are
cobbling
together
some
kind
of
rough
hand,
wavy
estimates,
but
it's
oftentimes
difficult
to
know
whether
it's
people
that
are
using
you
know
an
hotel
sdk.
We
also
have
a
pretty
large
customer
just
sending
us
like
just
use.
You
know
to
open
telemetry,
protos,
raw
and
okay,
not
using
an
open,
telemetry,
sdk
yeah.
I.
A
Was
looking
at
the
collector
repo,
like
many
of
the
collector
cone
drip
components
are
about
logging,
but
they
are
not
from
the
actual
sdk
collected
logs,
send
it
via
otl.
It's
all
like
the
otp
collector
has
many
receivers
quite
like
audiops,
just
one
of
them
it
can
receive
access,
log
and
other
things
like,
and
then
it
forwards
to
convert
city
totally
and
then
exports
to
like
some
places.
A
So
it's
possible
that,
like
a
good
amount
of
use,
cases
are
coming
from
that,
like
you,
people
simply
write
like
to
a
file
or
something
and
then
have
fluency
or
something
pick
it
up.
And
then
you
have
hotel
collector,
pick
it
up
and
then
convert
that
over
to
tlp.
A
So
it's
very
difficult
to
know
whether
it's
actually
the
application
loading
which
is
done
using
some
application
level,
loading
library
and
then
collected
by
open
elementary
and
then
send
videotape.
So
that
distinction
is,
you
know,
I
mean
at
least
I
don't
know
whether
it's
possible,
but
yeah,
like
the
proposal
you
had
about
sending
the
user
agent
would
definitely
help
confirm
whether
it's
coming
at
least
from
the
application,
not
from
the
collector
itself.
B
A
Okay,
yeah
yeah,
good
nice
topics,
I'll
just
put
some
notes
here.
Once
we
end
yeah
see
you
all
next
week.