►
From YouTube: Benchmarking WG Meeting Aug 28 3 EST.
A
A
A
No
okay:
in
terms
of
the
stand
up
since
last
meeting,
we
held
the
session
that
I
was
involved
in
along
with
Cornell.
In
terms
of
you
know,
trying
to
get
make
some
progress
on
setting
up
DCIS
I,
followed
up
with
Miles
on
the
startup
fix.
As
we
know,
we
made
a
security
fix
that
affected
startup
significantly.
It
seems
to
have
come
back
and
master,
but
not
note
version
eight.
A
It
did
turn
out
that
there
was
a
fix
missing
and
we're
still
waiting
for
that
to
hit
a
release
to
see
that
recover
in
the
chicken,
the
benchmarking
charts
and
some
investigation.
You
know
just
this
morning,
I
was
taking
a
look
to
see
why,
if
I
can
figure
out
why
DCE
is
isn't
starting
up
and
I'll
try
and
find
some
more
time
to
do
some
of
that
as
well
now,
I
think
you're.
Next
in
the
list.
B
C
Gareth
yeah
not
too
much
I'm
still
be
good.
If
we
can
get
a
bit
more
feedback
on
me
running
the
micro
benchmarks,
I
did
looking
to
an
issue
that
the
benchmark
and
that
Mattia
kicked
off.
It
seemed
to
be
not
exactly
an
issue
with
the
rapper,
but
the
rapper
looks
out
that
it
can
get
a
decent
set
of
numbers
and
he
didn't
find
that
so
it
it
quit
out.
C
But
it
wasn't
a
particularly
helpful
error
message
to
tell
you
that,
so
it
took
a
bit
of
discovering
where
that
one
out
I
might
have
a
look
at.
Maybe
I
didn't
something
else
to
the
rapper
to
those
people.
No
I
did
listen
through
the
recording
from
the
last
meeting,
and
another
thing
was
a
suggestion.
Maybe
to
have
some
drop-down
boxes
rather
than
having
to
know
there's
sort
of
different
areas
that
you
can
run
the
benchmarks
on
so
I'll
have
a
look
at
expanding
the
Jenkins
job
yeah.
A
D
Well
hi,
can
you
hit
me
yep,
okay,
so
yeah
I
looked
at
the
of
course,
I
spend
looking
at
the
no
DC
problem.
Why
we
weren't
able
to
run
that
first
of
all,
MongoDB
the
one
issue
was
I,
wasn't
able
to
start
the
MongoDB
on
different
port
non-standard
port,
which
was
two
seven
zero
one
eight
when
I
commented
that
I
think
I
can
now
run
MongoDB
process,
no
DC
runs
complete,
run
completes,
but
the
script
doesn't
exit
and
I
did
find
the
problem
today,
I
fixed
it
I
ran
it
on.
D
E
A
F
D
A
A
A
C
A
C
A
C
A
F
C
If
you've
got,
if
you
presume
you're
in
the
great
way,
well,
I
think
fact
I
think
anybody
in
the
benchmarking
group
can
go
and
launch
jobs.
Lankans
there's
a
job
in
that
core
benchmark:
node
micro
benchmarks.
Okay,
you
go
in
there
press
build
with
parameters,
there's
a
little
bit
of
text
below
each
one,
but
a
sale
aim
either
later
today
or
tomorrow.
Try
and
put
a
very
brief
guide
together,
at
least
in
that
status
of
the
rating
benchmarks
and
then
have
a
go
track
of
the
instructions.
If
it
doesn't
work.
A
D
C
There
and
essentially
the
Magette
that
the
main
bit
of
work
that
we're
trying
to
do
is
to
make
the
set
of
benchmarks
easy
to
use
when,
for
example,
review
in
a
pool
request
or
investigating
further.
If
there's
some
question
about
performance
between
two
like
that,
so
the
main
bit
of
work
is
Jenkins
and
just
arrived,
checked
into
the
benchmarking.
F
A
D
F
D
Can
spawn
as
many
processes
you
want,
so
how
do
I
say
if
you
were
to
try
that
out
on
benchmarking,
just
to
highlight
the
benefit?
Should
I
write
a
new
scrape?
It
is
something
I
can
modify
the
existing
scrip
and
try
to
set
that
up.
How
do
you
a
big
and
then,
as
you
know,
we
are
also
I'm
working.
Quite
this
server-side
rendering
use
case
right.
I.
F
A
So
I
guess
it
depends
on
what
makes
sense
like
okay,
let's
duplicate
in
the
better
right.
So
if
it's
something
where
you
could
have
a
single
script,
then
you
pass
it
an
option
that
tells
it
which
way
you
want
to
run
and
that
cut
to
cuts
down
the
duplication
of
what's
in
the
script.
That
would
be
good
if
it
means
that
now
the
script
is
ten
times
more
complicated
than
maybe
two
different
scripts
makes
that's
right.
Okay,.
F
A
D
A
A
I
removed
I
removed,
like
just
the
agenda
tag
because
I
think
the
idea
is,
we
put
it
on
the
agenda
if
there's
something
we
need
to
talk
about
in
this
meeting,
as
opposed
to
like,
if
we're
working
at
making
progress,
then
doesn't
need
to
be
tagged.
Okay,
okay
issues,
review.
The
next
thing
is
in
no
J's
bill
to
install
of
our
script
under
benchmarking
machines
that
one
I
think
was
tagged
because
we
needed
some
somebody
to
do
something
it
was
done.
A
If
it's
going
to
be
installed
on
the
machine
so
that
one
I
think,
though
it
doesn't
need
to
be
tagged
in
our
agenda
necessarily
either
I
think
I,
dare
you
opened
outright
so
I
think
in
terms
of
the
intent
of
tagging
it
you're,
probably
just
I,
guess
I'm
asking
if
you
agree
whether
removing
the
tag
makes
sense
or
not.
Yeah.
C
D
A
They're
they're
both
connected
into
the
CI
okay,
they're
being
used
for
general
jobs
right
now,
they're
also,
you
know
the
the
one
the
job
that
Gareth
is
working
on
is
targeting
them
and
we
can
also
you
know,
use
them
for
whatever
other
things.
So
it's
like
you
know
the
the
network
version
of
what
you're
doing
makes
sense
to
run
there.
Okay,
the
only
like,
so
we
in
terms
of
what
Garrett
is
running
is
setup.
A
You
know
it's
potentially
that
collaborators
will
kick
off
a
run,
that
goes
for,
say,
24
hours
or
something
right
and
I
could
interfere
with
regular
runs.
So
we
may
just
want
to
figure
out
how
we,
you
know
manage
that
across
those
two
machines,
but
otherwise
we
want
to
you
know,
put
them
to
the
best
use
we
possibly
can
I
just.
D
H
A
E
A
E
We
do
in
detail
something
in
minds
so,
first
of
all,
there's
the
octane
benchmark
yeah,
which
is
officially
deprecated,
but
that
doesn't
mean
that
we
don't
track
it
anymore
right
now,
we
finally
caught
up
after
the
switch
to
talkin
with
the
previous
performance,
and
we
might
even
get
to
javascriptcore
level
soon,
but
we
will
probably
stop
them.
Looking
at
to
obtain
specifically
I
saw
you
set
it
up
now
or
do
some
time
now.
E
So
at
least
we
catch
regressions
in
note,
if
there
is
something
different
than
in
chrome,
which
happened
in
the
past
yep,
but
this
is
only
for
regression
testing,
mostly
so,
for
driving
performance
work
in
the
future.
One
thing
we
would
like
to
set
up
is
what
we
would
call
it
a
tuning
benchmark,
which
is
this
benchmark
suite
that
contains
everything,
people
or
most
common
stuff
people
use
nowadays
to
build
web
pages
like
expert
like
table,
like
whatever
everything
that
runs
on
out
there,
so
like
10
or
15
of
the
most
part,
would
rank.
E
A
What
that
definitely
sounds
good
like
if
you
look
at
this
page,
like
sort
of
our
you
know,
the
overall
goal
of
this
group
is
to
try.
You
know
we
tried
to
identify
use
cases
and
key
attributes
that
we'd
like
to
measure
and
so
any
new
benchmarks
that
help
fill
in
the
gaps,
because
there's
still
a
lot
of
gaps
in
terms
of
oh,
you
know
even
the
use
cases
we
identified
on
and
if
there's
other
use
cases
you
think
are
important.
A
A
E
A
E
So
I
talked
to
the
Beck
developers
and
they
will
call
developers
already
and
they
are
willing
to
provide
benchmarks
or
have
already
provided
one.
So
we
have
already
have
one
from
the
back,
although
I'm
not
really
happy
with
that,
we
obviously
also
have
the
typescript
one
and
there's
a
table
front
end.
This
is
part
of
the
area,
six
benchmark
bubble
benchmark
if
you
find
with
that
that
will
probably
start
trying
to
pull
something
together
and
next
month
or
in
general
qq3
end
of
q3.
Instead
of
chief
on
what
item
for
us,
it's.
B
I
just
wanted
to
check
what
are
your
thoughts
on
measuring
the
memory
performance
or
the
garbage
collector
performance,
because
most
of
the
benchmarks
to
cover
like
the
CPU
time
and
I'd
know
like
octane.
We
had
like
couple
of
test
cases
that
says
the
garbage
collection
time,
but
I
think
it
will
be
good
to
have
those
kind
of
benchmarks
or
those
kind
of
metrics
that
we
measure
all
those
benchmarks.
So
if
you
see
those
or
the
link
that
you
have
shared
Michael,
the
matrix.
J
B
E
B
G
K
E
B
You
should
just
compare
the
memory
and
like
we
like
it.
Just
fades
I'd
like
for
all,
given
back
my
studies
at
600
Meg,
but
Chicago's
like
one
geek
and
downs
and
like
after
the
benchmark
is
done
running.
The
memory
footprint
after
page
mug
is
done,
is
same
for
chakra
and
v8,
but
how
would
we
know
that
in
fact,
when
the
bank
was
actually
running,
why
so
angry
then
we
ate?
So
we
should
have
some
way
to
measure
that
yeah.
B
E
A
Well,
I
think
the
the
idea
was
we
can
get
together
in
these
meetings
to
talk
about
those
things
as
opposed
to
waiting
to
know
coffee,
you
and,
and
also
not
everybody's,
going
to
be
there.
Okay,
so
I
think
you
know
if
I
don't
know
Benedict
if
you're
planning
to
come
every
other
week
now,
which
is
when
we
meet
oh.
A
I'm
very
interested
to
turn
your
welcome
either
way.
Anyway,
you
don't
need
to
be
an
odious
collaborator
to
join
in
the
work
group.
Theorist
feel
free
to
yeah.
So,
okay,
that's
good
to
hear
so.
Yeah
I
think
we
can
just
you
know
you
can
give
us
an
update
as
as
where
you're
at.
Ideally,
you
know
in
the
end
we'll
end
up
with
a
PR
that
learns
the
script
that
we
can
then
set
up
to
run
like
nightly
like
we
have
the
rest
of
them
and
publish
results
and
go
from
that.
The.
A
E
A
A
E
A
E
Okay,
so
the
one
or
one
of
the
main
motivations
to
improve
our
collaboration
is
also.
We
don't
want
to
end
up
in
this
situation
again
that
we
left
with
some
of
them
where
it
was
just
impossible
or
at
some
point
it
was
impossible
to
decide
whether
it's
okay
or
not
right,
I'm,
not
saying
we
would
in
a
compiler
or
anything,
but
just
making
decisions
like
switching
to
the
new
GC
is
happening.
E
A
A
I
mean
I,
guess
it's
like
the
you
know,
the
we've
been
making
I,
don't
know,
I
would
say
slow
progress
over
the
you
know
this.
Since
we
started
this
workgroup
and
if
you
guys
have
a
little
more
horsepower,
we
can
make
slightly
faster
progress
and
that's
what
I
think
we
should
just
continue
to
do.
Push
push
in
new
benchmarks
as
we
have
time
and
and
you
can
get
them
up
and
running.
E
F
E
E
A
E
It
would
be
nice
if
we
wouldn't
know
so
for
us,
it's
it's
difficult.
It
depends
on
who
you
talk
to
when
you
get
a
completely
different
answer
like
oh
yeah.
The
interaction
with
the
network
is
the
worst
thing,
because
we
have
to
switch
between
best
possible
JavaScript
to
John's.
Can
you
do
something
about
that?
So
yeah?
Okay?
We
could
look
into
that
and
the
next
thing
is
yeah,
but
this
is
not
so
important.
What's
more
important
is
to
JSON
processing,
this
is
still
or
right.
D
Yeah
can
I
some
instant
interject
here
a
little
bit
so
yeah
as
Mike
said
correctly.
It's
just
so
many
ways.
No
J's
can
be
used.
It's
really
like
a
shell
script.
You
know
so
with
no
DC
is
what
one
thing
we
try
to
do.
We
under
dueño
trance
a
single
instance
application.
Then
you
have
a
cluster
so
that
application
can
take
at
twenty
of
multiple
Kol
machines.
Then
we,
it
also
runs
in
a
microservices
mode.
D
So
same
application
is
divided
among
functionality,
so
you
can
run
in
different
machines
and
whatnot,
and
as
part
of
that,
we
measure
it
measures.
Memory
usage
is
over
every
tense.
It
collects
the
performance
for
every
10
seconds
400
seconds.
So
maybe
there
are
ten
samples
or
you
can
run
it
for
200
seconds,
but
it
collects
the
data
for
evidence
they
can
period
within
that
it
collects
the
memory
data.
How
much
memory
was
used
during
that
last
10?
Second,
and
also
it
tried
to
measure
the
the
network
bandwidth.
D
D
You
just
talked
about
the
networking
we
are
working
on
the
server-side,
rendering,
which
is
mainly
a
string
processing,
render
to
string
type
of
functionality
and
we,
as
we
see
fit
I
guess
every
time
do
we
need
to
write
a
new
total
new
application
just
to
address
that
one
use
case
or
how
should
we
that's
something
we
can
talk
about?
Should
we
extend
existing
one
or
should
we
keep
on
writing
a
new
one?
Just
read
us
a
new,
simple,
independent
use
case.
E
D
See
couple
runs
overnight
and
then
we
can
also
try
to
set
up
in
micro,
services
move
the
cluster
mode
and
it
also
generates
his
own
data
points
or
graphs,
and
we
can
look
at
those
data
points
and
see
whether
that
is
something
useful,
and
maybe
we
can
take
that
as
a
some
way
we
can
implement
as
a
infrastructure
for
the
everything
we
run
on
these
machines.
We
can
calculate
the
network
speed,
we
can
calculate
the
bandwidth.
The
memory
usage
is
anything
we
run
on
this
machine.
D
We
can
collect
those
data
points,
GC
starts,
and
then
we
can
maybe
learn
something
from
that
over
time.
So
there
can
be
some
common
properties.
We
really
want
to
track
from
the
node
core
and
then
then
comes
application.
What
fit,
what
how
we
can
implement
for
which
we
can
cut
it
across
any
application
it
doesn't
have
to.
The
application,
doesn't
have
to
do
that.
We
toward
node
DC
does
today
it
doesn't
have
to
do
but
there's
something
infrastructure
which
does
that
anything.
You
run
on
that
machine.
You
can
think
like
that.
Maybe.
D
C
D
C
Think,
especially
from
the
point
of
view
that,
if
obviously
the
more
benchmarks
and
tests
that
we
start
running,
if
we
then
start
sort
of
filling
out
onto
multiple
machines,
we
would
have
to
be
very
careful
that
we
don't
then
go
and
try
and
make
comparisons
between
Aaron.
That's
run
this
week
on
machine
one.
The
essays
are
in
the
strength
to
need
to
go
on
machine
to
don't,
if
that's
the
felt
one
of
the
things
you're,
meaning
by
having
sort
of
a
common,
a
comment,
sort
of
things
that
we're
trying
to
collect.
D
The
reason
I'm
saying
a
lot
of
times
all
these
writing
these
small
apps
is
very
easy.
It's
not
a
very
table
difficult
job,
but
coming
with
everyone
trying
to
implement
their
own
way
of
Coligny
all
these
different
data
points,
and
then
that's
what
probably
takes
most
time.
That's
what
we
realized
when
we
were
writing
this
note
DC
is
the
application
has
not
changed
since
we
wrote
first
time,
maybe
a
couple
of
times,
but
the
most
of
the
change
of
the
time
spent
was
on
developing
his
client.
D
The
driver
program
which
collects
different
statistics
generates
graphs,
litigate
the
data,
and
that
takes
most
time.
So
that's
where
I
was
thinking.
Would
it
be
a
good
idea
to
have
some
we
spend
thinking
at
least,
can
there
be
common
infrastructure
we
enable
when
we
have
anybody,
run
node?
That's
it
right.
A
D
I
mean
notice
he
doesn't
have
to
be.
It
can
be
anything
else
if
he
come
up
with
a
very
simple.
That's
fine,
I'm,
just
thinking
how
we,
how
we
can
minimize
the
time
spent
on
developing
this
infrastructure
and
just
writing
different
use
cases.
You
know
service
and
and
and
Jason
processing,
micro
services
and
all
those
things,
but
we
don't
have
to
then
every
single
day
we
don't
have
to
worry
about.
Are
we
getting
the
right
kind
of
data
back.
D
Yeah,
it's
a
very
simple
Python
client
driver.
We
wrote
which,
unlike
a
run
spec
no,
it
is
kind
of
based
on
a
spec
CPU
how
the
spec
CPU
has
a
runs,
make
purple
script,
but
it's
just
a
Python
script,
which
it
populates.
Your
data
base
generate
the
URLs
randomize,
the
URLs
good
enough,
so
that
it
can
get
request
and
the
post
request
on
the
delete
requests
are
distributed
equally
in
every
10.
Second
runs
because
there's
so
many
ways
you
can
skew
the
results.
If
you
don't
really
run
a
consistent
ways,
if
I.
F
D
The
get
request,
for
example,
application
we
get
a
different
data
there
in
verse
so
but
we'll
be
collecting
10
seconds
data.
We
want
to
make
sure
that
10
seen
as
a
good
enough
sample,
so
it
can
give
a
good
result
at
end
of
the
run.
So
those
things
are
probably
common
across
like
may
know
DC
and
may
be
something
new
tomorrow
and
also
the
what
is
the
end
result.
What
is
the
moment
for
a
memory
footprint?
D
E
Yeah,
so
we
have
been
quite
successful
using
these
users,
what
we
call
user
stories,
which
is
just
some
web
page,
that
you
need
to
load
and
then
some
user
actions
that
are
recorded
and
replayed
on
something
similar
like
that,
and
something
as
easy
to
use.
Leon
would
be
great
for
note
as
well,
and
obviously
telemetry
also
provides
you
with
all
the
so
you
can
provide.
E
A
D
A
That's
a
good
idea,
so
I
mean,
if
you
can,
you
know,
submit
a
PR.
That
is
like
the
starting
point
of
what
you
think
we
should
document
in
terms
of
that
list.
We
could
then
have
everybody
else
chimed
in
and
hopefully
end
up
with
the
document,
because
these
are
the
these
are
the
kind
of
data
things
that
you
know
we're
looking
to
capture.
Okay,
I,
don't
maybe
suggest
like
a
separate
document,
a
succoring
like
in
that
say
the
benchmarking
doc,
like
the
supper,
m.d.
document.
D
Yeah
I
mean
my
thinking
overall.
Is
they
said
someone
new
customer
comes
along,
they
download
the
node
benchmarking
just
to
verify
how
the
Machine
our
hardware
they
have
is
behaving.
They
can
run
this
and
it
generates
all
these
data
points
and
then
they
can
understand.
Where
are
the
this
at
the
infrastructure
level?
What
are
the
bottleneck
is
my
network
is
not
good
enough.
Do
I
need
a
higher
bandwidth
network
card.
A
Well,
we've
only
gotten,
as
far
as
using
this
regression
tool
to
say,
let's
at
least
try
and
track
things
and
make
sure
we
don't
regress.
But
that
means
we
wouldn't
like
be
trying
different
network
cards
and
stuff.
But
what
you
suggest
is
also
an
interesting
thing
for
people
to
say
how
good
is
what
I've
set
up
and
where
the
shortcomings.
D
But
these
sort
of
things
as
long
as
we
are
making
it
easy
for
the
end
users,
we
can
have
more
adoption
of
these
things
and
what
we
want
is
with
people
we
want
people
to
use,
node
and
whatever
node
comes
with
testing
wise
performance,
wise
and
benchmark
wise,
they
don't
want.
You
don't
want
customers
to
go
and
maybe
write
a
benchmark
and
try
to
understand
it.
Just
too
much
time
spent
on
that.
D
A
Yeah
I
mean,
from
my
perspective,
I
I
agree.
That's
useful
and
just
my
prior
prior
priority
would
be
covering
off
the
existing
use
cases
that
we
have
just
to
make
sure
we
don't
break
anything
before
we
cover
the
other
one,
but
not
any.
You
know
it'd
be
good
to
make
progress
on
both
just
if
I
have
time,
I'd
probably
work
on
the
the
first
one
until
we
get
that
cover
off
a
bit
better.
C
We
should
all
say
so
if
remember
that,
as
we're
talking
at
the
beginning
about
the
very
wide
range
of
ways
that
people
are
using
node,
we
shouldn't
ever
think
that
we'll
be
able
to
measure
absolutely
every
single
way
that
somebody
is
using
node
and
expect
to
be
able
to
find
absolutely
any
regression.
That's
gonna
hit
any
end
user.
Obviously
it
can
be
a
goal,
but
I
think
with
the
sort
of
downloads
that
we're
seeing
of
nodejs
I
think
we
would
be
very.
A
It's
okay,
so
we
got,
we
have
ten
minutes
left
I
mean
I,
think
Benedict.
It
sounds
like
you're,
maybe
gonna
come
back
with
like.
Is
there
a
write-up
or
anything
that
you
can
share
in
an
issue
to
say:
hey,
you're,
going
off
and
working
on
this?
Just
so
that
there's
a
bit
more
information
that
if
somebody
else
comes
on
they
can,
they
can
see
yeah.
E
I
just
started
actually
and
to
be
honest:
I
was
on
vacation
last
week,
so
I
didn't
even
stop
my
new
job
and
I
start
a
little
leftovers
from
my
old
one
right.
So
this
is
really
so.
The
only
thing
that
we
are
doing
for
sure
now
is
that
ruling
benchmark
and
everything
beyond
is
up
for
planning.
Okay,
we
are
trying
to
get
an
idea.
A
A
Get
some
feedback
it
informs
what
you're
doing
and
then
you
know
at
the
end.
We
also
add
that
benchmark
into
the
results
we
can
kind
of
report
and
keep
an
eye
on
yeah
definitely
sounds
good,
okay
and
I
think
otherwise
we're
gonna.
You
know
the
between
now
and
the
next
meeting
likely
the
the
focus
will
be
getting.
The
DCE
is
stuff
up
and
running
anything
else.
People
have
on
the
go
that
you
should
talk
about.
I.
A
E
E
You
have
no
idea
what's
running
note
service,
because
all
of
this
is
usually
closed,
source
yeah
some
contractor
work
right
so
that
so
Netflix
promised
us
so
give
us
some
insight
into
their
huge
note
right
application
and
we
have
some
insights
from
but
yeah
the
Bloomberg
scripts.
This
is
really
crazy.
What
they
are
doing
so,
probably
not
representative,
okay,
they're.
B
Actually,
a
neg
Mary
bench,
money
itself
is
so
like
simple
one
pile
benchmark,
but
the
interesting
part
is
the
new
models
that
it
use,
especially
X,
yes,
and
that
has
like
lot
of
type
system
in
things
are
on
as
investigating
Oneg
merit
performance
or
check
recover.
So
I
know
that
so
yeah
I
mean
like
unless,
like
we
have
something
similar
like
well
the
world
kind
of
benchmark,
we
can
swab
the
deck
man.
B
B
B
A
A
But
it
though
the
thing
the
thought
that
does
come
to
mind
is
I
know
through
the
community
committee.
One
of
the
things
that
focuses
I'm,
gonna,
try
and
put
some
effort
into
is
like
community
outreach,
and
so
that
might
be
something
we
try
and
sync
up
with.
You
know
to
say:
okay.
Well,
if
we
want
to
get
more
direct
feedback
from
end
users
into
what
make
the
right
benchmarks,
we
might
try-
and
you
know,
if
they're
doing,
surveys
and
stuff
like
that,
we
could
probably
add
some
content
for
surveys
along
that
line.
A
So,
like
you
know,
for
example,
if
you,
if
you
think
you
can
think
of
like
the
top
ten
questions,
we'd
like
to
ask
end-users
in
terms
of
you
know,
you
know
creating
a
benchmark.
If
we
can
open
that
up,
discuss
it
at
an
issue
here,
then
maybe
we
can
go
over
to
the
community
committee,
slash
foundation
and
say:
okay,
you
know
how
could
we
get
this?
These
questions
in
as
part
of
the
survey.
G
A
If
you
guys
can
go
and
run
the
survey
yourself
already
and
make
sure
that
that's
great,
so
don't
don't
stop
that.
But
if
you,
if
you
think,
leveraging
the
foundation
overall
in
the
community
to
try-
and
you
know,
make
those
surveys
go
out
on
behalf
of
nodejs
community,
we
I
think
there's
opportunity
to
do
that
as
well.
I.
A
Community
right
so
I
think
if
we
can
come
up
with
the
question
like
if
you
know
all
the
questions,
if
you
said
you
know
here
ten
good
questions
we
should
put
in
the
rest
of
the
people
in
this
group.
You
know
we
can
try
a
minute.
We
have
any
other
ideas
and
then
we
can
go
and
see
how
we
can
make
that
happen.
A
A
A
E
A
A
To
make
it
yeah
and
then
over
time
like
you
know,
we
can
talk
about
just
it.
This
time
really
isn't
doable.
We
can
talk
about
like.
Can
we
adjust
it
to
another
time?
It's
a
bit
well
like
for
all
the
groups,
it's
hard
to
find
one
that
works
perfectly
for
everybody,
some
of
the
other
ones
we've
like
had
alternating
times
and
stuff
like
that.
So.
A
E
A
Okay,
so
I
think
our
time
is
up
we're
at
the
end
of
the
hour.
Now
we
do
usually
give
people
on
who
are
watching.
If
you
have
any
questions,
just
chime
them
put
them
into
the
original
issue
for
the
meeting
which
I
don't
know
if
people
were
watching
can
see
the
shot
but
I'll
paste
it
there.
Just
in
case
it's
issue
134
in
the
benchmarking
workgroup.