►
From YouTube: OCI Weekly Discussion 2020-02-05
Description
No description was provided for this meeting.
If this is YOUR meeting, an easy way to fix this is to add a description to your video, wherever mtngs.io found it (probably YouTube).
A
A
Hopefully
this
week
won't
be
the
same
so
yeah.
The
question
was
about
the
maintainer
trimming.
It
was
really
just
a
heads
up
since
OCI
kicked
off
and
whatever
it's
2015
and
then
some
of
them
hadn't
had
a
lot
of
activity
like
imaging
on
time.
Spec
haven't
had
a
lot
of
activity
since
2017
when
we
went
be
one
with
those
two
specs
where
it's
that
I,
like
I,
like
a
lot
about
the
governance
of
how
we
set
out
so
see,
I'm,
it's
it's
opinionated
and
fair.
A
It's
got
a
few
things
that
had
made
some
people
grumble
and
it's
still
kind
of
awkward
of
like
no
more
it's
it's
it
favors
having
like
whoever
is
the
person
that
has
those
responsibilities.
It's
you
personally
bear
it's
not
just
the
company
you
work
for,
but
at
the
same
time
it
also
pays
heed
to
not
letting
certain
companies
just
overpower
the
conversation.
So,
like
various
roles,
you
can't
have
like
too
many
seats
held
by
people
who
work
for
the
same
companies.
A
A
You
gotta
stop
requiring
the
maintainer
seriously
yeah,
but
the
the
biggest
thing
has
been
that
there's
a
few
contexts
in
which,
particularly
like
voting
for
the
releases
of
specs,
it's
in
the
the
docks
that
it's
not
just
a
majority
has
to
LG
TM,
which,
when
we
did
that,
when
I
put
up
for
the
runtime
spec
10.2
the
other
week,
we
had
four
out
of
eight
LG
teams,
but
we
her
the
governance
or
whatever
for
the
spec
versions
had
to
have
a
two-thirds
quorum
and
so
for
people.
Just
in
the
last
two
years.
A
Don't
are
not
active
at
all
they're,
not
checking
anything
about
OCI
in
their
inbox
either
because
they're
not
working
in
it
anymore
or
change
companies
or
whatever
the
email.
In
some
situation
the
past
the
email
is
like
balanced,
because
people
no
longer
worked
at
the
company.
Their
email
was
with,
so
that
was
more
or
less
just
administrative
stuff.
That
I
was
airing
in
the
open
so
that
there's
no.
A
Idea,
no
indication
that
it's
done
behind
scenes
are
managed.
You
know
like
in
some
special
way
and
the
way
that
I
handled.
That
is
that,
after
immediately
after
that
conversation,
I
went
to
the
maintainer
x'
file
for
the
runtime,
spec
and
email
started,
a
private
email
between
the
people,
just
so
that
they
didn't
feel
like
have
to
feel
like.
A
This
is
a
public
of
like
public
archive
conversation
to
just
see
you
know,
because
there,
as
a
you
know
one-to-one,
is
this
something
you're
still
interested
or
able
to
do
or
whatnot,
and
from
that
a
couple
people
we
you
know,
were
able
to
be
like
extra
this
something
I
can't
do
anymore.
So
this
this
new
from
Google
and
Brandon
Phillips,
who
was
chorus
now
Red
Hat,
now
IBM,
both
of
them,
we
had
PRS
merged
to
remove
them.
So
the
runtime
spec
has
a
maintainer
base
of
six,
which
is
not
great
for
quorum,
quorum
pieces.
A
It
could
deadlock,
a
vote
or
whatever,
but
yeah.
At
least.
If
we
put
up
a
motion
for
a
vote
again
and
the
four
act
of
maintainer
x'
voted
again
that
that
that
things
wouldn't
be
Stahl,
dowels
or
deadlocked,
that's
all
I'm
sure
the
same
might
happen.
If
you
look
at
some
projects
like
go,
diet,
go
digest,
it's
a
software
and
not
spec
project,
but
most
of
the
people
in
that
maintainer
--zf
all
are
not
participating
regularly.
A
B
C
A
A
E
A
C
E
If
any
folks
who
we
they
use
the
Nvidia
hook
so
use
some
other
custom
hooks
can
play
around
with
the
PR
and
also
give
it.
Our
view
would
be
really
good,
because
yeah
I
mean
I
can
only
test
for
the
model.
I
can
say,
is
this
conform?
It
with
this
back
is
basically
the
only
thing
that
I
can
really
test,
because
I
don't
have
any
of
the
hardware
necessary
for
some
of
the
more
esoteric
hooks,
but
that
are
fairly
widely
used
nowadays,.
A
E
Yeah
I
mean
West
case
I
need
his
carry
any
basis,
but
yeah
I
mean
dong-su
was
the
reason
why
we
actually
managed
to
get
the
hook
stuff
over.
The
line
could
be
approached
me.
I
think
with
all
systems
go
and
was
like
well,
can
we
please
get
this
managed
and
I
was
like
alright
fine
with.
A
E
A
E
E
E
Yeah
I,
don't
know,
I
mean
long
time.
We
should
bump
them,
but
I
think
they're
going
for
actual
removal.
I
think
is
gonna,
be
fine.
It
we
never
think
about
much
longer
and
it's
probably
not
a
good
form
for
you
to
have
like
aspect
that
he
released
one
point:
oh
I'll
still
like
three
years
of
work
and
then
1.01
and
then
2.0
seems
a
little
bit
yeah.
E
I
just
want
to
mention
one
thing
with
regards
to
I'm:
sorry,
I'm,
sorry,
unless
wrestling
with
you
guys
for
the
maintain
and
shrinks
one
other
thing
that
is
is
a
possible
consideration.
Is
that
yeah
I
think
this
is
a
problem
we
should
push
possible
to
arm
like.
E
Put
this
office
candidates
so
thin
you
maintain
is
the
one
thing
is:
is
that
anything
that
we
should
fully
discuss
as
I
can
actual
like
change
to
Howie's?
E
Is
a
I,
don't
know
if
it's
how
to
change
about
the
highly
doing
some
reviews
I
think
one
thing
that
actually
is
holding
releases
or
uncia
very
annoying
to
do
now
suspects
sense
to
have
opposing
thinkers
in
print
all
the
ways
of
the
way
the
votes
are
meant
to
Seoul
particular
problem,
which
is
you
know,
you
know
we
make
sure
whole
is
it
sex,
but
for
actual
project
like
the
the
whole
voting's?
That
means
that
we
ended
up
delaying
releases
by
god
knows
how
long?
Luckily,
we've
never
had
a
you.
E
E
A
E
A
E
A
E
E
I
mean
I
think
it
was
nice,
but
to
have
like
a
standard
wave
in
relations
across
all
those
guys
projects
like
it,
and
we
all
it's
all
done
by
either
like
some
sort
of
well
as
honor
of
people
surgery
but
really
seen
but,
like
you
have,
you
know,
want
it
to
be
below
responsible
for
doing
releases.
Each
release
hold
on
has
to
be
less
like
a
couple
of
days,
Stu
period
or
something
but
yeah.
The
the
quorum
thing
is
in
the
Basilica,
for
instance.
E
Actually
right
now,
image
tools
has
been
broken
in
Debian
to
like
two
years
now,
because
because
there
was
a
a
soldier
stood
out
from
unisex
the
version,
number
Decree
decrease
and
so
they're
requesting
updates
to
like
100
off
c3
or
something
to
fix
this
problem
buzz
because
Parliament
votes,
it's
been
broken
and
dead.
Things
like
yeah,
so
I
think
that
it's,
if
you
just
clarify
that
it's
necessary,
then
I
mean
like
I,
can
put
the
release
right
now
finish.
E
E
Yet
the
child
lost
touched
in
2016,
but
yeah
I
get
your
point.
Its
own
yeah
I
think
I
will
I
will
double-check
the
achada
to
make
sure
that
it's
not
specified
in
the
Charter.
But
if
it's
also
sign
the
Charter,
then
we
can
put
in.
We
can
just
add
it
into
the
government's
the
governance
top
for
the
OCA
org
it
just
mentioned,
like
the
quorum
rules
are
only
for
the
specs
I
think
would
be
fiction.
I.
A
C
F
That
for
the
run
times
back,
we
we
have
this
exact
problem
right
and
the
way
Vincent
has
resolved.
It
is
by
reaching
out
to
said,
maintain,
errs
and
saying:
hey,
you're
inactive.
Can
you
please
step
down
and
they
made
the
PR
themselves
so
I
think
that's
why
we
worry
about
66%,
because
if
they're
saying
I
want
to
step
down,
who
are
we
to
stand
in
their
way
right
but
I?
E
E
F
A
And
the
only
other
thing
that
I
think
is
kind
of
comedic.
Sometimes
it's
sometimes
the
maintainer
czar,
the
ones
that
are
proposing
PR
is
not
always.
Obviously
it
is.
Things
are
very
active,
but
some
things
that
we
like
come
to
an
agreement
we
want
to
push
through
and
then
we
don't
have
an
effective,
maintain
or
even
to
get
LG
teams
in
a
timely
manner
that
one's
cute.
B
There
is
something
to
be
said:
if
you're
not
active,
then
we
should
have
something
written
because
I
mean
even
the
artifact
stuff
was
I
just
started.
It
and
I
was
having
a
hard
time
getting
responses
from
people
that
they
want
to
be
tape.
Then
we
didn't
get
responses,
so
general
I
loved
ciencia
for
Linux
Foundation
already
has
some
verbiage
that
helps
address
inactivity
but
seems
like
we
need
something.
B
E
Souter
democracy
but
yeah
the
other
option.
You
know
getting
all
these
things
on
PL
I
mean
I,
originally
pushed
for
requiring
good
to
say
for
disallowing,
LG
teams
and
I
think
on
balance.
That
was
actually
a
bad
idea,
because
I
think
that
if
some
maintainer
pushes
a
patch,
especially
for
such,
has
been
already
a
grab
some
sort
of
consensus
elsewhere,
it's
a
bit
annoying
just
a
way
for
people
to
go
to.
E
They
get
up
notifications
to
lgt
Anna's
like
it's
just
with,
and
nothing
mentioned,
I
think
Indiana's
can
push
to
Moscow
like
it's
like
it's
good,
I
wouldn't
say
it's
formality.
I
mean
the
point
of
it
is
to
get
review
and
press
to
it.
But
I
think
you
know,
if
maintained
has
already
pushed
a
change
understand
it
enough,
so
they
only
one
more
review
but
I
don't
open
that
kind
of
webs
again
or.
B
B
B
E
E
He
owes
a
lot
in
suspect,
I
mean,
like
you've,
said
up
to
it,
I'd
be
down
for
it,
because
I
think
that
the
problem
is
that
we
have
is
that
we
haven't
added
I,
mean
I,
think
said,
except
for
new
projects,
all
of
the
old
projects
haven't
added
maintain
is
in
like
three
or
four
years
from
memory.
Although
the
image
spec
added
me
two
years
ago,
I
think
and
that's
the
last
time
I.
E
Remember
that
my
data
being
added
and
there's
like
in
terms
of
new
blood,
there
hasn't
been
anyone
for
at
least
a
couple
of
years.
Who's,
a
maintainer,
so
I
think
that
another
problem
is
that
people
start
working
on
other
things
and
we
haven't
replaced
the
outflux
of
maintaining
is
with
new
ones
as
well
as
the
fact
we
have
maintained
us
who
are
in
the
Medina's
file,
but
have
basically
stopped
working
on
what
they
eat
to
maintain.
Yeah
I
think
it
I
think
it's
not
a
matter
of
getting.
E
E
Yeah
yeah
this
thing
I
just
want
to
mention,
but
yeah
I'm
gonna,
consider
proposal
to
achieve
put
in
those
yeah
I.
Think
you
just
about
it
and
given
that
it's
it's
widely
use
nowadays
and
yeah,
they
could
be
a
discussion
about
the
discussions
about
merging
image
tools
and
watching,
because
obviously,
the
verification
stuff
that
initial
does
is
quite
useful,
so
I
think
just
useful,
but
all
basically
necessary
from
what.
E
From
a
like
childhood
standpoint
standpoint,
so
yeah
I
think
I'll
post
a
video
help
us
PR
to
the
TOB,
but
you
know
I
think
the
only
the
only
thing
that
well
there
are
two
open
things.
One
of
them
is
maintained
is
again.
This
is
a
let's
say
this
is
a
cult
fraction.
If
you
would
like
to
be
infinite
of
much,
we
can
discuss
it
because
yeah
right
now,
the
only
maintainer
is
me
a
new.
G
E
E
E
Yeah
but
I
think
once
we
have
a
couple
my
debt
is
I
can
think
it
would
end.
These
are
the
police's
I
will
check
the
Charter,
but
if
the
Charter
distance',
that
releases
has
to
go
with
the
voting
thing
and
I'll
just
keep
releasing
much
Italy
I've
done
it
before,
which
is
on,
which
is
just
you
know
you
feel
released
whenever
you
need
to
do
a
reason.
There's
not
really
much
fanfare
about
it.
E
E
That's
I
think
yet
maintain
is,
is
one
thing
and
I
think
we
tie
it
with
at
least
two
people
I
think
is
enough
for
it
to
sort
of
be
put
into
people
forward,
and
then
the
other
thing
is
is
merging
validation,
stuff
I
mean
I'd,
be
down
to
just
like
literally
just
merge
like
subtree
merge
the
validations
from
initials
or
maybe
even
make
it
in
much
you
hold
it.
I,
don't
know
what
the
best
we
can
discuss.
Lets
me
in
the
Universities.
A
E
Yeah
and
there's
a
big
good
thing:
the
image
tool
stuff
that
I
did
you
fix
like,
for
instance,
the
invalidation
requires
pulling
stuff
in
the
internet,
so
we
need
to
like.
So
there
are
things
there
that
need
to
be
fixed,
but
because
of
interactive
maintain
is
and
all
that
so
that
they
haven't
been
touched.
So
yeah.
G
E
Mean
obviously
I'm
one
of
them
is
that
maintained
is
who
is
you
know
so
I
mean
I'm
stead
of
playing
myself
on
this
one
but
yeah.
It's
like
I,
think
I.
Think
if
it's,
if
it's
part
of
a
more
active
and
widely
used
project,
we
can,
we
can
get
more
development
and
possibly
even
more
active
from
other
maintenance
who
are
also
interested.
A
Yeah
I'm
keen
on
it,
because
it's
people
use
it
and
it's
it's
pretty
much
serving
the
purpose
that
I
think
we'd.
A
few
people
had
hoped
for
and
when
image
tools
to
come,
but
yeah
I'm
curious
on
my
on
the
other
releases,
because
I
think
we
got
a
the
that's
pretty
much
the
same
issue.
There
were
seeing
in
a
couple
places
right
now:
I
need
I
need
to
do
it.
A
I
mentioned
a
couple
weeks
ago,
doing
a
release
for
go
digest
and
the
thing
that
we
want
to
get
in
there
before
the
making
of
one
dato
release,
which
is
literally
just
updating
the
copyrights.
The
copyrights
are
still
healing
from
when
they
codebase
was
moved
over
and
it's
a
really
cute
cute,
lineup
of
of
maintained,
errs,
take
a
walk
down
memory
lane
to
look
at
all
the
code
digest
and
you're
maintainer
there
yeah.
C
D
C
D
E
Second,
one,
so
what
would
what
will
do
is
that
we
can
preserve
the
history
of,
but
will
it
will
basically
like
on
yeah,
remove
the
repo
from
openSUSE
to
open
containers
and
then
we'll
do
a
manual
like
it's
called
a
subtree
merge.
Basically,
where
you
take
the
history
of
one
repo
and
then
jam
it
into
another
repo
and
then
we'll
just
take
just
the
conformance
bits
from
initials,
because
that's
the
main
thing
that
the
people
use
it
for
and
it's
the
thing
very
much
you
can
do
it
that
way.
A
E
A
E
C
E
E
The
one
thing
that
the
the
main
argument
I
have
for
basically
archiving
image
tools
is
that
the
and
pointing
people
from
initials
to
much
is
that
the
extraction
encodes
the
extraction
generation
for
initials
is
like
completely
broken,
and
and
it's
just
because
it
was
renamed
like
the
simplest
way.
That
said
it
could
be,
but
it
like
it
just
it
doesn't
work.
The
reason
why
I
wrote
to
much
is
because
the
expression
code
just
doesn't
work
like
it
doesn't
correct,
like
it
doesn't
currently
do
time
stamps.
It
doesn't
want
two
things
indirectly,
yeah
sure.
A
Don't
think
that
osya
image
tools
has
any
brand
or
momentum
or
whatever
if
people
are
importing
from
it,
and
even
with
the
funny
name,
one
there's
probably
more
people
using
emoji.
So
how
to
give
credence
and
put
those
two
effectively
put
those
two
together,
whether
that
means
bringing
in
a
new
project
for
a
while,
it's
really
within
the
day
fixing
one
implementations,
broken
approach
with
another
was
just
a
merge
away:
yeah,
okay,
all
right!
Well,
figure
out
your
ideal
on
that
and
I'll
expect
to
see
a
team
out
of
the
TIV
one
way.
I
do.
E
A
B
Try
not
to
shake
your
there.
We
go
okay
chase
your
cursor
there,
so
first
I
was.
It
has
been
a
little
dormant
because
we've
been
busy
with
the
notary,
stuff
and
others,
and
the
first
thing
I
should
mention
is
I
similar
to
the
conversation
prior
is
making
sure
we
have
enough
containers
on
it.
So
just
because
I'm
busy
with
something
else,
doesn't
stall
it.
So
that
is
one
of
the
first
things
that
I'm
working
on
there's
a
couple:
people
that
have
asked
to
be
part
of
it.
That
will
be
acted.
B
So
that's
one
of
the
things
that
I
want
to
work
on.
The
other
was
the
blocking
part
of
artifacts
was
how
do
we
deal
with
the
root
media
type?
So
I
did
spend
some
time
reading
through
the
INR
work,
but
that's
like
we
said
or
whatever,
and
it
looks
extremely
promising,
so
it
for
people
that
already
know
it.
B
It's
like
duh
for
me:
I,
just
hadn't,
gotten
to
it
not
just
following
the
pattern,
so
I
need
to
continue
reading
finishing
reading
through
it
and
actually
go
through
the
process,
because
there
is
quote
experts
that
have
to
approve
things
and
I,
don't
know
exactly
how
that
works.
So
I
want
to
continue
to
make
progress
there
and
I
believe
that
is
the
last
thing
that
was
kind
of
blocking
us
getting
to
a
point
where
helm
and
others
can
actually
finalize
what
their
media
type
is.
So
we
could
actually
kind
of
declare
it
a
one.
B
Oh
so
people
can
start
using
it.
I
know
that
a
couple
of
other
clouds
are
actively
working
on
and
lenders
also
we're
actually
working
on
supporting
that.
So
we
want
to
get
that
out
there
and
actually
get
home
and
others
to
have
a
solid
media
type
that
they
could
use
the
what
builds
a
top.
That
is
in
the
notary
conversations
without
making
this
a
notary
conversation
and
we've
been
talking
about,
and
how
do
we
integrate
all
these
different
things
together?
So
we
want
to
sign
artifacts
that
are
in
the
registry.
B
Those
artifacts
were
obviously
container
images,
but
there
are
also
things
like
s
bombs
so
that
I
have
a
list
of
content
and
I
know
what
is
in
that
artifact,
because
some
things
you
can't
tell
by
scanning
it.
You
actually
not
only
know
if
you
write
that
out
beforehand
like
what
compiler
may
be
compiled
the
code
as
well
as
what
packages
are
in
it,
so
that
we
can
look
back
and
say.
We
know
that
this
package
has
a
vulnerability.
B
We
only
discovered
the
packages
at
MoMA
building
today,
but
what
about
the
content
that
was
deployed
last
month
and,
of
course,
having
this
s,
bomb
is
only
valid
if
it
can
be
signed.
So
oh
excuse
me,
so
that's
kind
of
the
aspect
there
and
then
what
we
were.
The
recent
conversation
we've
also
been.
Having
is
great
I've
got
these.
You
know
what,
if
I
want
to
constrain
my
environment,
to
only
support
my
sequel,
7.1
and
above
well,
we
don't
really
want
to
make
that
enforcement.
B
Obviously
part
of
notary
notary
is
just
saying
it
is
what
it
says.
It
is
what
it
is
we
don't
care.
Should
we
look
at
stuff
like
open
policy
agent
as
one
of
policy
enforcement's
that
can
validate
that
or
enforce
some
policy?
So
we've
been
thinking
about
this
from
an
end
to
end
and
basically
back
to
the
artifacts,
it
says
things
like
gas,
bombs
and
oppa's,
and
even
this
the
last
one
I'll
come
to
the
GPL
for
a
lot
of
these
licenses
requires
source
to
be
in
the
same
location
as
the
binary.
B
It
doesn't
have
to
be
with
the
binary
and
I
think.
This
is
the
thing
that
you
were
mentioning
a
couple
times
Vincent,
because
our
open
source
legal
team
came
to
me
and
was
like
hey.
How
do
we
close
that
loop
on
some
of
this
information
talking
to
one
of
the
people
over
at
one
of
the
companies
everywhere?
You
mentioned
it
and
basically
what
the
thought
process
there
is
the
source
for
the
package
can
be
put
as
an
artifact
in
the
registry
and
whether
we're
using
indexes
to
reference
these
things
or
otherwise.
B
C
E
A
A
Yeah,
do
it
this
to
navigate
the
the
initial
naive
implementation
is,
is
actually
I'll
do
a
finishing
a
write-up
of
the
layout
right
now,
but
the
naive
implementation
is
actually
putting
different
source
into
e
into
the
layer
tarballs.
So
even
naive,
old
client
tools
can
can
push
and
pull
them
to
registries
and
stuff
like
that,
but
the
next
approach
would
be
effectively
to
have
them
enumerated
as
their
own
descriptor
with
their
own
mime
type.
So
you
know
application
and
/v
MD,
sly
dot.
You
know
rpm
or
whatever
X
rpm,
so
it'd
be
a
source.
A
Rpm
pushed
right
at
the
registry
or
you
know
whatever
it's
it's
mime
type
is
so
that
it
wouldn't
it
wouldn't
be
runnable.
It
wouldn't
need
to
even
ever
touch
a
graph,
a
graph
storage
driver
of
a
container
runtime.
It
would
just
be
fetched
and
unpacked,
because
you'd
see
that
this
thing
is
a
certain
mind,
type
or.
B
B
Moche
tools
don't
have
to
pull
it,
so
it
really
makes
for
a
great
use
case
and
then,
as
a
registry,
you
know
whatever
package,
you
pick,
there's
probably
a
couple
of
thousand,
if
not
more
images
in
that
registry,
that
reference
it
and
because
most
of
us
do
various
types
of
deduping,
we
could
all
keep
track
of
its
pointing
to
the
thing,
but
we
don't
need
to
have
hundreds
of
thousands
of
copies
of
that
same
source
for
the
same
package.
So
to
me
it's
a
great
example
like
he
was.
B
You
know,
showed
me
a
slide
with
a
bunch
of
pictures
and
the
some
of
the
named
references
he
put
on.
There
might
be
a
little
bit
differently,
but
functionally
it's.
What
distribution
early
supports
and
what
the
artifact
stuff
all
it
does
is
just
put
some
semantic
meaning
to
some
of
the
stuff
that
they
were
doing.
Yeah.
B
So
anyways
that
was
just
the
quick
update.
I
do
you
know,
while
we're
busy
get
nerdery
going,
I
didn't
want
to
be
slack
on
the
artifact
stuff.
In
addition
to
just
all
the
stuff,
we've
been
really
busy
here
for
the
beginning
of
the
year,
so
I
just
wanted
to
give
a
quick
update.
My
I
don't
know
if
I
could
promised
a
date
by
which
I'm
trying
get
some
of
the
stuff
done,
because
the
Ayana
stuff
is
a
little
long,
but
it
has
all
the
roots
of
everything
that
seemed
to
make
perfect
sense.
E
Yeah
I
guess
two
questions,
or
one
is
more
of
like
a
comment.
/
thing
we
might
have
to
do
yeah
the
first
one
is
that
these
are
so
with
regards
to
it
being
a
description,
all
the
rest
of
it.
It
does
make
sense.
The
only
thing
is:
is
that
what
can
like
hard
cut
optional,
pulling
forearm
for
for
the
source
for
the
source,
descriptive,
but
I?
E
Let's
imagine
that
I'm
Red
Hat
osuzu,
who
will
canonical
or
something
and
I'm
stiffening
and
I'm
shipping,
my
ie
my
face
images
or
some
derivative
image
is
based
on
facing
it,
opens
it
for
the
best
example
this,
because
we
could,
because
we
have
two
OBS,
which
is
publicly
accessible
but
doesn't
matter.
The
point
is:
is
that
you
have
some
user
generates
an
image
or
generated
the
source
is
stored
in
OBS
in
this
case?
E
A
Of
the
conversations
even
from
yeah,
because
I
mean
rightfully
so
some
of
the
folks
like
on
the
quay
team
balked
at
this,
because
they
don't
want
all
the
all
the
pushed
item
and
if
you
had
something
like
because
right
now
we're
just
abusing
the
the
image
manifest.
So
when
you
see
the
the
array
of
layers,
they're,
not
they're,
not
filesystem,
layers
per
se,
you'd
have
to
walk
through
the
mime
types
and
treat
each
object
accordingly.
A
And
so,
even
if
we
have
this
array
of
sources
and
you
see
the
digest
of
new
sha-256
digest
of
it
or
whatever,
if
there
was
some,
you
know
either
abusing
the
earl
URL
filled
or
some
annotation
to
say,
here's
where
you
can
find
that
same
object,
the
the
legal
distinction,
the
only
the
only
thing
that
makes
my
brain
hurt
when
dealing
with
some
of
the
lawyers
is
the
LGPL
version.
2
section,
1,
subsection
d,
I
think
it
is
I'm
remembering
quickly,
painting.
A
Works,
the
the
source
of
a
works,
the
works
being
the
binaries.
The
source
of
the
works
should
be
available
in
equivalent
access,
so
it
as
a
distinction
from
some
of
the
other
like
MIT
and
GPL,
and
otherwise
they
just
say
the
sources
have
to
be
generally
available
that
license
in
particular
and
I.
Think
there's
a
couple
other
that
have
copied
that
wording
equivalent
access
lawyers
are
saying
should
generally
mean
that
it's
fetched
in
the
same
way
that
the
binaries
are
fetched.
A
So
if
you
pull
an
image
on
the
same
client
tools
that
should
be
able
to
pull
an
own
image
should
in
likeness
be
able
to
pull
those
sources
of
that
image.
So
if
we
did
do
some
kind
of
like
redirection
to
here's,
this
thing
that
here's
the
digest
of
the
thing
and
here's
a
URL
that
you
might
be
able
to
find
it
should
be
likewise
available
for
pulling
the
image
without
a
lot
of
like
pervasive,
is
needed.
A
E
In
that
case,
we
can
go
tech.
To
my
to
my
last
question,
which
is
a
which
is
a
second
question.
So
I
don't
want
to
I.
Don't
know
block
this
a
little
on
things
that
we'd
like
to
get
done,
but
one
I
just
want
to
make
sure
of
is
that
is
quote
produce
heavy
too
or
like
the
the
new
archive
format?
E
I
would
just
want
to
be
able
to
double-check
that
the
bill
of
materials
that
you're
talking
about
would
be
usable
for
that,
because
it's
because
I
mean
we
can,
if
you
can,
just
because
with
the
tar
archives.
Basically,
you
sort
of
trust
the
builders
to
generate
a
valid
set
of
build
materials,
but
we
don't
spiritual,
like
any
other,
like
open,
open,
transparent
format
for
OCI.
You
could
actually
do
the
all
of
the
checking
just
really
based
on
the
like
the
OCR
level
metadata.
E
So
you
could
actually
verify
that
the
build
materials
is
actually
complete
for
the
entire
archive
and
yeah.
One
thing
I,
just
wanna
double
check
is,
and
we
can
never
discuss
about
this
elsewhere
is
yeah.
Is
that
the
Bill
of
Materials
format?
We
use
that
we're
gonna
use
for
this,
for
the
for
the
old
style.
E
Images
of
the
current
style
ages
would
also
work
equally
well
for
the
other
one
because,
because
it
would
be
silly
to
after
now
migrate
build
materials
in
like
a
year
or
two
if
we
end
up
getting
the
the
new
fall
out,
so
that
was
one
thing
I.
What
about
moving
up.
A
B
The
thing
so
there's
a
couple
different
angles:
right,
there's
the
from
a
distribution
from
a
registry
operator
and
the
distribution
spec.
There's
part
of
me,
that's
like
if
I
wear
that
hat
as
the
artist
er
person-
and
you
know
what
saying
I'm
another
ticular
from
of
that
perspective,
we
don't
really
care
what's
stored
in
it.
We
just
want
to
know
what
it
is.
So
we
can
help
customers
under
no
underst
and
what's
in
this
storage
system
and
help
us
understand
when
they're
scanning
it,
we
would
love
my
branch
out.
B
I
love
the
a
bomb
and
all
these
other
things
that
could
actually
pull
together,
really
solid
end
in
my
stereos.
But
then
we
quickly
get
into
like
which
what
is
the
right
one
to
bet
on
and
bet
on
it
heavily
so,
for
instance,
the
three
TS
bomb
as
now
it's
being
referred
to,
because
that
particular
exist
bombs
too.
Generic
there's
a
group,
that's
working
on
it
I,
you
know,
I
happen
to
you
know,
participate
when
I
have
time
in
it
and
it
sounds
good.
B
But
you
know
red
has
been
working
on
theirs,
which
I
think
they've
had
for
a
longer
time,
just
because
one
does
well
shouldn't
exclude
the
other.
So
the
thing
that
makes
me
a
little
nervous
is
thinking
the
distribution
spec
would
have
to
adhere
like
would
endorse
a
particular
one
and
I'd
have
to
support
it.
As
opposed
to
the
artifact
approach,
says:
hey,
we
can
store
things
alongside
an
artifact
and
it
could
be
the
three
TS
bomb.
B
It
could
be
the
ibn
s
bomb
and
then
the
tooling
around
it
just
says
it
knows
how
to
work
with
these
various
different
types
of
formats,
and
then
the
individual
ecosystems
can
evolve
on
their
own
without
to
your
point,
having
to
revolutionize
the
the
spec
over
time.
Just
getting
an
index
have
a
media-type
there's
been
a
discussion
and
just
getting
artifacts
to
pretty
much
just
you
know
demonstrate
how
the
original
OCI
artifact
OCI
index
and
manifest
were
designed
to
pretty
flexible
in
support
in
99%
of.
B
B
D
B
G
A
B
You
should
I
think
you
want
them
saying
you
know.
Let's
just
say:
I'm
not
even
suggesting
you're
do
I,
know
you're
working
on
stuff
as
well
and
let's
say
I
there
could
be
three
es
box,
so
there
would
be
IBM.
You
know
the
NDS
IBM
s
bomb
there'd
be
three
tes
bomb.
They
need
me
to
ask
them
I'm
just
making
up
something
here.
The
point
is:
is
that
registries
know
how
to
store
different
artifact?
They
can
say
what
they
are
and
then
tooling
could
come
along
and
says.
B
Oh
I
don't
had
a
handle
these
two
of
these
two
of
the
three
s
bombs
and
then
I
could
go
on
and
do
its
thing,
and
the
idea
is
that
we
can
store
things.
The
thing
should
be
named
so
they're
various
tool.
You
can
understand
them
and
as
far
as
a
category
that
we
all
want,
but
I
don't
know
if
we'll
be
able
to
get
everybody
to
agree
to
one
so
from
an
artifact.
In
a
registry
perspective,
we
don't
need
to
bet
on
which
one
we
just
say
we
enable
all
of
them.
E
D
Was
so
I
was
I
was
gonna,
say
that,
okay,
that
that
makes
sense
from
a
higher
level
perspective
I
am
having
a
hard
time
figuring
out.
How
one
would
these
properties
of
all
the
artifacts
that
are
stored,
whether
there
s
bombs
or
whatever
in
the
image
such
that
a
tool
would
be
able
to
figure
out?
Okay,
I
found
I
found
this
media
type.
What
what
kind
of
media
type
is
it
and
can
I
do
anything
with
it
or
should
I
even
know
it
I.
E
So
any
question
is
the
way
that
we're
network
is
that
yeah,
just
like
media
media
isn't
tied
for
the
teeth
bomb
and
that's
a
tool.
The
stance
even
does
that
meet
the
type
and
knows
how
to
help
and
it
handles
it.
Why
is
it
which
is
basically
how
if
I
kind
of
would
inspect
öthey
exactly
here,
immediate
IP
is
that
you
can
specify
is
kinetics
tensions.
He
respects
which,
in
the
actual
image
tag,
that
different
tool
can
support.
E
Legal
separate
from
a
guy
like
you
have
like
Francisco
uses,
they
use
squash
is
sort
of
spot
for
their
helices,
something
that
you
can
do
just
by
an
immediate
type
of
the
layers
from
are
descriptive
us
and
now
other
tools.
Don't
support
that,
but
that's
an
extension
outside
suspect
the
thing
that
I
disagree
with
when
it
comes
to
the
Bill
of
Materials
stuff
in
a
good
supports,
sorry,
those
we
should
support
all
of
them,
but
we
should
not
explicitly
support
one.
E
My
problem
with
this
is
that,
while
this
is
okay-
or
this,
is
this
viewpoint
with
reasonable
when
it
comes
like
home
char,
something
where
we're
like,
we
don't
care.
What
is
it
Oxford?
Whatever
my
problem
with
is
scanning
and
having
package
lists
or
building
material
for
arm
for
images,
is
something
that
basically
every
build
tool
cares
about,
or
at
least
every
every
build
tool
that
you're
likely
to
see
I
mean
distribution,
build
tools,
all
care
about
that
at
the
very
least,
base
images.
E
Basically,
every
distribution
already
have
a
mechanism
in
some
way
to
generate
package
list,
I
mean
in
openSUSE.
We
have
OBS
generates
this,
for
you,
I'm
sure,
Red
Hat
has
a
simple
saying:
conical
was
another
thing,
so
I
think
that
the
ability
to
know
what's
inside
an
image,
I
think
much
more
fundamental
to
what
finds
an
image.
E
What
we're
actually
I
think
we
should
have
a
bill
of
materials
which
is,
and
we
endorsed
within
each
stack
now,
I
I
I
would
be
fine
without
having
like
who
was
the
options
in
the
same
way
that,
in
principle
like
we
have
multiple
like
compression
algorithms,
we
support
with
in
OCI
with
an
image
SPECT
I,
think
that
similarly,
which
have
some
you
should
say
these
are
to
bill
of
material
types,
that
every
tool
support.
If
it's
a
possible
material
sky
angle.
E
This
is
one
that
everyone
support
usable
bill,
material
standing
because
I
think
the
risk
you
run
with
having
having
like
a
laissez-faire
approach
to
just
knowing,
there's
pretty
or,
in
my
opinion,
to
what
makes
an
engine
image
based
image
filter.
An
image
blur
is:
is
that
you
end
up
with
balkanization
of
effectively
different
caps,
doing
different
things,
and
now
you
have.
E
About
well
there's
between
the
now,
oh,
if
you're
an
image
building
tool,
yes,
use
full
five
build
materials,
I've
spent
known
to
man,
because
then
because
you're
not
sure
if
the
consumer
will
support
one
of
them
and
I
would
like
to
avoid
that,
if
possible,
that
it's
giving
off
the
paint
stuff
with
artifacts,
it's
I
could
see
why
you'd
like
well,
let
me
do
the
middle
impossible.
I
do
I,
do
a
standard
concern,
but
that's
thing:
I
wanted
a
flag
in
that.
B
A
great
contain
think
about
this
play.
All
I'm
saying
is
I.
Don't
think
that
we
should
stall
the
project
by
trying
to
pick
one
because
I
don't
know
if
IBM
su
se
and
the
three
tedious
bomb
group
will
even
agree
on
one
and
if
they
can
awesome
like
I'm,
almost
saying
we
can
support
any
of
them
and
let
the
that
community,
you
know
that
are
working
across
the
end.
End
figure
out.
What's
the
right
one
like
is
JPEG
the
right
format
is
it.
It
is
a
gift
for
jiff.
B
E
And
I
will
say
to
me
to
be
fair
from
a
distribution
standpoint.
I
agree
that
you
shouldn't
really
care
I,
guess
I,
guess
my
my
argument
is
more
from
the
image
spec
viewpoint
of
like,
given
that,
if
we
want
to
make
image
scanning
like
a
very
cool
feature,
we
would
want
to
start
putting
stuff
in
image
spec
related
to
Danny,
especially
where
those
do
what
that
I.
That's
that's
been
already
stayed,
but
yeah.
B
I
think,
if
you,
if
we
think
of
a
proposal
that
I
think
Vincent's
been
talking
probably
related
to
this,
is,
should
an
S
farm
be
an
optional
layer
so
that
it
actually
goes
with
the
core
artifact.
But
when
I'm
trying
to
deploy
to
a
garbage
can-
and
it's
very
small
footprint,
don't
make
me
to
put
that
layer
on
that.
You
know
Debono.
A
A
B
A
D
A
D
I
went
on
top
of
the
hour
and
I'd
like
to
continue
this
conversation
because,
like
from
my
work
on
turn,
I
feel
like
these.
These
proposals
are
not
very
practical
because
they
come
from.
You
know
a
system
like
distro
perspective.
So
what
about
all
the
language
package
managers
that
people
use
to
install
stuff?
D
E
D
So
this
is
why
it's
I
mean
it's
easier.
If
there
was
something
like
the
tool
was
able
to
see
like
a
link
from
the
image
to
an
artifact
that
says:
okay,
that's
the
thing
that
I
need
to
just
read,
and
you
know
munch
into
whatever
format
is
required
for
the
end-user
or
follower
tool,
and
that's
okay.
It.
My
concern
is
just
this:
you
know
the
linking
from
the
image
to
the
artifact
and
then
are
we
able
to?
D
Are
we
able
to
have
like
maintain
provenance
of
the
the
links
from
the
end,
the
end
container
that
you
get
after
doing
you
know
from,
for
example,
I'm
using
docker
files,
but
that's
basically
how
I
see
the
ecosystem
working
is
that
you
start
from
starting
image.
You
add
your
layers
on
top
of
it,
the
starting
image
has
an
S
perm.
The
layers
have
their
own
s
bomb.
A
A
Am
but
it's
not
it's
not
a
design
doc
to
that
extent
of
how
they
all
plumb
together.
Let
me
what
yeah,
let
me
Oh
I'll,
follow
up
honestly
a
follow
up
with
you
on
the
list
about
it,
because
this
may
be
what
timezone
the
any
are
you
in
central
or
cific?
You
were
Pacific
your
own
person,
okay
and
I.
Will,
let
me
so
I
didn't
feel
like
this
would
just
be
I.
B
We're
trying
to
get
a
bunch
of
people
in
the
various
groups
that
talk
to
you
there
exactly
about
these
kind
of
things.
There's
a
I
have
heard
those
conversations
you're
bringing
up
exactly
in
the
various
s
bottom
conversations.
The
conversations
we've
been
having
a
notary
for
with
Justin
on
how
we
would
you
know
in
Sam
on
how
we
would
track
and
link
these
different
artifacts
together
in
a
loosely
coupled
way
is
exactly
the
kind
of
things
we've
been
talking
about.
Maybe
we
we
make
this
for
next
week's
agenda
and
actually
block
out
time
or
I.
A
I'm
not
I'm
not
available.
Next,
we
cut
up
I
mean
if
somebody
else
chairs
the
call,
that's
fine,
but
at
moments
just
like
put
post
something
to
list
and
I'll
open
up
the
bridge,
and
we
can
walk
through
this
specifically-
maybe
not
quite
open-ended
like
that.
As
far
as
time,
wise
goes,
but
to
walk
through
some
of
those
concerns
and
see
how
these
would
map
together,
because
I
think
they're
distinct,
but
they
they
should
be
able
to
touch
and
really
don't
relate
to
each
other.
I'm.
B
Hoping
that
each
to
how
they're
thinking
with
the
things
that
Nisha
is
talking
about,
and
then
we
just
got
to
figure
out
how
they
get
stitched
together,
like
an
s
bomb,
is
useless.
Unless
it's
signed,
an
s
bomb
should
be
able
to
link
to
an
artifact
which
also
needs
to
be
signed.
So
that's
there's
a
posto
linked,
but
something
I
refer
to
as
separation
of
concerns
on
notary
that
tried
to
delegate
these
pieces
to
the
individual
groups
that
are
focused
on
them
and
then
just
figure
out
what
the
connecting
pieces
are.
B
D
B
Image
grows
yeah,
so
Kate
from
SPD
acts
as
part
of
the
three
TS
bond
group.
That
I
was
talking
about.
Okay,
an
SPD
X
liked
I
think
they
referred
to
it.
Where
is
the
basis
of
three
TS
bomb,
and
that's
one
of
the
reasons
why
we
want
to
make
sure
s
bombs
can
be
disconnected
so
that
when
they
do
get
big
I
don't
have
to
deploy
that
big
thing
down
to
that
garbage
can
for
the
same
thing,
we
were
just
talking
about
the
soil
yeah.
Well,
there,
the
s
bomb
shouldn't,
have
to
go
there.