►
From YouTube: Node.js Build WG meeting - March 12 2019
Description
A
A
There
has
been
some
discussion,
though,
that
we
should
start
out
with
at
least
shorter
a
short
discussion
on
platforms
versions
and
stuff
like
that
for
no
12,
since
that's
timely,
so
we'll
start
with
that,
and
if
we
have
time
we'll
switch
back
to
the
the
other
issue,
that's
as
opposed
to
following
our
regular
agenda,
which
we
agreed
we'd
sort
of
skip
for
this
week,
so
rod
I,
think
you'd
mentioned
or
not
sure,
but
maybe
it
was
Richard.
Who
mentioned
that
we
should
look
at
this
and
you'd
sort
of
echo.
B
C
D
A
D
D
A
D
B
So
so
this
we
have
to
be
careful
with
it.
We
got
two
things
here:
we've
got
what
we
officially
support,
which
is
basically
what
we
will
run
tests
for
yep
and
then
what
we
build
on.
So
there's
two
things:
ranker.
What's
the
minimum
support
for
each
of
these
things?
Yes,
I'm?
So
how
about
I
copy
that
then
system.
A
B
C
B
C
A
A
B
D
C
B
And
the
other
thing,
that's
that's
important
to
look
at
is
actually
the
version
select
a
script
in
Jenkins.
We
should
highlight
that
for
being
one,
that's
listening
in
to
see
what
this
that's
the
status
of
our
test
and
release
in
frustration,
yep
and
I
might
give
you
a
link
to
that.
So
you
can
click
on.
If
you
don't
have
it.
C
A
B
A
A
Well,
I've
been
keeping
I,
I
mean
I,
have
a
on
that
one
I've
been
tracking
the
discussion
because,
like
we
need
to
get
ready
to
upgrade
our
machines
with
the
right
compilers
and
we're
starting
to
look
at
that,
the
you
know
in
terms
of
GCC
they're,
worse
I
think
it
was
smart.
Ass
doesn't
have
GCC
eight,
which
is
one
of
the
suggestions
and
IBM
I
only
has
up
to
six
three
know.
So
if
we
do
update
my
suggestion
as
we
go
to
six
three
and
no
higher.
E
D
A
E
They're
not
a
little
bit
of
turbulent
issues.
One
is
what
do
we
test
with,
and
the
other
is
what
we
build
with
you
know.
So
there
are
many
benefits
to
building
with
the
latest
compiler,
which
includes
the
latest
optimizer
and
which
can
pick
like
there
are
certain
if
depth
inside
the
code
to
use
better.
E
C
E
That
will
allow
us
to
at
least
you
know
not
get
stuck
with
the
situation
that
we
are
now
like
we're
building
with
four
point:
eight,
which
is
an
ancient
compiler.
So
like
get
like,
they
say
like
align
ourselves
with
it's
not
even
state-of-the-art,
because
they're
like
like
clang
gate,
C++,
twenty
get
ourselves
situated
for
relief
and
tried
it
and
keep
support
for
for
things
like
yeah.
A
E
A
E
E
C
A
I
mean
my
main
concern
is
that
we
don't
push
the
minimum
to
the
point
where
we
can't
do
that
on
some
platforms
and
that
we
test
it
well
enough,
so
that
it's
you
know
we're
not
using
say
six
only
four,
you
know
subset
of
the
platform,
so
we
actually
test
out
on
all
of
them.
So
it's
it's
basically,
you
know
needed
its
ensures
will
continue
to
run
on
that
level,
but
then,
if
we
want
to
make
the
release
ones,
something
higher
that
that
doesn't
seem
like
a
bad
idea.
A
E
E
E
D
A
E
A
I
think
I'm
understanding
the
proposal
that
would
be
like
our
options
are
leave
for
nine
four
is
the
minimum
level
move
up
to
something
like
6.3?
Is
the
minimum
level
or
push
for
something
higher?
It
sounds
like
what
you
might
be
suggesting
is,
like
we've
moved
the
minimum
level,
2
6
3
ish,
but
then
in
our
release,
machines
actually
look
to
build
with
something
new
where,
if
we
could
yeah.
E
A
E
E
D
B
He's
security
in
theory
he's
the
other
concern
that
we
we
can't
lose
sight
of,
which
is
that
there
are
people
out
there.
There
are
lots
of
people
out
there
that
build
this
themselves
and
for
the
for
the
platforms
that
we
don't
officially
support
in
our
binary.
They
will
build
themselves.
So,
on
the
one
hand,
we
got
pressure
to
upgrade
C++
level
and
better
compilers.
B
B
Fact
that
v8
ship
clang
themselves
in
their
repose
talks
about
the
difficulty
of
that
whole
tool
chain.
It's
not
as
simple
as
saying
I
am
just
great
playing
this,
but
this
minimal
version
of
compiler.
It's
it's
really
not
a
trivial
process.
I
mean
piling.
A
compiler
is
not
a
fun
particular
but
just
say:
clang,
clang,
I
think
about
systems
that
are
resource,
slash,
brain
played.
E
B
Towards
newer
stuff
and
I
understand
the
pressure
from
v8,
although
I
don't
agree
with
it,
I
think
they
need
to
just
hold
up
a
bit,
but
let's
accept
that
we've
got
your
precious
here,
so
let's
not
gallop
for
too
much.
Now,
if
you
look
at,
if
you
look
at
some
of
the
life
cycles,
I've
just,
for
instance,
are
putting
in
the
top
one
there
in
Linux,
x64
I've,
put
in
some
proposed
example
minimum
distros
that
we
would
support
and
then
the
EO
else
for
them
now.
B
D
B
B
Stop
comes
even
testing
on
CentOS
six,
so
that
by
that
same
metric
we
would
stop
testing
no
12
on
Debian
8
and
only
ship
binaries
on
Debian
9.
But
these
people
running
these
operating
systems
have
got
a
year
and
a
bit
of
legitimate
support
from
their
end
coming
in,
but
then
binaries
from
nodejs
that
org
that
probably
won't
run
on
their
systems.
C
E
They
want
us
to
community
effort
and
community
feel
so
that
that's
out
there
and
it's
a
possibility
and
for
instance,
we
will
dropped,
lena
36
compatibility,
but
God
hasn't
broken
and
I,
and
it's
like
it's
over
a
year
that
we
haven't
tested
or
shipped.
It
is
it's
binaries
and
then
the
cord
works.
So
it's
it's
a
trade-off.
It's
not
that
we're.
Like
really
cutting.
E
E
E
E
B
A
Is
what
we
currently
have
my
goal
is
to
try
and
get
I
mean
I
have
some
CentOS
7
6
machines
now
that
those
are
available
at
OSU,
but
getting
people
getting
you
know
lined
up
to
get
those
configured
in
building
gonna
take
some
time.
So
first
I
want
to
get
the
compilers
to
whatever
level
they
need
to
be
so
we
can
continue
to
build
and
then
we'll
work
on
moving
over
to
CentOS
x7
sex
I
mean
that
shouldn't
affect
the
binaries
in
terms
of
it
actually
sent
us
76.
B
B
We're
still
going
to
be
supporting
these
platforms.
Do
we
still
do
we
think
we're
gonna?
Have
the
manpower
to
do
that
and
are
we
gonna
be
around
and
all
those
four
things,
the
more
compatibility
we
keep
the
more
we
are
signing
up
for
a
lot
of
work
and
we've
felt
that
pain
in
the
past.
So
it's
it
is
worth
us
creeping
forward
just
for
our
own
sanity,
just
I.
E
Could
also
like
release
the
hard
dependency
that
we
made
between
OS
and
release
and
I
know
that
in
the
past
it
was
a
hard
constraint,
but
if
it's,
if
it's
gonna
be
you
know
good
enough
to
switch
release
os's
midstream,
then
we
could
do
it.
So
if
it
doesn't
break
quiet,
it's
a
possibility.
It's
not
inconceivable
that.
A
A
E
A
B
Mean
Raphael
is
but
I
ride
that
they
did
technically
everything.
We
know
it
should
be
okay,
but
the
tragedy
of
having
a
release
go
out
that
people
can't
run
is
pretty
hang
on.
Yes,
particularly
as
it
gets
longer
answer.
It's
not
it's
not
like
we've
done
book
releases
in
the
past,
though,
like
we
have
yeah
the
bad
and
then
we've
had
to
backtrack
yeah.
We
could
do
that.
It's
just
that
that
trust
level
in
LTS
is
such
a
such
a
huge
commodity
for
for
node.
Well,.
E
D
D
C
D
E
E
A
A
A
C
A
C
A
B
B
C
E
E
B
D
E
A
A
It
does
sound
like
later
versions
of
the
dev
tools
that
may
be
available
on
CentOS
for
power
and
for
us
390
axe.
Unfortunately,
it
may
be
bigger
than
higher
than
sex.
So,
to
start
with,
we
may
you
know
start
with
what
we've
done
already,
which
is
to
get
the
compilers
build
the
compiler
GCC
six
separately,
and
it's
not
getting
any
worse
than
it
has
been
for
existing
versions.
A
A
My
plan
is
just
to
say:
we're
gonna
continue
to
do
what
we're
doing,
which
was
like
we
installed
the
494,
which
wasn't
the
default
compiler
and
we'll
do
the
same
thing
for
six
on
on
the
idea
of
platforms
to
start
and
then
hopefully
for
the
next
release.
If
we
can
get
the
the
dev
tool
set,
I
hope
that
we
could
move
up
to
those.
B
D
B
A
B
C
E
E
E
B
Like,
let's
not
lose
sight
of
the
fact
that
Debian
is
used
as
the
base
of
other
distros
as
well,
so
it
is
important
for
many
reasons,
I,
but
you
know
having
been
there
having
to
administer
Jesse
systems
and
watch
that
support
slip
slowly
away.
This
is
more
than
a
problem
for
more
than
just
node.
It's
it
is
it's
a
mess.
C
B
A
D
B
B
A
B
A
I
mean
I
guess
arm.
Does
sound
like
it's
gonna
be
problematic,
though,
because
the
comment
in
the
arm
six
was
that
the
cross
compilers
don't
even
support.
Wasn't
it
yeah
that
the
cross
compilers
don't
support
targeting
arms?
You
know
the
newer
compiler
newer,
cross
compilers,
don't
support
targeting
arm
sex
anymore.
C
B
E
B
B
B
E
C
E
E
B
E
E
A
E
A
E
E
Like
they
return
as
doing
releases
similar
to
two
releases
per
year
and
like
every
fourth
release,
is
gonna
be
in
FPS,
but
there
they
are,
they
are
pushing
equal
to
upgrade
to
like
the
need,
mid-cycle
releases,
and
we
have
none
of
those
like
we
have.
The
our
latest
is
2018,
which
is
already
like
three
releases
behind
the
latest,
and
we
have
like
we've
been.
E
D
E
B
A
E
I'm
not
sure
that
that's
they're
like
well,
that's
the
priority
they're
getting
like
specifically
I'm
talking
about
janae
I'm,
not
sure
that
that's
the
priority
they
are
getting
from.
Microsoft
we've
been,
we've
been
taking
them.
You
know
some
level
for
granted
and
and
Microsoft
has
given
them
other
priorities
to
focus
on
and,
as
a
consequence,
we've
gone
down
and
their
priority,
and
we
need
to
push
that
back.
A
little
bit
put
back
up.
E
B
C
E
B
C
E
A
E
And,
and
and
like
the
matrix,
not
exactly
the
matrix,
but
the
condition
for
what
we
test
with
what
is
it's
like
this
1k
character,
if
sentence
in
the
job.
B
C
B
B
E
C
A
I
think
it's
in
and
for
me
it's
what,
if
the
people
you
know,
the
people
from
Microsoft
should
be
able
to
best
comment
on
well
at
least
what
they
think
is
important
right.
Cuz,
like
rod,
said
they
do
a
lot
of
the
work
and
I
can't
see
us
substituting
and
doing
replacing
that
work.
If
we
think
something
different
is
important
right
unless.
E
E
E
It's
a
really
big
problem,
so
it's
a
big
problem
problem
that
if,
if
there
was
a
possibility,
I'll
push
out
to
do
the
foundation
because
Windows
usage
for
tooling
and
for
development
outside
of
the
US,
it's
like
it's
8090
percent
and
and
the
u.s.
is
very
chauvinistic,
never
mind.
That's
about
the
same
windows.
C
E
So
I'm
not
saying
that
it's
a
fault
like
I'm,
saying
that
they're
so
big
that
Peter
coordinate.
We
need
to
be
like
their
project
manager.
We
should
we
need
to
tell
them
like
what
we're
missing
I'm
sure
that
they'll
be
very
accommodating
all
the
people
that
I've
talked
with
very
happy
to
help,
etc.
We,
like
the
node
project,
hasn't
been
pushing
enough
to
get
those
things
and
I
was
gonna,
say
entitled
and
we're
entitled
to
them,
because
it's
a
huge
thing,
although
lesson
production
but
but
definitely
development
usage,
it's
a
huge
thing.
A
C
A
A
We
are
five
minutes
past
seven
for
me
anyway,
just
wondering
if
we
can
loop
back
see
if
we
get
closure
on
the
table
here
and
I
guess
arm
was
one
of
the
things
we
still
needed
to
talk
a
bit
about
right.
B
B
It's
just
it's
been
annoying
to
have
to
support
older
versions
of
smile
OS
for
long
periods
of
time,
so
I
just
created
an
instance
to
see
if
I
could
figure
out
what
the
compiler
level
was
on
the
button
time.
Anyone
seen
any
reason
why
we
shouldn't
just
like
we
can
pass.
We
can
pass
this
on
to
:
yeah
I
would.
A
E
C
A
E
A
E
E
A
E
E
E
A
E
So
I'm
I'm
come
ambivalent.
On
the
one
hand,
I
said
I
wanna
say
I,
don't
care.
On
the
other
hand,
it
would
keep
the
the
level
of
support
we
do
right
now,
which
is
a
pure
64
machines
hardware.
We
just
test
it
as
if
it's
32
bit
and
that's
the
guarantee
that
we
do
that's
what
we've
been
doing
for
a
sorry.
E
E
A
E
C
C
B
E
B
E
C
E
B
A
A
D
A
D
A
You
know
and
I
don't
I,
don't
know,
I,
think
well,
I!
Guess
it's
like
how
much
of
a
priority
would
be.
Would
it
be
versus
all
the
other
stuff
that
we
need
to
get
up
like
if
we
moved
our
building
up
to
12,
which
we
already
have?
Is
that
going
to
be
a
you
know,
a
disaster
versus
trying
to
push
for
13th
and
it
sounds
like
14-
is
just
not
worth
the
effort
yet
right
now,
anyway,.
B
Mekka
waste
deployment
target
10,
7,
you're
right
so
I
mean
I,
just
don't
I,
don't
know
if
we
can
know
any
compelling
reasons
about
this.
Maybe
this
is
a
question
for
the
VA
team
and
maybe
early
would
be
able
to
answer
this
for
us
if
we
left
it
as
10
11
with
deployment
target
team
7
is
that
gonna
cause
even.
E
E
We're
building
that
we're
using
a
new
version
of
of
Xcode
and
the
Xcode
build
tool
set
and
that's
what
we
do
create
again.
They
compile
their
own
client,
so
we're
having
problems
that
Apple
doesn't
want
to
fix
because
they're
saying
you're,
using
like
a
ridiculous
all
Xcode
or
like
all
the
Xcode
Oh
new
Xcode
on
an
old
OS.
E
B
I
idol:
this
is
the
quick
thing
I'd
like
to
post
in
github
and
pull
in
v8,
probably
Ally,
specifically
and
libuv
probably
been
particularly
to
help
solve
that
one
out.
I.
Just
don't
think
we
can
answer
that
here.
It
would
be
nice
to
bump
it
up
further,
but
and
what
are
our
options
we
can?
We
could
go
to
ten
twelve
ten.
A
Twelve
is
I
think
easy
because
we
already
have
those
machines
in
the
farmers.
Ten
thirteen
I
think
we
have
images
on
in
Mac
stadium.
We
just
have
to
provision
them
and
configure
them,
and
you
know
just
find
the
time
to
actually
do
them.
Fourteen
it
sounds
like
from
what
we
fact
is
saying:
there's
just
a
fundamental
issue
that
needs
to
be
solved
before
it
be
virtualized,
virtualizer,
ball,
yeah,.
A
A
B
A
B
C
B
With
Janee,
so
let's
get
an
arm
yeah,
so
I
I
think
we
can
locking
in
Debian
nine
is
gonna,
be
sensible,
forearm
for
our
v7
I.
Don't
think,
there's
any
problem
with
that.
V6
goes
back
to
that
discussion
on
github
and
as
as
we're
discovering
it's
the
PI
zeros,
a
good
idea,
but
very
non-trivial
to
set
up
a
cluster
of
them
to
do.
The
testing
and
I
haven't
responded
to
Brian's
question
there
about
failure.
Do
they
track
for
the
failures
on
our
v6
track
failures
in
monthly
seven,
but
my
response
would
be
yes.
B
The
only
the
main
failures
we
get
on
arm
is
to
do
with
resource
constraints
and
I
can't.
Over
the
last
time
we
had
an
arm
failure
that
was
arm,
specific
or
particularly
arm
v6.
Specific
right,
I
would
say.
Moving
it
to
experimental
is
very
like
unlikely
to
cause
too
many
problems
for
arm
v6
users
and
the
only
the
only
thing
is
providing
binaries
for
them
right.
B
B
Yes,
no!
No,
but
there's
no
point
is
that
so
the
question
here
is:
do
we
just
drop
them
for
notes
no
twelve,
but
find
a
way
to
provide
them
with
wineries?
That
would
keep.
That
seemed
seems
to
me
that
that
would
keep
the
community
happy
and
that
would
be
the
best
compromise
between
our
needs
and
their
needs
now
needs
of
maintenance
and
then
in
the
cost.
Yes,.
E
B
My
proposal
would
be
let
my
present
would
be
for
now,
let's
put
our
v6
as
experimental
for
no
12
and
commit
ourselves
to
finding
a
way
to
get
them.
Binaries
and
I.
Don't
think
that
I,
don't
think
we
had
at
commitment
has
to
be
we're.
Gonna
provide
you
with
node
1200
binaries
on
the
day
it
comes
out,
but
I
think
working
towards
a
channel
for
these
non-standard
binaries
and
I
mentioned
things
like
yeah,
well,
you're
in
your
city
in
x86,
we
fail
exactly.
E
B
X64
with
Musil
would
be
a
huge
boon
for
alpine
people
if
we
can
provide
binaries
for
that
through
some
other
channel,
because
we
I
have
been
and
I
think
others
have
been
resistant
to
providing
them
through
the
official
channels
because
of
just
this
proliferation
of
official
binaries.
We
have
to
support,
and
you
know
if,
if
something
breaks
or
you
know
something
doesn't
work
for
release,
then
everyone's
got
to
be
all
hands
on
deck.
B
A
B
Maybe
so
what
I
was
imagining
is
a
a
github
repo,
maybe
not
even
then
in
the
node.js
org,
possibly
not
that
people
could
contribute
scripts,
maybe
docker
containers
things
that
can
be
true
triggered
and
maybe
the
node.js
org
could
provide
some
infrastructure.
But
if
you
cuz,
we
could
do
a
lot
of
these
things.
In
docker
containers
like
you,
can
spin
up
a
docker
container,
grab
cross-compiled
tool
chains
and
compile
armed
v6
binaries
in
a
docker
container
same
with
Musil.
All
these
other
things.
B
We
could
do
a
lot
of
these
things
in
docker,
so
you
could
have
a
repo
that
contains
all
of
that
and
some
basic
scripting
that
you
know
tests
for
new
releases
and
then
triggers
these
things
on
some
infrastructure.
We
could
have
one
server
doing
all
of
these
things.
You
know
sequentially
and
they
just
because
there's
no
hurry,
and
so
the
the
burden
of
having
hardware
is
lower.
Unless
you
want
to
come
in
with
something
like
FreeBSD
or
net
bsd
or
something
else.
B
Yeah
and
yeah,
but
if
it's
unofficial
then
also
the
security
burden
is
lower,
so
we
can
weaken
the
notice
or
can
put
disclaimers
saying
we
are
not
officially
related
to
this
project.
This
is
done
by
some
of
our
members
and
others.
If
you
want
to
trust
these
binaries,
then
it's
up
to
you
and
we
could
put
a
big
disclaimer
on
the
right
on
whatever
page.
We
have
for
these
downloads
that
this
is
and
I've
seen
that
the
play
near
places
like
you're,
going
to
repose
for
all
sorts
of
things,
people.
A
And
I
guess
the
real
difference
would
be
that
the
the
real
one
of
the
big
differences
also
would
be
like
it
wouldn't
be
tested
for
every
pull
request.
So
it's
something
snuck
in
that
was
arm
six,
but
specific
it
might
break
one
of
these
things
would
would
you
know
eventually
it's
not
cover
it,
but
you
know
that
could
take
a
few
weeks
to
resolve
and
but
whatever
right,
yeah.
B
And
collaborators
would
be
welcome
to
set
to
go
into
an
issue
that
says
the
Oh
even
get
a
binary
for
these
versions
to
say:
look
they're
not
officially
supported.
If
you
want
to
make
it
work
again,
you
can
put
in
the
work
or
find
some
yeah
Ken
that
should
be
okay,
so
I'm
open
enough
to
do
which
is
open.
A
new
issue
with
some
rough
proposal
around
that
and.
A
D
A
C
B
If
this
thing
has
problems
like
trust
or
security,
or
even
time
problems,
we
we
don't
want
to
have
to
take
responsibility
for
it
right
and
we
and
we
and
we
and
we
would
prefer
a
community
to
develop
around
it.
So
these
are
all
these
aren't
v6
people
that
showed
up
in
that
issue,
yeah
they
that
can
be
the
community
that
relatives
around
those
aren't
v65
binaries
and
maintains
them
great
like
there
will
be
a
time
when
the
tool
chain
that
we
using
doesn't
work
anymore
and
somebody
needs
to
fix
it
who's
gonna
do
that.
C
A
B
A
I
mean
and
yeah
and
encourage
that
group
to
come
up
with
like
if
there's
enough,
people
that
come
together
to
say,
let's
come
up
with
a
build
option
that
maybe
less
frequent
and
whatever
that
makes
sense,
then
that
sounds
good.
I
think
the
build
option,
though,
will
need
to
be
like
the
PI
zeros
don't
help
because
they
just
don't
have
enough
memory.
I,
don't
know.
E
E
A
E
B
Yeah
well,
I
would
hope,
see
if
we
constrain
it
enough
and
sends
and
hold
on
an
arm's
length.
I
would
hope
that
others
show
up,
and
so
you
might
get
somebody
showing
up
with
Solaris
machines
and
saying
look
I
we're
a
company
that
does
Solaris
we'd
like
to
have
binaries,
but
we
and
we've
got
machines
because
one
of
the
problems
we
have
now
is
that
all
of
our
energies
and
all
of
our
sponsorship
focused
on
these
official
pipe
platforms,
and
you
don't
have
room
for
these
others
to
come
and
contribute.
E
B
There's
a
lot
there's
a
lot
of
activity
happening
in
risk.
Five
right
now.
May
you
know
perhaps
we'd
see
a
risk.
Five
movement
for
node
I,
don't
know
if
it's
even
possible
with
the
v8,
but
then
it'd
be
cool
to
see
that
emerge
and
if
we
could
help
foster
that
by
having
a
space
dedicated
to
these
quirky
things,
that'd
be
great
anyway,
I'll
formulate
an
issue
for
that.
We
can
have
a
discussion.
Yeah
sounds
good.