►
From YouTube: Platform Sync: 2021-05-19
Description
Meeting notes: https://bit.ly/38pal2Z
A
Let's
record:
hey:
what's
good
welcome
to
platform?
Sync,
I'm
doing
it,
I'm
doing
this
we'll
do
it
my
way
all
right!
First
of
all,
I
just
want
y'all
to
say
hi.
Let's
do
some
status
updates.
Okay,
me
first,
I
know
I
have
a
lot
of
things
in
flight.
I
have
my
rfc,
which
I'm
rewriting
it's
in
draft
mode,
but
if
y'all
could
take
a
look,
it's
pretty
digestible,
it's
really
just
the
rfc,
but
it's
not
really
interactive.
It's
just
you
know.
A
Pack
builds,
shows
a
bunch
of
funny
terminal
things
and
then,
and
then
it
exits-
that's
all
it
is,
as
we
discussed
the
last
time
other
than
that
you
know
just
some
little
stories
here
there.
I
know
javier
responded
to
my
life
cycle
issues
and
that's
about
it.
Maybe
we
can
get
into
the
details
later
if
it's
important
cool
next
on
my
screen
is
a
big
t.
Terrace.
B
I
don't
have
too
many
updates,
especially
not
really
the
platform,
but
I
did
start
the
bill
pack.
B
Authors,
sub
team
rfc
so
excited
to
kind
of
see
that
going-
and
I
think
there'll
be
definitely
some
overlap
with
this
team
and
pack,
especially
with
the
create
command
and
figuring
out
kind
of
how
to
divide
responsibility
or
contributions
into
pack.
C
All
right,
let's
see
what's
relevant
to
this
meeting,
been
working
on
creating
similar
to
like
image.
Util
supports
the
daemon
image
and
a
remote
image.
I've
been
creating
an
interface
implementation
for
oci
layout
right,
and
the
purpose
of
this
is
to
enable
the
build
kit
proposal
that
we
have
where
the
lifecycle
and
build
kit
will
interact
using
oci
layout.
C
So,
like
file
system,
oci
images,
I
did
find
that
we're
gonna
need
the
run
image
to
also
be
exported
out
by
build
kit
and
consumed
by
the
life
cycle
and
oca
layout
format.
So
there's
that
yeah,
I
think
that's
pretty
much.
It
I'll
probably
be
focused
on
that
for
quite
a
bit
to
get
it
off
the
ground.
A
How's
it
good
daniel.
D
E
E
Oh
I
guess.
Actually
we
don't.
We
support
everything
after
platform,
api
o2,
I
believe
which
seems
like
something
we
want
to
keep
doing.
That's
not
too
difficult,
and
I
guess
other
than
that
I've
been
working
on
trying
to
enable
pac
to
work
with
some
of
these
enterprise
configurations
where
they
have
a
proxy
registry,
so
that
they're
not
hitting
docker
hub
thousands
and
thousands
of
times
that's
going
pretty
well.
C
I
did
have
a
question
about
that
and
I'm
not
sure
if
we
need
to
go
like
too
much
into
depth,
but
aren't
there
certain
images
that
life
cycle
itself
doesn't
look
up
where
it
wouldn't
do
the
transformation
for
the
proxy
and
may
this
might
be
really.
You
know
specific
to
your
implementation
of
it.
But
what
I
was
thinking
about
is
it
does
like
a
run
image
look
up
based
on
the
run
image
mirrors
available
and
it
kind
of
picks
a
run
image
on
its
own.
C
D
E
There's
like
it
will
be
a
little
awkward
right,
because,
ideally
most
of
these
things
are
like.
What's
these
registry
or
proxy
registries,
you
can't
push
to
right
they're,
just
for
like
pull
through
purposes.
E
So,
if
you're
gonna
have
to
pull
through
the
run
image
and
then
re-upload
it
to
a
new
registry,
you
wanted
to
go
through
the
proxy,
but
if
you're
uploading
to
something
like
docker
hub
or
gtcr,
that
already
has
a
run
image
mirror
you
don't
want
to
change
the
name
at
all,
because
you
just
want
the
registry
to
be
able
to
do
a
cross
repo
blob
mount,
but
yeah
yeah
I'll
definitely
have
to
do
something
for
the
the
like
streaming
down.
In
the
case
when
we
don't
have
a
run
image
mirror.
A
E
A
Okay,
quick
question,
so
I
know
I'm
working
on
a
story
to
bump
the
life
cycle.
Is
that
something
we
want
to
squeeze
into
the
next
release?
Or
we
don't
care.
D
A
A
A
Sorry
give
me
a
second
just
trying
to
you
know,
make
things
safer,
safer
work
here,
all
right.
We
have
oh
interesting.
So
how
does
oh
I
see
so
things
would
have
to
be
labeled
platform,
so
we
have
this
interaction
mode
again,
like
we
said
last
time.
I
I
want
this
to
be
not
looked
at,
I
guess
or
be
disregarded.
A
There
is
sort
of
another
draft
in
its
place
that
doesn't
have
the
label.
I
don't
know
if
someone
can
put
the
label
on
it.
For
me,
that'd
be
that'd,
be
appreciated,
but
that's
supposed
to
be
the
replacement
for
this
one
sounds
good.
C
And
probably
the
subteam
rfc
label
as
well.
D
D
D
A
Okay,
hopefully
that's
going
on
last
one.
The
application
mix
is
still
in
draft.
B
Of
the
loop
on
this
one,
sorry,
I
I
feel
like
this
is
going
to
be
in
draft,
probably
until
stack
packs
gets
much
further
along
and
then
it
affects
kind
of
the
platform
team
once
the
life
cycle
stuff
kind
of
goes
through
is
the
impression
I
get.
I
don't
think
joe
as
far
as
I
know,
joe's
not
actively
working
on
this
rc,
which
is
why
it's
in
draft.
A
Sorry
can
we
get
like
cliff
notes
on
what
this
is
supposed
to
do?
Application
developers
can
specify
mixins,
so
the
build
pack
is
going
to
val.
You
know
verify
that
the
mixins
of
the
app
as
well.
C
If
I'm
not
mistaken,
the
way
I
understood
this
is
based
on
quote-unquote,
make
sense
that
you
could
add
during
the
build
process
it
would
actually
install
stuff
for
you
right
so
like
that
example
right
there
with
ffmpeg.
A
A
C
Yeah
but
so
in
go
for
instance,
right,
like
your
software
dependencies
are
for
go
itself
like
these
are
os
level
dependencies.
D
C
Right
so
like
ffmpeg,
it's
not
something
that
you
could
declare
at
the
application
level
you
declared
at
the
os
level.
Sorry
turns,
I
think
I
interrupted
you.
B
Yeah,
I
was
just
gonna,
probably
repeat
what
you
said:
yeah
I
think
you
alluded
to
this,
but
it
does
require,
for
this
to
function,
does
require
a
basically
stack,
build
pack
that
has
root
level
access
to
process
this
mixin.
B
So
if
your
builder
does
not
have
that-
and
you
put
this
in-
it's
not
going
to
magically
install
fmpeg
in
that
example,
out
of
nowhere
so
isn't
just.
It
is
a
thing
like
if
you
are,
for
instance,
like
a
healthcare
company
right
and
you
you
want
to
basically
lock
down
and
don't
want
your
application
devs
to
be
able
to
install
whatever
packages
like
it
is
a
thing
as
a
a
platform.
B
You
know
that
that
is
managing
that
stack
kind
of
has
control
over
that.
So
that's
why
this
is
kind
of
dependent
on
all
the
kind
of
implementation
life
cycle.
Changes
to
support
stack
packs
to
actually
have
stack
packs
that
can
process
this,
and
this
would
be
kind
of
the
I
guess
platform
side
of
supporting
the
ability
to
for
an
application.
B
End
user
to
kind
of
tie
into
a
stack
pack
like
you
can
have
stack
packs
that
do
root
level
things
that
don't
require
mix-ins.
But
I
think
nixon
was
probably
one
of
the
big
use
cases
for
stack
packs.
A
Okay,
I
completely
dig
that
that
was
great.
That
was
the
missing
piece
for
me.
I
100,
I
will
just
say,
as
an
outsider,
newest
outsider.
Looking
in,
I
think,
mix-ins.
Originally
the
stack
mixes,
the
regular
mix-ins
were
probably
the
most
confusing
part
and
some
other
thing
called
mix-ins.
Is
it's
not
going
to
help?
Maybe
the
learning
team
can
help
us
with
the
docs
on
that
all
right
all
right.
A
E
So
I
think
the
easiest
thing
for
me
to
do
is
say
we're
going
to
support
everything
until
we
start
needing
to
deprecate
stuff.
There
was
like
an
idea
thrown
out
last
time
about
doing
like
an
n,
minus
1
or
n
minus
2.
D
C
So
yeah
I
mean
I
totally
agree
with
that.
I
I
do
wonder
like
this
specifically
is
calling
out
platform
apis,
but
the
one
that
I
was
really
more
concerned
about
were
other
apis,
specifically
project
descriptor,
right,
which
I
think
is
like
you
know
way
worse
state
right
now.
C
Other
apis
there
are,
but
like
build
builder,
is
going
to
be
an
api
as
well
right
and
so
like
again,
I
envision
somewhere
right
like
we
would
have
like
all
the
apis
and
then
the
listed
versions
of
their
support.
C
B
D
C
Guess
I
wonder
dan,
I
know
that
this
is
something
I
guess
on
your
plate,
but
if
there's
anything
that
we
can
do
to
to
better
like
track
all
this
stuff
right
like
it
doesn't
have
to
be.
You
like,
maybe
just
formulate
a
place
where
we
could
do
this,
and
then
we
could
all
collaborate
to
putting
all
this
information
in
there.
E
Yeah,
okay,
that
that
sounds
good.
I
guess
like
I
don't
think
we
have
do.
We
have
builder,
we,
we
don't
have
builder
ap.
We
have
some
stuff
that
isn't
versioned
right
now
right
and
I
feel
like
figuring
out
how
to
version
that
stuff
is
like.
D
E
C
Yeah,
so
I
think
that's
why
we,
the
platform
api,
was
a
really
good
first
step,
and
you
know
thank
you
for
going
through
that
effort.
But,
like
you
know,
making
that
statement
was
like
the
first
one,
because
that
one's
easy,
that
one
is
already
project
descriptor,
I
think,
would
be
the
second
one,
because
we
do
have
an
o1
version
and
we
have
the
o2
version.
That's
about
to
be
released
and
then
builder
is
like
kind
of
looking
into
the
future.
B
C
C
Release
notes
does
sound
like
a
pretty
good
place
right,
but
there
might
be
others
as
well,
but
I
definitely
don't
think
this
is
like
a
end
user,
end
user
or
like
an
app
developer
kind
of
thing,
but
then
again
it
might
be
depending
on,
like
the
builder
like
how
bad
they
they
land
into
this
point
in
time,
where
the
builder
that
they're,
using
isn't
supported
right,
like
pointing
them
to
release,
notes,
might
not
be
the
most
appropriate
place.
C
So
it's
almost
like
a
thought
experiment
at
this
point,
but
the
first
thing
again
is
like
just
putting
it
somewhere
right
and
then
from
there
we
could
kind
of
figure
out
exactly
where
the
best
place
may
be.
A
I'm
just
gonna
put
this
in
here:
just
want
to
be
clear,
explicit
about
next
steps
on
this
asset
packages.
Pr,
I
know
it's
sort
of
looking
ready
to
review
for
quite
a
while.
I
was
just
hoping-
maybe
we
could
say
some
yes
knows
maybe
knows.
As
far
as
when
the
review's
connect.
A
So
there's
dance
asset
cash's
pr
which
I
think
is
ready
to
review,
and
you
know
I
know
it's
a
big.
It's
a
big
pr,
but
I
was
wondering,
like
you
know,
as
opposed
to
just
letting
it
linger
there
like
what
what
do
we
think
next
step
should
be.
E
E
The
places
where
this
vr
requires
anything
that
to
be
done
to
the
spec
is
extremely
small
right
and,
in
fact,
because
all
this
would
be
experimental,
we
would
have
to
wait
for
like
a
life
cycle
release
anyways
to
like
use
things.
I
guess
the
way
that
they
should
probably
be
used
at
the
very
end
of
this,
but
like
that,
won't
change
any
of
what's.
C
C
I
guess
the
point
where
the
spec
matters
is
when
we
want
to
try
to
take
it
out
of
experimental
like
at
that
point.
The
spec
should
be
merged
in
and
then
it
kind
of
ties
into
this.
You
know
cataloging
of
what
apis
we
support
and
this
one
would
be
another
one
that
we
would
add
there
as
officially
supported.
C
C
Nope,
cool
yeah,
I
guess
dan,
do
you
mind
adding
that
sort
of
statement?
What
I
just
said
like,
because
this
is
experimental,
we
don't
want
to
wait
for
the
spec
changes
to
be
finalized
and
then
once
we
do
want
to
make
it
non-experimental.
That's
when
we'll
have
the
spec
changes
done.
E
I
think
there's
a
comment
in
there
that
kind
of
outlines
like
what
all
the
compatibility
is
and
where
things
are
right
now
and
how
we
would
move
things
forward
once
the
specification
is
merged
and
how
we
could
like.
Okay,
it's
kind
of
like
a
more
smooth
transition.
D
D
A
Okay,
last
one
I
just
wanted
to
talk
about
this
rsu
really
quickly.
I'm
sure
nobody's
had
a
chance
to
look
at
it,
yet
it
was
completely
fine
right.
I
just
wanted
you
as
always
to
just
hear
it
in
person
right
before
you
re
read
another
another
document
on
the
internet.
I
just
wanted
to
show
like
at
a
high
level
what
it
is
and
I'm
not
expecting
any
statements.
A
Just
like
you
know,
red
flags
or
something
or
something
you
want
is
that
okay,
I'm
gonna
share
my
screen,
so
I
mean,
for
the
most
part,
it's
pretty
similar
right.
That's
why
I
wanted
to
do
this.
I
just
wanted
to
show
the
differences
for
the
most
part.
So
let's
say
you
on
the
command
line.
You
do
build
with
your
image
name
and
interact
dash
eye
or
whatever.
A
A
All
this
information
gets
filled
out.
I
get
my
little
check
marks
and
then
I
control
c
out
of
it.
That
is
that
is
the
bulk
of
this
rc
right.
Maybe
other
things
can
come
forward,
but,
like
that's
all
we're
going
for
here,
so
this
there's
four
things
one
is
when
I'm
sorry
one
is
when
I'm
I'll
put
this
in
rc2
right.
One
is
when
I'm
pulling
right
when
I
show
these
cool
pulling
things
which
reminiscent
of
docker
another
is
when
I'm
detecting,
which
you
know
detecting.
A
Presumably
this
stuff
would
stop,
and
you
wouldn't
see
any
of
this,
but
it
would
just
say:
hey,
I'm
detecting.
So,
oh
as
a
user,
I
know
something's
happening
for
me
and
then
there's
when
I'm
building.
This
is
what
it
looks
like
when
I'm
building
the
logs
are
going.
This
is
what
it
looks
like
with
the
image
name
I
put
in
and
then,
when
it's
succeeded,
I
have
a
little
time
stamp
when
my
image
was
built,
which
actually
I
don't
even
know.
A
But
that
this
is
the
gist
and
I
don't
know
again
not
expecting
any
comments
right
now,
just
chew
on
it,
and
you
know
next
time
you
look
at
the
rc
or
something.
E
E
A
Right,
this
is
the
smallest
version
I
can
think
of
of
getting
the
the
user
workflow
in
mind.
D
E
Maybe
I'm
missing
something
on
this,
but
it
seemed
to
me
like
when
we
talked
about
this
last
time
that
the
consensus
was.
We
definitely
want
to
do
this.
It
seems
like
we're
going
to
need
we're
at
the
point
now
where,
in
order
to
start
implementing
something
that
does
this,
we
need
to
have
an
rfc
to
get
like
just
the
base
level
of
like
we
can
pause
in
between
these
steps.
A
No
you're
not
wrong.
That's
my
understanding
too
to
me,
but
you
know
I
thought
that
was
going
to
be
a
follow-up
rfc
right.
If
this
thing
gets
through
makes
it
in
right,
then
we
can
add
the
hey,
I'm
pausing,
I
you
know
I
get
to
pause
and
as
a
user,
I
get
to
click
the
build
button
over
and
over
and
over
again
right.
That
is
to
me
easy.
A
D
A
D
C
Yeah,
I
do
just
real
quick.
This
is
definitely
probably
a
stretch,
but
I
just
wanted
to
kind
of
maybe
throw
it
out
there
for
some
some
thinking,
but
has
there
been
any
thought
I
know
dan?
You
had
some
work
that
you
did
previously
with
dive
kind
of
being
interactive,
inspect
utility
and
then
we're
gonna.
Have
this
other
ui
like
how
are
those
two
or
how
are
these
two
things
going
to
like
marry?
C
E
So
I
can
give
some
background
just
on,
so
I
am
so.
I
worked
like
quite
a
bit
on
this
like
dive
thing
and
I'm
not
super
confident
that
this
is
going
to
be
pulled
in
by
the
maintainer
just
since
it
basic
like
basically
most
of
the
pr
is
adding
a
windowing
system
right
to
this
tool,
so
you're,
not
just
picking
squares
on
the
screen
and
printing
to
it.
E
It's
adding
like
a
communication
component
so
that
different
things
can
have
this
process
where
they
decide
how
much
of
the
screen
real
estate
they're
going
to
take
up
where
they're
going
to
be,
and
it
involves
a
big
rewrite,
and
I
don't
think
this
person
is
super
interested
in
actually
reviewing
it.
Unfortunately,
just
since
it
changes
their
code
based
launch.
C
Cool,
so
then,
is
it
safe
enough
to
say
that
we
should
look
to
implementing
our
own,
essentially,
and
especially
with
maybe
some
of
the
stuff
that
anthony
does
right
and
just
use
that
as
the
quote
unquote
ui,
I
don't
know
what
are
you
using
anthony
for
the
the
ui
stuff,
I'm.
A
Actually,
using
the
same
library,
the
dive
is,
I
think,
it's
pretty
standard
for
golden
at
this
point,
but.
C
Cool
yeah
so
then
like
it
would
be
a
lot
of
that
right.
First,
as
as
far
as
the
visual
goes,
the
ui
interaction
that
could
all
be
using
the
same
thing
and
then
the
actual
bits
of
like
how
the
commands
are
done
and
what
things
we
could
do.
We
could
just
use
dive
as
an
inspiration,
but
like
re-implemented
ourselves.
E
I
mean
I
I
feel
like
we
could
do
this.
I
think
that
the
like
thing
that
anthony
rivo
tv
looks
like
the
thing
that
you're
advocating
and
that's
basically
what
I
did
the
rewrite
of
dive
in
right.
It's
like
cool,
so
I
think
there's
most
of
the
like
extra
windowing
functionality
we
would
need,
is
probably
already
written
since.