►
From YouTube: 2020-08-10 C/C++ SIG
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
B
C
C
C
C
C
D
C
E
A
Yeah,
actually,
we
are
not
sure,
but
I
think
lolita
is
in
another
meeting.
Maybe
she
will
attend
this
several
minutes
later,
I'm
not
sure.
What's
the
internal
plan
for
our
code
review.
C
A
No
it
so,
including
this
week.
I
still
have
three
weeks:
yeah,
okay,.
F
Hey
hi,
so
basically
what
I've
been
working
on
this
week
is
implementing
the
trace
state.
F
I
had
a
definition
pr
that
merged
and
I
also
implemented
event
attributes
and
links
in
span
data
as
well,
and
then
one
question
I
had,
and
I
was
just
trying
to
find
the
pr
is
regarding
the
otlp
exporter,
specs,
there's
a
pr
that's
been
going
on.
I
can
add
it
here,
there's
a
pr
that's
been
going
on
for,
like
I
think
about
a
month
now
that
I've
kind
of
been
like
waiting
on.
So
I
can
start
actually
using
these
specs
to
to
update
the
otlp
exporter.
C
C
This
one,
can
you
see
my
screen,
there's
a
property
tag
here.
Oh.
C
Yeah,
so
p3
is
unlikely
to
make
good
progress
because
people
are
fighting
with
p1
and
p2.
So,
okay,
so
the
result.
I
I
think
number
one
thing
is.
I
don't
believe
that
the
otl
px
porter
has
to
get
blocked
on
this
spec,
because
a
lot
of
leaders
they
have
the
otlp
exporter
and
they're
working
today,
like
they
open,
telemetry,
python
and
donna.
Actually
like
we
have
that
exporter
in
both
places.
C
C
C
Maybe
there's
miscommunication,
so
I
let
me
repeat
so,
even
without
the
otro
px
power
spec,
I
think
the
oplp
protocol
is
clear
and,
and
the
majority
of
the
work
should
be
should
not
be
affected
by
by
the
fact
that
content,
we
don't
have
a
clear
spec.
I
think
what
the
spanx
is
trying
to
do
is
to
give
some
configuration
consistency
across
all
languages,
and
even
without
that,
I
think
you
can
have
a
working
otlp
explorer
similar
like
what
donald
and
python
has.
F
Okay,
so
I
can
start
to
work
on
adding
the
configuration,
even
though
it's
not
fully
the
spec
isn't
fully
completed.
G
C
C
H
On
hi
raleigh,
hi
yeah,
so
just
following
up
from
our
previous
sig
meeting,
I
released
a
design
dock
for
the
working
support
for
both
the
processors
and
the
exploders.
The
basic
api
is
ready
for
both
of
them.
The
four
cover
processor
and
exporter.
Now
I've
exposed
three
functions
to
every
processor
and
explorer
in
the
repository
and
yeah.
H
Now
I'm
just
gonna
be
working
through
what
actually
happens
when
you
export,
when
you
fork
in
the
simple
processor
in
the
batch
processor
like
what's
supposed
to
happen,
and
I'm
going
to
be
writing
tests
for
the
same
okay,
so
yeah,
I'm
in
the
process
of
that
right
now,.
C
Okay,
thanks
I'll
unusually
reveal
this
today
sure
thank.
C
I
Okay,
so
last
week
I
made
a
pull
request
that
contained
the
adding
of
you
know
a
span
to
the
runtime
context,
implicitly
upon
the
start
spent
and
johannes
mentioned
that
he
thought
we
should
be
storing
the
span
context
instead
of
the
span.
Do
we
have
any
resolution
about
which
it
should
be.
C
I
So
well,
I
think
it's
all
based
in
when
so
when
we're
using
the
start
span
method,
I
I
we're
we're
attaching
the
a
shared
pointer
to
the
you
know
the
current
span,
so
that's
creating
a
scoping
issue
where
you
have
to
call
end
for
end
on
each
of
the
spans
and
johannes
was,
was
raising
the
thought
that
that
was
problematic
and
instead
of
storing
the
span
itself,
we
should
be
storing
the
span
in
context,
so
we
could
still
keep
the
unique
pointer
to
this
band,
so
it
would
automatically
end
when
going
at
a
scope.
I
I
just
wanted
to
to
see.
If
that's
the
approach
we
should
take,
because
I
felt
like
there
wasn't
clear
resolution
on
the
pr
or
if
we
should
continue
with
the
current
approach.
D
And
storing
the
span
in
the
context
of
the
command,
you
would
have
to
return
a
sharepointer
from
start
span,
as
you
did
in
your
pr
and
then
we
would
lose
the
ability
to
kind
of
automatically
control
like
lifetime
of
this
bank,
just
via
the
scope.
Just
this
raii
idiom
that
we
couldn't
use,
then
so
we
would.
D
The
user
would
have
to
call
and
explicitly
on
every
span,
and
I
just
like
raised
the
possibility
to
just
store
the
span
context,
because
it
would
enable
us
still
to
use
to
use
a
unique
pointer,
but
it
would
not
allow
us
to
implement
the
get
currency
bank
function
on
the
tracer,
which
is,
as
I
see
the
spec
as
it
should.
But
I'm
not
sure
how
strong
the
this
requirement
is
when
it's
just
a
weighing
off
of
either
having
for
me.
D
C
C
I
didn't
see
yeah
so
so
in
python.
What
people
has
done
is
there's
a
functionality
where
you
can
decide
whether
you
just
start
a
span
without
putting
that
and
you
can't
spend
or
you
can
start
as
a
current
span.
C
C
C
D
I
thought
about
that.
I
mean
when
we
have
to
store
the
span
into
the
context
I
mean
we
could
provide
something
similar
in
c
plus,
plus
that
we
just
return.
Oh,
that
we
have
something
similar
like
to
the
block
guard.
Yes,
that
library
just
has
to
lock
kind
of
scope
to
the
current
scope,
which
is
I
mean
again,
something
that
the
user
has
to
write
in,
but
I
think
it's
kind
of
as
far
as
I
could
think
the
most
elegant
solution
for.
C
C
D
Have
some
kind
of
a
card
to
just
that
the
card
goes
out
of
scope
and
then
calls
and
automatically,
but
I
think
that
is
not
in
the
scope
of
sam's
pr,
so
with
that
here,
then
that
we
have
to
store
this
brand
in
the
context,
then
sam
can
just
go
ahead
with
the
solution
that
he
had
already
anyway.
I
And
then
the
other
thing
I
was
working
on
is
starting
with
the
grpc
propagation.
However,
I'm
sort
of
blocked
on
that
until
the
span
context
is
finished,
and
I
know
that's
in
tianlin's
pr
which
is
waiting
on
the
trace
state,
so
we're
we're
both
sort
of
blocked
on
the
trace
state.
There.
C
I
Oh
well,
I
myself
am
not
blocked
on
trace
date,
but
I
am
blocked
on
the
span
context.
Implementation
which
exists
within
tianlin's
pr
tandem
would
have
to
explain
why
he's
blocked
on
tri-state.
C
J
So
for
me
for
the
s4d
http
trace
context,
I
think
everything
is
done,
except
for
two
two
parts.
So
the
first
one
is
the
question
that
sam
has
raised
before
and
like
what
value.
Does
the
context
accept
is
very
important
to
me
as
well,
because
right
now,
I'm
expecting
the
context
to
take
in
to
take
in
a
span
as
a
value
instead
of
a
span
context
right
now.
J
The
only
way
in
http
trace
context
to
get
access
to
a
spam
contact
is
to
extract
the
span
context
from
the
span
which
is
stored
inside
of
the
span,
and
if
that
is
going
to
change,
I
think
I
will
need
to
adjust
my
code
as
well,
and
the
second
part
is
about
the
trace
date
and
the
reason
that
I'm
kind
of
blocked
on
trace8
is
because
I
mean
because
I'm
trying
to
implement
a
fully
functional
trace
context
and
in
the
testing
there
are
a
couple
of
tests
that
are
requiring
the
tree
state
propagation,
also
working
fine
as
well,
so.
J
So
that's
why
I
kind
of
need
to
need
a
tray
state,
otherwise,
I'm
I
could
only
use
a
standard
map
to
replace
that
and
other
than
the
hdb
trace
context.
I'm
also
working
on
the
conformance
testing
and
right
now.
I've
used
lip
curl
to
to
make
a
http
client,
and
I
had
managed
to
look
at
max
code
on
the
server
http
server
and
I'm
trying
to
modify
upon
his
server
and
try
to
see
if
it
can
fit.
The
purpose
of
the
testing
for
conformance
tests.
B
E
C
J
Yeah
and
actually
I'm
not
that
too
worried
about
the
trade
state
problem,
because
I
might
need
to
take
some
time
to
to
finish
the
like
to
finish
the
conformance
testing
setup
and
by
the
time
I
finish
it.
It
might
be
the
case
that
the
trace
data
is
already
resolved,
and
then
I
don't
need
to
really
accommodate
it.
C
I
Yeah,
if
you
could,
you
know,
get
a
pr
merge
as
soon
as
possible.
I
would
appreciate
it
yeah
sure,
but
then.
J
Okay,
I
see
and
then
there's
one
final
problem
I
have
before
I
could
emerge,
which
is
which
is
that
when
trying
to
pass
the
past
the
test,
I
it
seems
that
I
will
fail
the
os
10
test
and
I
think
a
lot
of
people
have
that
same
problem
as
well.
I
don't
think
that's
something
really
to
the
code,
like
my
coding
specifically,
but
something
some
problem
existed
on
on
the
public
repo.
C
Yeah
there
are
many
threads,
so
I'm
I'm
like
loading
track
of
the
the
thrust.
I
think
there's
some
transient
ratio
where,
if
you
rerun,
you
may
pass
that
there's
like
random
issue
introduced
by,
for
example,
like
external
dependency,
like
some
external
test
package,
got
updated
or
the
tool
got
updated,
and
I
I
think
those
got
fixed.
K
Yeah,
so
I
have
the
the
mac
exporter,
which
has
been
renamed
in
memory
span.
Exporter.
That's
finished,
currently
we're
using
kind
of
a
workaround
we're
using
relative
imports
because
it's
required
the
exporters
are
required
in
the
sdk
for
tests
and
your
honda
suggested
using
relative
imports
as
a
workaround,
but
that's
not
working
for
bazel
tests,
so
there's
kind
of
two
sets
of
imports
that
are
required
and
it
will
only
work
with
one
or
the
other.
K
I
think
we're
going
to
need
to
modify
the
global
cmec
file,
but
I'm
not
sure
exactly
what
needs
to
be
done
there.
So
that's
blocked
on
figuring
out
what
needs
to
happen
with
that.
I
was
waiting
to
hear
back
on
my
pr
from
johannes
and
then
I'm
trying
to
get
oliver's
final
pr
for
documentation,
merged
he's,
got
that
ready
and
just
trying
to
get
that
merged
before
friday,
because
it
was
my
last
week
and
in
the
meantime,
I'm
just
working
on
some
optimizations
for
the
samplers.
K
Thank
you
for
helping
over
the
course
of
my
internship.
D
Yes,
I
will,
I
didn't,
find
time,
unfortunately,
to
review
the
pyramid
looking
at
that
last
week,
but
I
will
look
into
that
during
the
next
days.
Okay,
awesome,
thank
you
for
this
one.
I
will
definitely
find
time
the
next
days,
because
I
think
that's
not
the
big
of
a
thing
and
I
will
just
try
to
get
it
to
work
for
me
locally
and
then
comment
on
the
vr
with.
C
Okay
is
eight
pages.
E
So
keisha
has
physical
therapy
right
now,
but
this
is
his
last
week
for
his
internship
and
he
has
two
prs
that
he's
hoping
to
merge
before
his
internship
ends,
just
one
for
the
wrapper
for
users
to
add
seed
pages
to
their
apps
and
then
yeah.
The
z
pages
example
that
kind
of
demonstrates
what
it
does
but
yeah.
E
Besides
that,
that's
all
that
keisha
really
had
and
on
my
end,
all
my
things
got
merged,
so
I've
just
been
mostly
working
on
documentation
and
I
have
like
three
more
weeks,
so
I
could
probably
pick
up
something
like
maybe
a
little
smaller
or
just
maybe
do
some
cleanup
for
z
pages,
because
there's
some
stuff
in
there
that
we
could
improve.
B
Yeah,
like
what
you
guys
had
shown
last
week,
is
good,
so
kind
of
functional
already
right.
Just
we
need
to
share
the
knowledge.
E
Okay,
yeah,
so
I
I
think
we
have
like
our
final
internship
presentation
on
thursday,
but
that's
only
like
within
google,
but
I
think
so
by
like
sharing.
Do
you
want
like
to
see
to
show
like
what
it
looks
like
like.
B
Yeah,
it's
like
I
don't
know,
is
there
anything
top
secret
that
you
guys
are
gonna
be
presenting
in
journal?
Can
you
maybe
record
it
and
share
with
us.
E
E
B
We
get
meetings
recorded
and
published
on
youtube
right,
like
my
understanding,
is
that
it's
happening
yeah.
C
Yeah
you're
right
so
either.
I
I
think
you
can
either
summarize
this
by
putting
a
document
or
you
can
either
tell
people
like
here
goes.
The
call
just
take
a
look,
or
we
kind
of
use
the
next,
like
community
meeting,
to
have
a
quick
demo
from
you
like
it's
up
to.
E
Okay,
yeah,
I
don't
know
I
haven't
built,
it
will
probably
take
a
while
to
build
to
like
just
do,
show
everybody
right
now,
but
there's
I
have
like
the
design
dock.
I
could
just
show
right
now
that
has
a
picture
of
what
it
looks
like
and.