►
From YouTube: 2020-08-17 C/C++ SIG
Description
Open Telemetry Meeting 1's Personal Meeting Room
A
B
Right
waiting
for
people
to
join
I'll
do
take
the
time
and
put
her
names
here
thanks
next,
I
just
want
to
quickly
check
with
you
like.
Do
you
have
some
like
demo?
I
think
you
mentioned
last
week.
You
want
to
do
a
demo.
Oh.
C
D
C
D
B
D
Okay,
can
you
see
my
screen
yeah.
D
Cool,
so
I
already
built
everything
and
I'm
just
going
to
run
the
executable
from
the
z
pages
example,
which
we
have
a
pr
up
for
and
the
wrapper.
D
So
it's
creating
the
spans
for
the
example,
and
this
is
what
it
looks
like.
So
we
create
an
error,
span,
a
running
span
and
then
like
a
span
that
takes
a
really
long
time
to
render,
and
we
also
have
other
spans
that
would
be
being
created
every
second
or
so
we'll
be
here.
It's
kind
of
emulating,
like
kind
of
like
a
chat
app.
D
So
this
you
know
just
be
like,
like
a
deadlock
of
also
sort,
our
instrumentation
error
and
here's
just
like
a
regular
error,
and
you
can
also
see
like
the
status
and
hold
hover
over
and
see
what
the
sas
means,
and
we
also
have
the
attributes
implemented.
And
you
can
see
the
rows
for
each
of
the
tables
and.
A
D
B
Yeah
so
just
curious,
so
so
the
page
like
so
you
can
refresh
the
page
on
the
on
the
top.
Like
there's
a
button
like
fetch
new
data.
D
If
you
look
at
the
network
instead
of
like
doing
the
server
side
like
fetching
the
page,
you
can
just
update
it
using
a
call
like
just
get
the
aggregations
at
the
table,
so
the
html,
css
and
javascript
doesn't
get
fetched
again
and
it
just
like
updates
the
dom.
D
D
B
B
So
I
also
probably
like
one
minor
suggestion
for
the
timestamp.
I
would
suggest
to
use
iso
standard
timestamp,
which
is
like
formatted
to
utc,
and
probably
you
can
consider
some
like
additional
ux
feature
where
people
can
switch
either
from
local
time
to
a
you
next
time
or
the
ep
system
at
a
time.
This
will
be
helpful
for,
like
folks
like
during
conversation
like
there
might
be
folks
from
different
like
time
zones
they
communicate
with
each
other,
so
just
full
convenience.
D
Okay,
yeah,
that
makes
sense.
I
think,
what
we're
doing
in
the
javascript
right
now
is
that
we're
just,
I
think
it's
a
unix
space
number.
It's
like
a
really
long!
It's
like
a
long
long.
It's
actually
even
had
to
do
a
custom
function
to
convert
the
number
to
a
stamp,
so
I
could
definitely
adjust
that
because
there
we
have
like
a
lot
of
lost
precision
in
there,
so
we
had
to
like
just
append
this
to
end
of
this
time
stamp
and
yeah.
This
is
definitely
something
we
could
change.
B
Time
anyone
online
from
aws
yeah,
I
am.
B
E
E
Yeah,
but
that's
that
was
in
terms
of
the
like
the
implementation
yeah.
So
actually
we
are
going
to
refactor
the
prs
we
have
filed
to
to
make
sure
that
each
pr
is
buildable.
So
for
each
pr
we
will
put
the
header
definition,
the
implementation
and
the
unit
test
file
for
for
one
single
component.
So,
ideally,
we
will
have
three
pr's
for
for
prometheus
exporter,
promises
collector
and
the
utility
class.
E
F
Yeah,
so
one
minor,
I
did
this
request
right
just
to
explain.
I
I
looked
over
the
prs
a
bit
like
end
of
last
week
and
it
would
be
kind
of
edited
request
just
to
make
them
buildable
and
make
the
best
running
in
our
ci,
because
now
they
were
just
like
separate
parts
and
the
tests
were
not
running
and
the
prs
like
were
depending
on
each
other,
to
actually
get
something
built.
So
I
think
it
would
just
make
reviewing
much
easier
if
some
something
could
be
built
and
just
run
in
our
ci.
B
Yeah,
I
agree
and
one
suggestion
from
my
side
so
like
this
is
like
when
I
was
working
on
the
opentimetrade.net
in
the
past
few
days,
and
you
can
see
one
thing:
is
you
put
a
clear
like
step,
one
two,
three
four,
so
people
know
like:
what's
the
ordering.
A
B
They're
trying
to
do
giving
the
like
the
overarching
description
of
what
you
try
to
do
and
how
do
you
split
out
the
work,
okay
and
then
make
it
as
small
as
possible,
but
you
need
to
make
that
self-contained,
so
people
can
instead
of
having
five
different,
pr's
and
running
the
circular
like
loop.
That
makes
it
harder
for
the
others
to
reveal
it's
just
minor
suggestion.
Yeah.
E
Sure-
and
another
problem
is
like
for
those
seven
pr's-
we
have
already
filed
what
we
have
done.
Was
we
forked
this
repo
and
created
new
branches
for
each
feature
and
file
pr
from
the
new
branches
in
order
to
make
all
the
prs
like
isolated,
and
it
does
not
contain
any
files
in
other
prs.
So
that's
why
some
of
our
pr's
cannot
build.
So
basically,
here
in
our
local
repo,
we
have
seven
different
branches
for
seven
prs.
First
things
we
are
going
to
do
the
refactor.
E
So
do
you
have
some
recommendations
for
us
in
this
case.
B
Yeah
so
yeah
I
I
do
so
like
for
this
donut
like
refactor,
so
basically
I
was
doing
changes
and
after
that
I
switched
like.
Basically,
I
have
different
branches.
I
have
step
one
step
two
or
step
three
branches
and
I
do
all
the
merge.
So
I
send
the
first
pr
it's
based
on
the
master
and
then
I
send
the
second
pr
which
is
based
on
the
step.
One
thing
and
the
third
pr
is
based
on
step
two.
So
it
gives
people
a
clear
view
of
like.
B
B
B
I
I
think
I
probably
will
will
be
able
to
start
to
reveal
the
the
first
like
two
prs
later
today.
I'm
sure.
B
Yeah,
so
if
you
want
to
like
clean
up
the
pr
make
sure
you
start
from
the
first
pr's
like
the
first
two
prs
and
if
the
other
pr's
you
figure,
you
want
to
spend
more
time
or
you're
going
to
make
dramatic
changes.
It
doesn't
make
sense
for
johannes
and
I
to
start
review
and
later
a
lot
of
changes
happen.
So
all
the
comments
got
reset.
So
probably
my
suggestion
would
be
mark
those
pr's
that
you
think
will
be
revealed
later
as
like
not
ready
for
reveal
either
change
them
to
draft
or
working
progress.
B
Okay,
anything
you
want
to
mention
in
addition
to
this,
for
communities.
G
Okay
hi,
so
for
me
last
week
I
was
working
on
the
trace
state
implementation
and
I
merged
it
on
friday,
and
then
this
morning,
like
I
just
fixed
like
a
small
bug,
and
I
have
like
a
few
more
finishing
touches
to
put
on
the
trace
date
and
then
other
than
that,
what
I'll
be
working
on
is
doing.
The
exporter
configuration
for
the
otlp
exporter,
taking
a
look
at
the.net
exporters
as
an
example
and
then
other
than
that.
G
F
They're,
like
I
would
also
want
to
make
a
suggestion
here
now
that
we
have
the
contact
stuff
in
place.
I
mean
nadia
could
work
on
like
actually
correctly
setting
the
parent,
ids
and
trace
ids
for
all
those
bands,
because
that
is
not
up
yet,
so
we
still
don't
set
like
trace
and
spanner
these
and
parent
ids
for
spain,
and
I
think
now
that
the
context
stuff
is
in
place.
I
think
that's
doable
in
two
weeks.
F
If,
if
that
sounds.
G
F
As
far
as
I
can
see,
everything
is
in
place
for
that
we're
just
missing
at
basically
setting
the
I
mean
you
would
need
to
like
create
a
trace
id
if
none
is
created
yet
and
then
on
each
like
creation
of
a
new
span,
basically
setting
a
new
span
id
retrieving.
The
current
span
context,
I
think,
that's
already
in
the
code,
but
what's
missing
it,
it's
setting
the
parent
id
on
the
span
and.
D
F
D
F
F
A
F
H
Right
all
right,
hi,
riley,
hi,
all
right,
so
I'm
doing
well.
Thank
you
yeah!
So
for
my
updates,
I
just
pushed
out
the
vr
for
forking
support
last
friday
and
I
got
some
comments
from
max
and
it
seems
like
we're
looking
for
a
more
concrete
solution
to
the
entire
thing.
Wait
and
we'll
have
like
a
small
platform,
even
handler
class,
which
will
implement
all
platform
specific
events.
So
I'm
waiting
for
a
draft
pr
from
him
still
max.
Are
you
there
on
the
call.
C
Yes,
I
would
suggest
you
to
implement
the
class
like
fork
aware
as
a
temporary
solution
to
try
things
out,
and
then
we
will
refactor
this
into
a
better
implementation
that
provides
platform
clouds
like
for
platform.
Events
such
as
unforked
unsuspended
or
resumed
terminating.
Something
like
this
in
a
generic
fashion,
and
the
logical
reasoning
is
that
fork
itself
has
features
mostly
linux,
with
some
quirks
significant
quirks.
H
Okay,
all
right
so
should
I
do
the
same
thing
for
the
exploders
too,
because
I
think
for
the
processors
the
pr
is
up
so
should
I
be
doing
the
same
thing
for
the.
C
Exploders,
so
I'm
I'm
actually
working
on
giving
you
some
thoughts
on
event
and
like
registration
for
a
triggering
an
event
as
well
as
handling
that
I'd
like
you
to
proceed
with
the
implementation.
Let's
say
you
have
a
fork
aware
class,
whatever
make
sure
that
all
tests
pass
with
that,
don't
wait
for
my
proposal,
I
think
yeah
all
right,
then
I
actually
will
refactor.
I
think
we
should
make
sure
that
it
works
first,
then
we
refactor
this
to
a
bridge
like
merge
into
one
common
implementation.
H
Okay,
I
have
a
design
dock
for
this
I'll,
just
share
it
with
you,
so
I
think
everything
can
go
in
one
place:
the
proposal
and
everything
so,
okay
and
yeah.
This.
B
The
link
in
here,
like
so
people
interested
in
the
community
can
also
take
a
look.
H
Sure
I'll
do
that
right
after
this
yeah
and
yeah.
This
also
is
my
last
sig
meeting,
so
I
think
my
internship
ends
on
this
friday,
so
just
wanted
to
thank
each
and
every
one
of
you.
I
think
it's
been
a
really
great
experience.
Learning
from
you
all,
and
hopefully
I
can
keep
contributing
to
this
even
after
my
internship,
but
yeah
just
wanted
to
take
this
time
to
thank
all
of
you.
B
A
Hi
so
want
to
say
this
is
my
last
week
in
this
internship
as
well,
so
I
hope
yeah.
So
basically,
this
has
been
a
wonderful
leadership
and
I
want
to
thank
everyone
who
helped
me
and
pointing
out
what
I
have
to
do
upon
myself,
but
other
than
that.
A
I
would
also
like
to
leave
some
of
my
contributions
being
seen
on
the
main
public
repo.
So
and
my
pr
is,
I
think
it
is
ready
to
merge
right
now.
The
pr
is
without
it's
on
the
http
trace
context
and
right
now
it's
the
version
that
is
without
the
trace
date,
and
actually
I
do
have
the
version
that
that
is
indeed
with
the
tristate.
But
I
would
like
to
go
step
by
step
and
it
would
be
great
if
you
could
merge.
A
You
could
go
over
and
merge
the
pr
on
this.
It's
number
143..
A
It's
only
failing
the
it's
only
fitting
the
co
code
coverage
test,
which
is
because
I
removed
everything
related
to
the
tree
state.
I
think
it's
fine
and
as
long
as
it
as
soon
as
I
didn't
merge,
I
will
bring
the
branch
with.
A
I
would
update
the
pr
with
excuse
me
with
the
version
of
the
of
the
branch
with
the
trace
date,
and
it
will
be
ready
to
merge.
A
I
think,
as
long
as
I
merge
that
branch
back
into
the
pr
yeah
and
as
for
the
conformance
test,
I
think
it's
going
steady,
although
it's
not
yet
being
like
it's
not
yet
passing
on
the
test,
but
at
least
it's
I
mean
the
test.
Validation
service
and
the
test
service
are
coordinating
with
one
another,
which
is
a
good
start,
and
I
think
the
things
are.
A
The
things
left
are
just
to
utilize
to
utilize
the
code
I've
written
to
and
put
them
into
the
callback
functions
of
this
service,
which
should
be
done
quite
like
with
not
without
a
lot
of
difficulties
and
all,
I
have
to
say.
B
I
Wonderful
good
to
be
back
all
right,
so,
as
I
was
saying
previously,
we
merged
the
span
context
and
placed
it
out
of
the
context
pr
and
I'm
currently
looking
at
the
grpc
propagation,
I
wrote
out
something
that
works
ish.
You
know
in
a
simplistic
form
based
off
of
the
open
census,
grpc
propagation,
but
I
had
a
quick
conversation
with
riley
this
morning
and
I'm
switching
over
to
look
at
how
python
and
go
are
doing
it.
I
You
know
to
see
if
I
can't
follow
this
back
a
little
better,
but
so
that's
in
progress-
and
it's
also
my
last
week,
so
I'm
not
a
hundred
french
chad
sure
I'll
be
able
to
get
it
done,
but
I'll
do
my
best,
but
otherwise
it
has
been
a
pleasure
and
yeah.
I
hope
I
was
able
to
add
something
of
value.
B
Yeah,
thank
you.
The
context
part
is
is
one
critical
piece
for
this
entire
project
and
thanks
for
your
contribution,
definitely
see
a
lot
of
value
from
that.
B
Much
yes,
great
thanks
a
lot
simon
yeah
and
for
the
grp
says
so
one
one
interesting
thing
is
that
today,
like
in
the
open,
telemetry
spike,
there's
only
a
clear
specification
about
how
the
text
format
should
be
so
there's
itex
format,
interface
and
that
covers
the
http
part.
But
people
see
the
value
of
having
a
binary
format
and
I'm
not
sure
for
grpc
people
prefer
the
binary
format
or
not.
B
At
least
the
binary
propagation
is
a
must-have
for
some
of
the
like
compact
product,
like
I'm
qtt
or
amqp,
so,
probably
like
check
the
other
repo
and
see
what
people
are
doing
there
and
get
some
idea
from
the
conversations
you
can
search
for
like
a
binary
format,
propagation,
probably
in
the
stack
and
other
like
recall,
or
even
in
the
in
the
spike
six.
They
have
a
similar
like
meeting
notes.
You
can
see
like
this
has
been
covered
to
give
some
idea.
B
At
least
I
want
to
make
sure
like
either
use
the
text,
format
or
binary
format.
You
got
some
ideas
from
where
the
community
is
heading
towards.
Instead
of
doing
some
throwaway
work.
B
Yeah
instead,
thank
you.
J
Yeah
me
too
thanks.
I
wanted
to
call
out
one
thing
about
the
conformance
test
work
that
gianlun
was
discussing.
I
think
that
it
relates
to
the
work
that
johannes
flagged
earlier,
that
nadia
could
work
on,
because
without
the
generation
of
trace,
ids
and
correctly
generating
spans
within
traces
on
the
fly,
the
performance
tests,
I
expect,
will
not
be
passing.
B
Thanks,
thank
you.
No
assembling.
K
B
Yeah
yeah
yeah.
So
so
you
take
a
look
at
the
merge
conflict
and
the
failures
here.
K
Okay,
yeah,
I'm
not
sure
what
those
are
caused
by
but
I'll
take
a
look
at
this.
When
I
have
a
chance
and
then
johannes,
you
said
that
the
other
pr-
that's
failing,
you
said
that
might
be
related
to
the
context.
Have
you
had
a
chance
to
look
into
that
at
all
or.
F
K
If
these,
if
I'm
gonna,
have
to
re-push
to
these,
I
might
have
to
close
the
bra
or
close
the
pr
off
and
create
a
new
one.
Just
because
my
google
email
got
deactivated
so
I'll
have
to
resign
the
cla
by
myself
with
my
personal
email,
so
I'm
not
sure
how
much
time
I'll
have
but
I'll
try
to
get
those
fixed
sometime
this
week.
K
I
have
both
of
those
it
was
linked
to
it,
but
when
oliver,
when
he
left,
he
had
some
issues
with
cla
when
his
google
email
got
deactivated,
it
wouldn't
let
him
create
or
update
his
existing
pr.
So.
B
Thanks
now,
the
pages.
D
Oh
hi,
everybody
again,
okay,
so
I
did
the
demo
so
yeah,
just
gonna
follow
up
on
a
couple
pr's
that
keisha
put
up
before
his
internship
ended.
So
for
the
rapper
and
example,
I
think
it's
a
little
bit
blocked
on
making
a
c
make
build
for
the
json
dependency
it
uses
or
that's
like
something
we
need
to
follow
up
in
a
later
pr.
It's
like
a
little
weird,
because
I
think
the
example
is
approved
right
now,
but
I'm
not
sure
about
what
we're
supposed
to
do
about
the
cmake
build.
D
C
Just
make
sure
that
that
works
for
the
bazel
build,
and
then
I
can
help
with
the
cement.
D
Okay,
yeah.
I
think
it
should
work
for
the
basil,
because
that's
what
we've
been
using
for
building
everything
and
getting
like
our
example
running
yeah,
that's
a
good
next
step!
Thank
you
for
that
unrelated
things.
There's
like
some
different
topics.
I
was
wondering
if
I
need
to
make
github
issues
for
I
know
you
mentioned
that
there
might
be
some
changes
to
the
hp
server
that
we
should
make
or
if
that
doesn't
need
like
an
issue
for
it.
D
So
like
the
thread
safety
just
because
you,
this
is
just
like
an
initial
iteration
of
the
http
server
and
you
wanted
to
add
some
extra
things
to
it
and
then
I
think
nadia
is
already
taking
this
over,
but
yeah
the
ids
for
the
spans
and
traces
they
aren't
set
yet
and
then
tracy
we
don't
have.
We
only
showed
attributes,
so
we
didn't
show
any
events
for
that.
Yet
so
maybe
that's
another
github
issue.
We
could.
D
B
D
Okay,
good
to
know
so
I'm
not
sure
about
the
ids
because
it
looks
like
nadia
is
going
to
be
working
on
that.
Should
I
not
make
an
issue
for
that
then.
B
F
D
All
right
nadia:
would
you
want
a
github
issue
created
for
that,
or
do
you
think
that's
like
something
good
to
go
without.
G
It
it
doesn't
hurt
to
have
an
issue.
I
would
appreciate
yeah
if
you
want
to
make
one
that
sounds
good
to
me.
D
Okay,
yeah
just
some
different
things
that
I
noticed
we
might
wanna
track
when
looking
over
all
the
z
pages
code
now
so
yeah.
I
have
two
more
weeks
also
so
my
host
and
co-host-
and
I
were
thinking
that
we
wanted
to
write
performance
tests
for
those
pages,
back-end
components
and
I'm
also
working
some
documentation
things
outside
of
the
c
plus
plus
repo,
and
I
could
once
I
finish
those
I
could
probably
ask
for
more
things.
If
I'm
up
for
it.
B
E
E
But
my
problem
is:
should
I
create
new
branches
and
do
the
refactor
and
make
new
prs,
or
should
I
modify
the
existing
branches?
So
if
I
continue
work
on
the
existing
branches,
some
of
the
branch
name
does
not
make
sense.
So,
for
example,
I
have
a
branch
for
headers,
but
now
this
branch
may
also
contain
implementations
in
the
tests.
E
B
Yeah,
I
I
think
there's
no
perfect
way,
but
if
you
see
like
existing
pr
already
has
a
lot
of
comments,
and
there
are
a
lot
of
follow-up
discussions,
then
probably
it
doesn't
work
like
makes
sense
to
create
a
new
pr
and
like
like,
have
two
different
conversations.
B
B
Thanks:
okay,
I
anything
else
before
we
jump
to
the
vrs.
F
There's
only
one
short
thing,
not
really
really
just
in
between.
I
thought:
jodie
varney
is
on
the
car,
just
another,
introduce
her
she's
like
a
product
manager
from
relic,
oh
yeah,
she's,
dialing
in
and
just
seeing
where
the
project
stands
and
how
things
are
moving.
Just
that
you
know
who
the
new
face
is
on
the
side.
L
Sure
sure
so,
I'm
responsible
for
the
phpc
and
go
languages
here
at
new
relic,
but
I'd
be
most
interested
in
whether
or
not
I
can
contribute
here
to
helping
with
like,
like
overall
project
management
activities
so
like.
Where
are
we
at
and
like
giving
people
very
clear
visibility
as
to
you
know
how
close?
L
Where
are
we
at
in
relation
to
releasing
something?
What
would
we
release?
You
know
what,
sir,
how?
How
closely
do
we
match
the
specs
right
now,
you
know
what
would
be
what
are
the
upcoming?
What
are
the
issues
that
are
in
our
backlog
that
we
really
recommend
people
work
on
first,
like
giving
so
people
have
very
clear
visibility
when
they
come
here
like
where
can
they
make
the
most
value-added
contribution
just
to
start
with,
so
anything
I
can
help
with
that?
It
would
just
let
let
me
know,
but
I'll
just
be
here.
B
Great
welcome
so
my
suggestion
would
be
try
to
join
the
the
spec
sig
to
get
some
idea
where
the
entire
project
is
heading
towards,
because
that
one
gives
overarching
direction
of
the
like
cross-ice,
dk
collaboration
and
for
specific,
specifically
speaking
about
this
ick
or
if
you
want
to
like
work
on,
go
sdk.
The
process
will
be
the
same,
so
I
initially
can
help
to
see.
B
So
I
think
any
like
sort
of
help
would
be
like
appreciated
it,
and
one
specific
role
that
the
open
time
trade
community
has
been
discussing
is
they
won't
have
a
triager
role
so
that
that,
like
with
the
being
a
treasurer,
that
means
you
need
to
have
idea
of
like
how
the
project
is
moving
and
what's
the
dependency.
B
L
B
Yeah,
so
it's
it's
not
like
per
like
sick
or,
like
personal
preferences,
the
general
practice
across
this
entire
project.
So
I
I
guess
we'll
just
follow
like
what
the
entire
product
is
using
fair
enough.
B
You,
okay,
I
think
we've
covered
the
premises
exporter
about
how
we
can
make
better
progress
on
the
prs,
so
I
probably
will
go
from
the
bottom
up.
So
for
this
oxygen
pr,
it
has
been
there
for
a
while.
So
I
wonder
like
if
there's
any
plan
to
make
progress-
or
it's
still
stuck
there.
E
Yeah,
I'm
still
stuck
there.
So
actually
I
don't
have
enough
time
to
work
on
this,
but
I'm
still
stuck
at
where
I
was
like.
So
the
the
open
telemetry
starts.
Namespace
is
a
macro,
so
I'm
still
looking
for
a
good
way
to
expand
the
macro
in
oxygen.
E
A
E
I
I
think
maybe
I
can
work
on
this
after
my
internship
ends,
but
for
for
the
coming
week
I
will
still
focus
on
the
prs
and
like
the
off-boarding
stuff,
for
my
internship
yeah.
C
Actually,
for
the
macro,
I
remembered
something
I'm
going
to
copy
a
suggestion
into
chat
window.
You
can
configure
the
oxygen
to
predefine
certain
macros
for
certain
values
for
the
macros,
and
that
might
might
help
you
with
the
like.
You
just
have
to
specify
what
the
value
of
that
macro
is,
and
then
the
oxygen
should
be
fine,
after
that,
like
I'll,
could
be
suggestions
to
the
children.
C
F
B
B
C
On
this-
and
I
need
your
opinion
on
these
guys
like
see
what
we
do-
we
kind
of
build
the
docker
image,
then
we
run
docker
image,
but
github
actions
actually
provide
some
pre-built
vms,
which
already
have
all
the
tools
installed.
So
we
can
actually
run
things
faster
if
you
don't
need
to
build
the
docker
image
and
the
vms.
C
Coincidentally,
the
two
vms
that
are
already
provided
by
github
actions
are
ubuntu
18
and
ubuntu
20
and
ubuntu
20
already,
and
can
they
install
the
right
compiler
gcc
with
c
plus
possible?
One
is
support.
I
I
shown
how
I
do
it
in
my
fork.
Basically,
you
can
have
one
github
action
config
with
so-called
matrix,
build
configuration
run
this
on
ubuntu
18
run
this
on
ubuntu
20.
C
and
then
one
of
whatever
a
standard
install
script
that
you
run
to
install
the
dependencies
such
as
gcc
will
actually
make
sure
that
your
ubuntu
20
build
is
already
kinda,
see,
plus
plus
and
then
in
cmake
there's
a
script
where
there's
a
part
where
you
say
whatever
tool
set
standard
and
if
you
set
it
to
20,
but
the
compiler
is
17.
It's
going
to
downgrade
to
17,
but
if
compiler
is
20,
it's
going
to
build
as
20.,
so
it
seems
like
we
can
actually
do
it
in
a
bit
different
way.
F
A
C
I
get
it
johannes,
my
feedback
is
that
pretty
much
like
I
do
have
similar
setup
and
there
should
be
one
script
which
you
can
run
either
on
18
or
20,
and
it
doesn't
really
matter
because
when
you
run
it
on
20,
it's
going
to
install
the
compiler
that
can
do
20
and
the
cma
flag
is
going
to
pick
up
20..
So
that's
why
I'm
kind
of
thinking
like?
Maybe
we
can
do
it
in
a
way
that
it
won't
really
like.
F
C
It's
like,
let's
say
if
we
can
take
some
upstream
ubuntu
docker
base
image
with
either
whatever
16
18
or
20.
16.
C
I
think
it
might
be
four
I'll
check
on
this,
but
then,
if
we
have
a
genetic
ci,
whatever
local
built,
to
set
up
script,
that
will
just
install
the
standard
tools
that
come
with
that
distro
seems
like
we
can
actually
cover
a
few
of
these
with
one
tool,
setup
script
and
then
run
the
flow
like
anyways.
Let
me
provide
some
thoughts
on
this
because
we
do
have
similar
setup
elsewhere,
and
I
also
showed
how
to
do
it
in
a
bit
different
way.
C
In
my
fork,
I
can
probably
split
it
into
kind
of
just
the
tools.
Pr,
because
I
sent
like
a
huge
pr
with
a
few
other
things
in
it.
Maybe
I
can
send
in
some
tools
the
pr
and
an
example
how
to
run
that
with
github
actions
like
on
168
into
any
and
sure
yeah.
Let
me
follow
up
on
this.
B
Thanks
and
for
the
sampler
documentation,
I
think
we'll
talk
about
this
so
once
the
ci
is
clean,
it's
ready
to
merge.
I
think
it's
probably
like
not
related
to
this
change,
but
please
take
a
look
and
for
the
prometheus
exporter,
the
the
series.
So
we
talked
about
that
before
johannes
for
this
one
I
think
there's
some
like
failures,
not
sure
if
you
got
a
chance
to
fix
that
or
last
time
I
I
checked
steel.
F
D
Yeah,
I
think
either
we
could
merge
it
right
now,
as
is
or
like
wait
until
the
json
dependency
for
cmake
is
created
and
I'm
gonna.
I'm
gonna
write
an
issue
for
that.
C
Merge
it
create
an
issue
and
I'll
help
you
guys
with
the
simulink
setup.
I
I
still
want
to
bring
up
the
topic
for
the
third
time.
Maybe
if
we
start
using
git
modules,
then
we
can
kind
of
say
that
this
third
underscore
party
directory
at
the
root
folder
and
then
there
we
had
whatever
json
hpp
like
pc
package
and
a
few
others
again
open,
telemetry
java
is
doing
this
already
and
I'm
using
them
as
an
example.
C
Even
though
we
don't
need
that
for
bazel
build,
it's
probably
going
to
be
a
bit
easier
to
for
cmake.
If
we
have
that
one
git
clone
that
already
recursively
brings
in
all
needed
dependencies,
such
as
json
hpp,
and
we
would
have
less
build
script.
Things
like
installing
json
hp,
because
we
would
already
know
that
it's
kind
of
located
under
third
party
under
some
some
known
path,
anyways,
open
the
github
issue
and
I'll
cover
the
cmak
part
of
the
build
for
that.
D
Okay,
yeah,
I
think
you
shared
some
resources
when
we
did
the
initial
pr
for
the
json
with
just
the
bazel
build.
Is
that
like
what
you're
referring
to
with
resources
or
the
other?
Are
there
other
things
we
should
other
ways?
I
would
get
information
about
that
or
share
that
with
other
people.
They're
curious.
C
So
let's
say
you
guys
cover
basil
for
now
and
whatever
cma
parts
at
the
red
options,
either
we
do
vc
package
install
whatever,
and
this
will
work
and
we
do
it
elsewhere
in
the
setup
scripts
right
now
or
we
start
using
sub
modules
and
then
whatever
build
script,
files
and
cmx
script
files,
we
hard
code,
the
path
to
third
party,
slash
json,
then
log
on
json
hpp,
like
both
ways
would
work.
I
would
probably
go
the
least
cost
route
yet
I'll
just
do
the
vc
packaging
style
for
now,
but
then
basically
I'll
think.
C
Maybe
we
guys
should
switch
to
sub-modules.
It's
like
more
of
an
organizational
decision,
because,
right
now
we
don't
use
sub-modules,
and
I
think
why,
during
one
of
the
early
meetings
we
discussed
that
we
shouldn't
and
again
I'm
showing
that
open
senator
java
does
it's
a
module.
So
maybe
we
should.
D
C
Yeah
for
the
static
files
like,
if
you
guys
have
another
public
repo
location
for
the
static
file,
the
I
think
sergey
should
help.
You
drive
this.
As
the
owner
of
the
pages
accrues
all
languages,
then
we
say
whatever
path
is
recursively
cloning
from
your
other
repo
and
you
post
your
index,
html,
whatever
javascript
files
and
styles
in
there
and
then
in
the
getting
started
and
contributing
we're
going
to
say
that
you
do
get
clone
recursive
with
some
modules
and
then
it's
pretty
much
one
command.
C
That
is
pulling
everything
together
into
one
kind
of
workspace,
three
yeah
as
an
option.
D
Okay,
yeah,
it's
cool
yeah,
definitely
could
put
that
up
as
a
future
request.
B
Okay,
thanks
and
for
the
using
like
struck
as
a
team
processor,
so
I
have
to
send
that
to
myself
I'll,
take
a
review
and
I
I
think
I
finish
his
internship
but
like
he
promised
that
he'll
help
get
this
pr
armored
so
we'll
see,
I
think
it's
a
small
er
and
the
14th
support.
We
talked
about
this,
so
there's
a
flop
from
from
max,
but
for
now
I
I
think
it's
not
blocked
right
max.
You
want
the
jar
to
move
forward
and
then
we
can
refactor
and
extract
them.
Yes,.
C
I
didn't
I
didn't
request
changes
on
this.
I
just
left
my
comments
right
now
and
my
comments
are
in
the
base
class.
Let's
not
call
it
like
fork
aware:
let's
yeah,
let's
find
a
good
way
to
abstract
it
away,
and
then
it's
like
a
platform,
specific
feature
that
is
neatly
abstracted
away
and
we
can
come
up
with
a
few
other
useful
features.
Yeah
sure,
let's
follow
up.
B
C
Guys
at
your
own
pace,
could
you
please
start
looking
because
what
I
did
for
the
no
std
classes
right
now.
I
made
sure
that
whatever
code
as
of
six
days
ago
requires
minimal
to
know
modifications
elsewhere,
like
everybody
is
still
using
no
std
classes
on
api
surface,
but
then
there's
a
compiled
time
option
that
allows
to
remap
everything
to
the
standard
library
and
again,
the
main
reasoning
is
for
header,
only
statically
linked
implementation
of
the
sdk.
C
C
I
don't
want
to
cause
any
split.
We
still
use
no
std
classes
on
api
surface
and
we
still
fully
support
that
and
dynamic
runtime
loadable
build
as
as
the
main
build.
I
just
want
to
add
more
flexibility
and
what
I'm
trying
to
illustrate
that
when
I
had
that
flexibility,
all
the
test
pass,
everything
still
works,
and
it's
like
one
to
one
mapping.
Everything
is
passing
the
same
way
and
operating
the
same
way.
C
C
C
I
realize
it's
a
bit
of
an
overhead
like
sometimes
it's
a
dumb
thing
like
a
integer
used
in
a
loop
right
and
something's
gonna,
say:
hey
size,
t
versus
end
type,
mismatch
and
all
that
or
sometimes
there's
an
assumption
about
truncation
64-bit
to
32-bit,
which
is
safe,
but
still
there's
a
compiler
warning
and
in
a
few
environments
my
customers
would
prefer
zero
earnings.
C
Even
if
these
are
silly
warnings-
and
you
can
right
away-
tell
that
it's
a
safe
thing,
maybe
we
should
try
to
enforce
this
more
strictly.
I
would
do
this.
B
C
I
can
clean
it
at
least
for
visual
studio
at
first,
and
maybe
we
can
take
another
step
at
other
platforms
because
there
might
be
quite
so.
It's
usually
developer.
Does
some
code
right
and
zero
warnings
in
gcc
or
zero
warnings.
Inquire
right
and
the
code
is
great
and
everything's
working
and
then
there's
a
warning
and
visually.
It's
like
man,
but
typically
it's
an
easy
fixer
like
yeah,
your
pr
is
gonna
fail,
but
then
it's
like
hey.
I
should
have
used
size
t
here
instead
of
it
right.
C
B
B
Sure
yeah
I'll
I'll
do
some
fixes
in
this
regard,
and-
and
we
did
find
like
three
or
four
like
real
issues
from
those
warnings,
so
I
I
think
they
are
helpful
and
also
that's
a
that's
a
quality
bars.
Basically,
when
people
come
to
this
reply,
I
think
they
we
would
want
to
make
sure
that
we're
doing
a
reasonable
job
here,
instead
of
like
say
like
we
like
we're
very
like
forgiving
on
those
like
warnings,
so
yeah.
F
C
Yeah
the
reason
why
I
ran
into
this,
like,
as
I
mentioned
to
you
guys,
I
do
some
internal
trials
with
just
api
headers
and
windows.
Customers
like
they
run
into
some
issues,
and
I
think,
ideally,
we
should
fix
this
in
the
main
line.