►
From YouTube: Platform Sync: 2020-09-02
Description
- Milestones
- Windows Runner
--- Moving WCOW from Packet back to GHA managed runner
--- Packer automation for both LCOW and WCOW
- WSL vs WSL2
A
A
A
A
That's
still
a
large
epic
of
work
that
is
involving
travis,
who
isn't
here
and
supra,
which
is
a
contributor
that's
helping
out
with
some
of
the
pieces
there
we
it
still,
you
know,
still
has
a
good
level
of
of
work
and
effort
that
needs
to
be
put
in
before
it's
fully
functional.
But
it
is
something
that's
top
of
mind
right
now.
A
C
I
was
out
for
a
lot
of
past
week
and
a
half,
but
recently
I
did
a
bit
of
work
trying
to
make
the
packed
cli
docs
a
bit
more
discoverable,
so
there's
two
pr's
open
for
opening.
For
that
I
did
I'm
trying
to
right
now.
There
was
a
request
about
a
bit
of
a
while
back
to
to
make
the
quiet
mode
like
spit
out
the
name
and
shot
at
the
end.
So
I'm
fixing
that
up
right
now,
it's
not
fully
quiet
right
now.
C
C
A
E
D
D
D
So
I
might
have
you
all
and
thank
you,
javier,
for
the
feedback
on
that.
I
was
going
to
tidy
it
up
a
little
bit,
but
if
y'all
want
to
give
it
a
test,
run
it's
in
the
issue.
I
can
share
a
link
around
too,
but
if
you
have
any
needs
for
windows,
vm
also
feel
free
to
reach
out
the
just
one
thing
to
be
wary
of,
and
I
want
to
get
your
feedback
on.
If
you
have
thoughts
on
this
is
the.
D
If
you
leave
a
vm
running
it'll
cost
you
about
100
bucks
a
month,
not
nothing
to
run
a
you
know,
spend
a
few
hours.
Unpack
is
a
dollar,
so
you
know.
Wouldn't
be
too
much,
but
if
y'all
have
better
ideas
for
making
sure
these
things
shut
down,
I
just
threw
up
timed
shutdown
is
the
first
command
you
run
on
these
things
eight
hours,
but
if
you
all
have
other
thoughts,
I'd
love
to
hear.
A
B
B
Over
the
past
week
we
were
working
with
natalie,
specifically
to
get
it
revived.
We
were
having
plenty
of
issues
with
just
running
normal
prs,
and
you
know
eventually
we
got
it
through
and
you
know
going
through
the
docks
and
pairing
on
it.
We
did
notice
that
the
ducts
could
be
spruced
up
quite
a
little
bit
just
to
make
it
a
little
bit
easier,
so
we
did
go
ahead
and
do
that,
but
in
general
I
think
you
know
we're
just
sort
of
cognizant
that
these
packet
workers
require
a
little
too
much
hands-on,
one-on-one
time.
B
A
A
All
right
going
once
gone
all
right.
That's
all
you
get
okay,
so
yeah,
I
guess
release
planning
there
really
isn't
anything
quite
yet.
We
do
have
a
milestone
for
o14o
for
pack
that
has
a
lot
of
stuff
based
on
how
it's
trending
a
lot
of
it
is
probably
not
gonna
go
in
and
just
to
kind
of
give
the
call
out
that,
as
we
get
closer
to
the
anticipated
release
date,
a
lot
of
the
things
that
aren't
completed
are
simply
just
going
to
drop
into
the
next
milestone.
E
A
So
I
think
we've
brought
it
or
the
discussions
that
we've
had
around
this
have
been
to
have
an
escalating
model
where
the
conversations
could
be
had
in
independent
teams
where
appropriate
right.
So
that
means
within
heroku
or
within
vmware.
Given
that
there
is
are
no,
you
know
broader
implications,
then
from
there.
It
would
be
these
meetings
that
we
have
here
and
then,
if
again
we
need
a
broader
audience.
Then
it
would
be
working
group
meetings
and
then
from
working
group
meetings.
A
E
B
Yeah,
I
actually,
I
honestly
have
no
preference
really
just
whatever
is
the
best
from
a
maintenance
perspective
right.
What
I
do
see
is
that
for
some
reason,
these
things
just
flake
every
once
in
a
while,
and
you
know
whether
the
containers
don't
have
outbound
internet
anymore
or
whether
you
know
configuration
has
been
blown
away
right.
So
you
know
honestly,
I
guess
I
was
just
thinking
about
fixing
these
particulars.
B
A
A
Maybe
there
is.
You
know
the
separate
conversation
of
trying
to
understand
the
reasons
why
these
things
are
flaking
out
right,
and
I
think
we
could
also
talk
about
those
at
length
given
a
very
specific
situation
or
problem.
So
if
we
talk
about
a
losing
network,
I
didn't
do
any.
You
know
research
into
that.
I
think
anthony
you
and
andrew
might
have
do
you
have
any
more
insight
into
what
happened
there.
B
Yeah,
we
have
no
idea,
you
know.
What's
actually
funny
is
that
it's
also
happening
on
the
alkyl
salad,
it's
just
only
when
you
switch
it
to
windows
containers,
so
we
briefly
switched
it
to
the
windows
containers,
one
on
the
daemon
and
we
were
seeing
the
same
behavior,
but
we
switch
it
back
to
linux.
So
you
know
I
feel
like
there
has
to
be
some
something
on
the
line
there,
but
I
mean
as
of
right
now.
I've
only
seen
it
once,
but
you
know
these
sort
of
one-off
issues
keep
showing
up.
A
It
would
be
interesting-
I
don't
know
if,
like
one
off
would
really
warrant
this.
But
if
we're
talking
about
network-
and
it
seems
like
it
might
have
happened-
and
maybe
you
were
able
to
reproduce
it
with
alcohol
as
well,
if
we
could
create
an
issue
for
it
right
and
that
way,
we
could
at
least
continue
like.
We
create
an
issue
address
it,
even
if
we
have
to
do
the
quick
fix
of
killing
it
and
starting
another
one
right
and
then
if
it
happens
again,
keep
adding
to
that
documentation.
A
B
A
D
B
D
Now,
just
to
play
devil's
advocate
and
we
chat
about
this
separately,
so
github
actions,
the
pack
test
should
all
pass
in
a
default
out
of
the
box.
Github
actions
runner
for
windows.
What
that
doesn't
give
us
is
ability
to
drop
in
and
debug
anything.
That's
failing
and
that's
a
pretty
crucial
feature
that
keeps
me
from
wanting
to
go
all
in
on
that,
but
there's
probably
eventually
a
cost
trade-off
value
of
you
know
what
does
it
cost
us
to
maintain
this
packet
worker
and
what
is
it
you
know?
A
D
I
think
we
still
could
work
around
it.
I
think
we
would
still
be
doing
the
same
thing
that
we're
doing
on
the
worker,
but
we
would
include
it
into
the
the
ci
scripts
of
modifying
the
insecure
registries.
Every
time
we
ran
that
thing,
not
a
not.
I
don't
want
to
downplay
that
there's
some
work
to
still
figure
out
there
potentially,
but
if
we're
you
know,
won't
have
any
parallel
options:
that's
not
all
packet
or
not
all
packer.
A
Yeah,
I
think
it's
definitely
worth
keeping
on
the
radar,
because
there
is,
you
know
to
the
point
of
having
to
maintain
the
windows
workers
in
a
healthy
state
that
is
totally
lifted
off
our
shoulders.
If
we
go
with
the
github
provided
route,
which
is
the
original-
and
you
know
reason
why
we
did
that
in
the
intention
there,
so
it
was
unfortunate
to
have
to
move
away
from
that
based
on
the
limitations.
A
D
Right
yeah,
I
was
thinking
like
if
it's
very
easy
for
it
to
have
to
go
and
spin
up
their
own
sort
of
stateful
vm
to
debug
an
issue.
Then
you
get
a
little
bit
of
that
debugging
functionality
that
you
lose
by
not
having
a
long
lived
machine,
but
I
do
want
to
go
and
validate
that
the
pack
test
can
still
run
on
a
default.
Github
actions
runner.
We
learned
a
lot
since
we
tried
that
explorer
too.
So
that
still
might
not
be
true
I'll
double
check
that
in
parallel.
B
A
D
D
A
D
Probably
yeah,
I
think,
where
the
best
source
of
that
data
is
the
I
mean.
The
funny
thing
is
that
pack
and
tools
like
it
are
really
going
to
be
the
motivator
to
get
people
to
use
some
of
these
tools,
and
it's
a
little
bit
of
prediction:
there's
not
a
ton
of
people,
probably
using
wsl
for
docker
at
all
right
now
today,
but
once
you
know
we're
we're
writing
the
wave,
that's
going
to
get
them
to
want
to
do
it,
but
they
will
still
be
the
part
that
is
isn't.
D
Prediction
is
the
versions
of
windows
that
they're
stuck
on.
We
can
get
more
hard
data
that
shows
that
you
know.
90
percent
of
enterprise
customers
are
using
windows,
10
enterprise,
but
I
don't
think
it
would
tell
us
too
much
about
wsl
users.
I
think
we
would
the
that
that
part
of
the
puzzle
is
still.
A
D
C
A
A
C
A
Not
ordered
right,
I
think
one
of
the
intentions
of
working
in
this
project
as
a
whole
was
to
move
away
from
any
sort
of
ordered
priorities
and
milestones
were
the
best
thing
that
kind
of
fits
right
in
the
middle
between
saying
we
should
have
focus
on
this,
but
then
at
the
same
time
say
there
is
no
order
within
this
focus
right.
So
I
think
it's
really
more
or
less
a
matter
of
importance
to
the
individual,
but
there
is
at
least
a
scope
versus
the
entire
project
as
a
whole.