►
From YouTube: Platform Sync: 2021-03-17
Description
Meeting notes: https://bit.ly/38pal2Z
A
C
Updates
can
kick
things
off.
There
is
a
release
candidate,
oh.
C
C
I
think
we
just
got
to
updates
so
there's
a
release.
Candidate
of
pack
that
went
out
on
monday
went
out
a
little
bit
late.
I
apologize.
I
was
traveling
on
friday
and
there
were
some
pipeline
issues.
C
C
B
B
There
was
some
some
good
feedback
on
potentially
different
names
for
it,
but
I
think
it
all
kind
of
lent
to
the
the
fact
that
we
should
probably
gather
some
more
feedback
on
the
future
as
it
is,
and
maybe
look
for
ways
to
fix
the
name
by
not
changing
the
name,
but
maybe
by
changing
what
it
does
to
make
it
fit
better
with
what
users
expect
it
to
do
so
yeah
separately
I'll,
go
and
lay
out
the
scenarios
and
see
if
we
can
tweak
the
behavior
to
fit
those
scenarios.
B
Apart
from
that,
I'm
just
trying
to
get
a
change
into
life
cycle
that'll,
hopefully
make
it
into
that
release
which
will
bubble
up
to
pax
release.
It's
a
fix
for
cache
images,
so
it's
slightly
user-facing
for
pac
users,
but
yeah
shouldn't
be
a
should
be
a
big
change.
C
A
A
A
C
A
I
guess
just
step
from
core
team
is
we're
finalizing
stuff
on
roadmap
and
I
think
it
should
be
coming
out
today
if
it
hasn't
already.
Hopefully
we'll
see.
Okay,
but
there's
a
pr
in
the
community
repo
that
jacob.
C
A
That
has
at
least
the
detail,
bull
points
that
is
not
the
nice
fancy
wording
stuff
that
we
tell
the
public
who
don't
go
to
our
github
repos.
C
Oh
oops,
all
right,
I
guess
oops
we
already
kind
of
talked
about
release
planning-
that's
previous
section,
so
I
guess
jump
on
to
the
needs,
discussion,
stuff
and
I'm
happy
to
share
my
screen.
A
C
Okay,
so
only
two
two
small
things:
it
looks
like
moving
the
analyze
phase
before
detect.
I
think
there's
like
a
huge
pr
on
the
life
cycle
that
needs
to
land
before
we
like
think
about
what
we're
gonna
do
here
or
start
making
any
changes.
It
still
looks
like
it's
going
through.
A
C
B
I
would
sorry
just
throw
in
here
related
to
this
change.
I
think
there
will
be
some
changes
that
have
to
move
around
in
pack
for
windows,
there's
a
prep
step
that
goes
and
runs
before
detect
that
ensures
that
the
volume
permissions
are
correct
and
I
suspect
that
I'll
have
to
move
to
running
before
analyzer
or
analyze.
I'm
sure
the
test
will
fail
and
we'll
will
show
us
how
that
goes.
But
I
guess
we'll
have
to
be
clever
about
how
we
deal
with
that
for
different
api
versions.
C
C
All
right-
and
I
guess
I'll
add
a
note
just
on
this
issue-
just
to
kind
of
like
capture
that
so
that
come
back
and.
C
C
All
right:
okay,
no
more
discussion
stuff!
It
looks
like
I
accidentally
just
ticked
every
single
box
in
our
notes,
but.
C
A
I
don't
know
if
there's
a
ton
to
discuss
besides,
probably
would
just
be
good
to
get.
A
People's
input.
A
C
All
right,
next
up,
we
have
this
package
refactoring,
which
I
think
is
also
kind
of
at
the
same
place.
C
I
think
it'd
be
good
to
get
an
extra
pair
of
eyes
on
this
stuff,
but
it
doesn't
really
outline
anything
too
dramatic.
Sorry.
A
C
C
Yeah,
I
guess
so
one
of
the
questions
I
have
is
a
little
bit
about
process.
Will
this?
Do
we
want
to
wait
for
this
to
be
approved
by
the
maintainers
and
the
core
team
members
in
this.
C
C
A
A
I
don't
know
if
there
has
really
been
or
fashion
rvc
in
the
past,
mostly
because
one
the
project
is
only
a
few
years
old.
So
I
think
this
is
the
first
attempt
at
dealing
with
the
fact
that
decisions
were
contacting.
Decisions
were
different
back
when
the
posture
was
created.
C
Yeah,
I
think
that
that
was
trying
to
shift
the
focus
away
from
like
what
things
will
actually
look
like
at
the
end
and
kind
of
more
be
like
hey.
These
are
some
really
generic
things.
We're
gonna
have
like
just
this
like
pkg.
C
This
is
like
standard
go
and
some
of
the
asks
that
we've
had
so
far
about
functionality
that
should
be
consumable
by
other
people.
C
So
I
think
trying
to
capture
those
two
points,
and
then
I
think
that
the
big
thing
is
just
trying
to
outline
like,
oh
in
the
future,
when
we're
trying
to
do
this.
This
are
probably
just
some
like
basic
steps
that
we
should
take
so
that
we
don't
have
to
write
a
new
refactoring
rfc
every
time
people
want
to
make
a
refactor,
because
yeah
totally
agree
that
agreeing
in
this
meeting
before
anyone's
actually
done
the
refactor
is
probably
not
the
way
to
get
to
codebase.
C
A
A
Like
I,
I
don't
think
like
every
single
refactor
goes
through
rfc,
but
I
think
kind
of,
like
you,
said
right.
It's
like
a
lot
of
high
level
kind
of
things
and
not
like
this
is
how
the
code
should
look,
or
these
are
the
exact
like
it
says
as
one
the
non-goals
right
like
these
are
the
exact
packages
that
need
to
exist.
C
C
Yeah
yeah
yeah,
I
agree,
but
yeah
totally
open
to
any
any
any
other
feedback,
but
I
guess
I'll
just
wait
for
people
to
feel
comfortable
about
on
this.
I
think,
probably,
unless
there's
anything
else
to
add
move
on.
C
Adding
builder
key
and
project
descriptor,
I
think
this
is
gonna
come
up,
it's
also
in
fcp,
and
then
we
just
have
application
accents.
Okay,
all
right!
Sorry,
definitely
not
as
fast
as
javier
going
through
this
stuff.
It
looks
like
we
have
a
couple
other
things:
the
image,
util
os
architecture,
aware
constructor
options.
B
B
B
So
now
you
would
explicitly
say
which
os
you
want
to
create
the
the
image
for
there's
this
slightly
different
edge
case
that
this
platform
gets
used
for,
specifically
in
the
the
small
case,
where
you're
actually
using
from
base
image,
you
can
give
it
a
preference
or
from
base
image
with
the
manifest
list
you
can
give
it
a
preference.
It'll
tend
to
prefer
whatever
platform
you're
asking
for
if
it's
ambiguous,
but
the
primary
case
is
really
to
for
your
brand
new
image
to
have
an
os
and
architecture
ahead
of
time.
B
B
So
even
if
it
does,
you
do
bump
to
the
new
version
of
image,
you
tell
you
won't
change
any
of
the
behavior
until
you
start
to
use
these
separate
options,
so
it
can
gradually
be
added
over
time
and
also
there's
places
in
pack
that
aren't
creating
images
this
way
through
mgtow
those
might
be
candidates
for
either
adding
or
porting
that
this
is
particular
for
build
packages
created
on
disk,
maybe
that'll,
you
know,
be
a
good,
forcing
function
to
move
that
functionality
into
image
util
or
come
up
with
some
other
holistic
way
to
solve
it.
B
But
it's
really
try
to
be
very
conservative
about
how
we
introduced
this
functionality,
so
you
weren't
forced
to
change
the
way
that
you're
currently
making
images
there's
more.
This
is
the
closed
for
the
merged
pr
in
image
detail,
but
there
will
be
some
examples
that
will
come
up
using
this
inside
of
lifecycle
first
but
yeah
like
I
said,
if
you,
if
you
bump
to
the
latest
image,
you
tell
you'll,
be
able
to
do
this
in
pack,
but
you
don't
have
to.
B
Yeah
one
thing
that
was
a
little
unclear
was
whether
we
should
actually
make
this.
Have
the
exact
same
option
do
the
same
thing
for
manifest
list
disambiguation
and
for
new
image
creation.
Potentially,
we
could
have
added
an
option
to
the
from
based
image
there.
So
if
it
feels
weird,
then
it
feels
like
that
should
change
that'd,
be
good
feedback,
potentially
tweak
how
it
works.
C
Oh
okay,
I
guess
the
next
thing
is
the
registry
helper
test
changes.
B
I
guess
I
got
all
the
all
the
topics
today.
This
also
came
in
with
similar
sort
of
alongside
those
changes
in
image.
A
B
But
not
specifically
for
the
same
purpose.
I
can
share
screen
again.
B
This
one
will
be
pretty
short
as
well.
The
so
current
state
of
affairs,
image,
util
and
lifecycle,
use
the
same
registry
helper
and
that
registry
helper
is
the
way
to
create
a
one-off
test
registry
for
tests,
usually
for
acceptance
tests.
I
think
you
can
write
a
bunch
of
stuff
to
it
and
then
it
goes
away
when
the
test
stops
running.
B
But
it's
and
it
does
that
trick
by
running
the
container
and
having
any
test
containers,
use,
dash
dash
network
host,
so
everything
kind
of
standardized
on
localhost
being
the
same
thing,
even
though
it's
not
this,
that's
all
well
and
good,
but
doesn't
work
very
well
on
linux,
around
windows
and,
interestingly,
we
saw
that
a
new
version
of
docker
desktop
had
that
broken
as
well.
They'll
probably
fix
it,
but
yeah
mac
was
was
broken
with
that
functionality.
B
So
the
hope
here
is
to
make
it
easier
to
not
have
everything
need
to
use
network
host
for
the
same
thing,
to
make
it
easy
to
stand
up
a
test
registry
container
and
then
inform
the
test.
That's
running
where
to
look
for
that.
B
The
way
that
it's
going
to
do
that,
at
least
in
this
current
implementation,
is
by
any
by
asking
the
daemon
that
it's
running
on,
if
it
has
any
insecure
registries
entries
and
specifically
ones
with
the
slash
32
feels
a
little
contrived.
But
unfortunately,
you
do
if
you
already
use
this
test
registry
this
way
you
need
to
have
it
as
an
insecure
registry.
B
Somehow
you
could
have
wild
card
entry,
allow
money
and
secure
registry
if
you
wanted,
but
it
felt
like
the
single
source
of
truth
is
to
have
a
single
entry
in
there
that
slash
32..
It's
very
secure,
you're
not
going
to
have
someone
create
another
registry
on
that
exact,
same
entry
like
maliciously
like
up
in
a
ci
instance,
or
something
like
that,
so
you
won't
inadvertently
trust
too
many
things
and
then,
at
the
same
time,
it
also
tells
the
test
where
to
find
that
registry
in
a
more
public
way
than
just
localhost.
B
I
think
that's,
probably
the
most
significant
part
of
the
change
there
there's
a
few
other
tweaks
that
came
out
of
this.
But
again
this
is
just
the
registry,
the
registry
helper,
that's
used
in
lifecycle
and
in
image
util.
So
if
pac
did
want
to
start
using
it
or
wanted
to
have
similar
functionality
technically
it
already
has
similar
functionality.
It's
just
slightly
different.
It
does
allow
the
windows
acceptance
test
to
run
against
a
registry,
that's
not
effectively
on
localhost.
B
You
could
potentially
adopt
this
registry
implementation
or
change
the
one
in
pack
to
be
similar,
this
one's
a
lot
more
hand-wavy
than
the
than
the
other
topic.
So
if
any
of
you
wanted
to,
if
you
want
me
to
explain
anything
more,
I'm
happy
to
do
that
also
no
need
to
adopt
it
too.
C
I
think
that
the
mac
this
functionality
is
still
busted
on
macs,
unfortunately,
so
would
be
nice
to
switch
off
of
it.
B
Yeah
this
seems:
I've
been
using
the
latest
version
of
docker
desktop
and
running
all
my
life
cycle
and
image
util
tests
locally.
Using
this
feature,
so
it
does
seem
to
work
around
it
for
the
time.
B
B
No,
that's
fine
yeah!
I
think
that's
about
it.
For
that
one,
and
the
last
topic
I
had
on
there
is
I'm
moot
at
this
point
too,
because
I
think
we
I
closed
the
pr
and
we
can
discuss
the
a
better
name
offline.
I
think.