►
From YouTube: OCI Weekly Discussion - 2023-06-15
C
C
A
C
C
That's
good
well
just
moments
before
this
I
finally
was
chewing
through
some
to
do
so.
I
was
I
was
re.
Reviewing
that
10
30,
that's
been
up
for
comment
for
a
little
while
from
Aaron
but
happy
to
jump
into
some
other
ones.
If
there's
high
priority
stuff.
A
A
good
first
one
or
set
of
them
to
look
through
might
be
wrong.
I've
ever
seen
your
name
right,
correct
me.
If
I'm
ever
mispronouncing,
it
he's
got
a
few
out
there
for
conformance
testing.
That
I
think
would
be
good
to
look
at.
E
So
the
in
this
test,
what
we
have
is
there
is
a
variable
called
Delete
manifest
before
blobs
and
that
was
initially
I
think
the
way
Josh
wrote
this
I
believe
was
he
set
it
as
Paul's
first,
so
he
wants
to
by
default,
delete
blobs
first
and
then
the
Manifest
later
so,
which
I
think
is
somewhat
of
a
dangerous
operation
to
have
by
default.
E
So
we
just
clip
that
around.
So
what
you're
saying
now
is
by
default
you,
because
we
have
the
blobs
API
and
the
Manifest
API
you
can
delete
blocks.
You
can
delete,
manifests
so
I
think
the
safer
option
is
to
enforce
the
Delete
manifest
before
blobs,
because
once
blobs
manifest
gets
deleted,
then
it's
no
longer
visible
to
users
under
normal
circumstances,
and
then
so.
This
is
the.
E
A
It
looked
good
to
me.
I
know,
we've
also
seen
there's
a
J
frog.
One
of
them
was
asking
about
making
the
whole
blob
delete.
I'm,
not
sure.
If
optional
is
the
right
word,
I
know:
they've
they've
asked
for
us
to
not
hit
it
so
hard
on
the
performance.
That's
why
their
performance
run
doesn't
even
do
some
of
the
content
management.
A
A
E
So
I
believe
the
the
author
here
intended
to
test
that
the
empty
layer
is
also
deleted.
So
when
you
push
an
empty
layer
in,
we
are
also
able
to
delete
it.
F
E
E
Yeah
I,
don't
I,
don't
know
the
the
author
is
going
to
circle
around
and
fix
this,
but
if
not,
we
have
to
adopt
this
pier
and
then
push
it
in
so.
But
comments
are
welcome.
A
A
There
was
the
whole
skimped
empty
layer
test,
and
so
they
might
not
have
actually
pushed
that
empty
layer
manifest.
Yes,.
A
The
other
option
is
to
take
their
PR
and
pull
down
the
code,
put
the
change
on
top
of
it
and
push
the
second
PR
on
top
on
a
new
branch,
and
that
way
you
can,
if
you're
e-basing
on
top
of
theirs
or
adding
a
commit
on
top
of
theirs,
then
it's
just
going
to
say
signed
off
by
both
people.
At
that
point,.
F
E
E
So
let's
go
to
setup.gov
first,
so
the
way
this
is
set
up
is
we
have
a
a
new
environment
variable
that
you
can
turn
on
or
off
with
it.
Based
on
that,
this
test
will
be
done,
and
this
is
all
the
setup
for
that.
So.
E
A
G
A
A
E
E
It
one
by
one
okay,
so
so
we
push
the
empty
Json
blob.
We
make
sure
that
this
is
that
it
matches
the
hardcore
value.
A
I
I
just
resolved
that
one
just
a
second
ago,
I
think
you've
gotten
all
of
my
comments
in
here,
and
so
it's
on
me
right
now,
just
same
as
other
one,
where
I
need
to
actually
run
this
against
a
couple.
Registries
double
check:
everything
before
I
put
my
thumbs
up,
but
okay.
Overall,
it's
been
looking
good
to
me.
Okay,.
E
So
the
the
idea
here
is,
we
pushed
the
the
empty
blob.
The
two
cases
we
test
is
we
push
the
image
and
then
we
push
a
reference
to
that
image.
E
The
other
case
we
are
testing
is
the
opposite.
We
push
the
reference
first
and
then
the
image
later,
so
both
cases
are
being
tested
here.
E
A
Every
time
I
read
through
this
one
I
find
one
more
thing.
So
I
had
to
put
a
nerd
in
there
for
Josh
to
take
a
look
at
the
worn
card
at
the
very
end.
That's
way
down
at
the
bottom,
but
I
just
want
to
have
him
give
a
second
set
of
eyes,
since
he
knows
a
lot
of
this
conformance
code
in
there
of
what
we're
doing
with
that
warning
function.
Just
to
make
sure
that
he's
happy
with
that.
You.
F
E
So
yeah
so
I
added
yeah,
so
I
added
a
new
method
yeah.
So
this
is
one
of
one
of
the
things
that
came
up
with
during
the
discussions
is
so
the
conformance
test
will
enforce
the
must
conditions,
and
then
it
will
run
on
the
failed,
should
conditions.
So
so
this
was
the.
A
Yeah
and
I
like
doing
this
for
the
non-must
use
cases
in
terms
of
hey,
you
should
be
doing
something
different
than
that
I've.
Just
I
think
Josh
in
that
one
to
say:
hey
is
this,
given
that
he's
done
so
much
work
on
some
of
this
code
in
the
past
and
noticed
some
of
these
other
libraries?
Are
there
different
ways
that
he
would
prefer
to
see
this
done
or
not?.
F
A
G
E
Scored
this
is
how
you
do
it
right,
so
the
OCA
filters
you
may
or
may
not
support
it
regardless
the
the
output
should
be
correct,
but
when
you
don't
support
it,
we
issue
a
warning
like
this.
A
Yeah,
at
least
at
some
point
I
was
debating
going
through.
Some
of
those
checks
instead
of
saying,
is
equal
to
two
or
four
to
make
sure
it's
at
least
two
and
at
least
four
and
then
to
look
for
items
within
the
list.
I
figured
that
was
more
effort
than
we
needed
for
our
performance
test.
No
one's
going
to
have
other
data
in
here
already
they
do.
They
just
need
to
give
us
an
empty
repo.
E
F
A
For
all
I,
really
like
this
itari,
or
maybe
even
from
the
docker
Hub
side,
if
you've
been
looking
at
this
at
all
thoughts
on
adding
this
conformist
code
against
your
Registries,
any
any
pros
and
cons
against
it.
D
Let's
just
do
push
that
first
before
and
after
right,
I
think
that's
awesome,
yeah
yeah,
we've
we've
had
this
scenario
suggested
internally,
so
I
think
it's
not
no
concern
from
our
side.
E
Yeah
so
so
feel
free
to
take
this
patch
and
run
it
against
your
Registries.
And
if
you
have
any
comments,
please
post
and
and
then
we'll
take
it
up.
E
So
at
least
at
least
we
have
the
tools
that
we
can
give
to
the
the
to
folks
outside
right
outside
of
this
room.
So.
A
B
D
D
D
That'll
break
but
I
think
in
general.
The
scenario
of
pushing
the
refers
before
and
after
is
covered.
D
What
about
the
oci
subject,
header
as
well
yeah,
that's
in
there
as
well
yeah,
so
those
will
break
for
us
because
we
haven't
implemented
rc2.
Yet
so,
okay,
but
but
in
general
I
think
it's
which
looks
nice.
E
And
then
you
can
control
it
right.
So
if
you,
if
you
can,
if
you
don't
want
to
test
the
reference
just
just
unset
the
the
environment
and
you're
good,
to
go.
F
G
E
So
what
this
is
testing
is
also
by
default.
We
are
testing
whether
a
zero
is
a
zero
layer.
Image
manifest
can
be
pushed
to
a
registry
and
the
zero
layer
image
manifest
is
the
should
condition.
So
that's
why
we
put
a
warning,
so
this
was
the
other
style
of
doing
a
warning.
You
could
just
skip
it,
but
the
the
issue
with
this
is
the
output
is
nice,
except
that
anything
passes
will
get
skipped
so
I,
don't
particularly
like
this.
E
A
A
That's
where
I
was
going
to
go
is
whatever
we
pick
get
both
of
these
on
the
same
same
design,
and
so,
if
folks
here
have
an
opinion,
great
chime
in
if
otherwise
I'm
ping
Josh
listen
just
because
I
was
hoping
he
might
have
some
ideas,
yeah
and
then
we'll
get
this
down
to
a
single
implementation.
Yeah.
E
So
this
is,
this
is
a
limiting
right.
This
is
the
the
this
functionality
is
limiting,
because
then
you
can
only
use
it
in
certain
situations.
I
I
prefer
the
other
generic
one,
the
in
the
other
PR.
But
you
know
that's
my
opinion.
A
A
D
Yeah
I
guess
you
and
wouldn't
have
the
most
comments
on
that.
So
I
think
it's
better.
If
you
folks
can
agree
on
that.
A
Yeah
and
I've
gone
back
and
forth.
I
went
ahead
and
threw
out
my
example:
I
need
to
go.
Do
pretty
pictures
like
other
people
doing
theirs,
but
I
did
put
somewhere
down
here
of.
A
There
is
an
example
sitting
out
there
folks
want
to
look
at
this,
that
it
just
implements
one
of
these.
Where,
since
we're
throwing
the
subject
on
an
index,
you
can
do
this
funny
little
thing
where
both
the
client
is
going
to
digest
and
the
subjects
point
to
a
digest,
and
if
those
were
the
same
digest
in
this
case,
where
you
go
down
to
this,
and
you
look
back
to
that
depending
on
client
implementations
and
so
I
wanted
to
give
people
a
chance
to
say
test
their
client
against
a
specific.
D
I,
don't
know
if
Vincent
is
still
listening
on
a
call,
but
one
thing
is
that
we
I
spoke
with
the
oras
library
folks,
and
they
said
they
don't
have
a
problem
with
this,
primarily
because
it's
walking
down
the
graph
and
it's
just
two
parts:
it's
actually
not
forming
a
loop
in
any
way
so
they're,
okay,
if
the
images
are
all
still
kind
of
like
pointing
downwards,
and
even
if
it's
the
same
image
or
the
subject
at
first,
it
might
be
okay,
that's
that's
one
piece
of
information:
I
can.
A
Add
you
so
the
case
here
is
you
would
look
at
this
digest
the
569.
You
would
find
this
image.
You
would
ask
this
image.
What
we
first
used
say
call
refers.
Api
and
you'll
get
back
this
index
up
here,
because
it's
got
the
subject
field
defined
in
it
and
then
you
would
go
say
well
we're
the
child
objects
of
this
one,
and
so
you
would
find
here
it
is
on
this
case
right
there
on
that
lines.
A
D
A
It
it's
similar
for
me
too,
when
we
were
worried
about
the
media
type
being
an
optional
field,
and
we
said
nobody
should
really
be
generating
a
manifest
that
has
both
manifest
and
layers
in
the
same
manifest.
But
just
because
they
shouldn't
doesn't
mean
done
me
is
not
possible
and
they
couldn't
do
something
bad.
If
they
did
that.
B
C
A
This
was
the
big
one
for,
for
my
side,
the
the
caveat
I'll
say:
maintain
a
list
gets
more
complicated
when
you
throw
it
on
concurrency.
G
C
That
sounds
like
a
you
problem,
not
a
me
problem.
No
I
get
that
I
can
be
challenged,
but
that's
solvable
problem.
A
The
other
big
one
that
I
had
turned
out
a
few
times
was
get
some
actual
use
cases
out
there,
people
showing
where
they're
doing
this
and
the
one
that
Jonathan
thrown
out
was
saying.
Hey
I
can
have
a
an
s-bomb
index
in
addition
to
the
image
index
and
have
all
my
response
in
one
place
and
responsible
as
well.
That's
nice!
If
you
do
that,
but
then
what
happens
when
someone
does
this
to
you
and
just
starts
throwing
other
s-bombs
out
there
that
you
think
are
there
but
aren't
really
there.
D
I
didn't
have
a
pretty
pretty
picture,
but
I
did
call
out
the
conversation
with
cnab.
They
couldn't
use
the
index
because
of
the
lack
of
heart
attack,
type
and
things
like
that
in
the
index
and
so
completing
this
would
have
enabled
them
to
do
that
as
well.
So
it's
more
than
just
the
hierarchical
index
or
respond
that
the
scenario
would
enabled
again.
D
A
D
So
one
of
the
things
was
they
wanted
one
of
the
Manifest
to
be
an
image,
and
so
they
wanted
to
use
the
index,
but
that
image
is
like
a
driver
image
that
kind
of
like
installs
stuff
and
run
stuff.
So
it's
almost
like
an
initialization,
but
that
is
not
the
actual
application
right
so
being
able
to
author
something
that
composes
multiple
images
as
well
as
calling
it.
Something
was
what
the
what
the
use
case
was,
and
they
had
to
step
away
from
using
the
index
primarily
because
there
was
no
artifact
type.
D
It
was
not
I
mean
it
had
to
be
sitting
inside
a
config
of
something
or
use
an
annotation
on
those
or
anything.
So
so
anyway.
The
point
is
that
clearly
there
was
a
use
case.
They
did
come
around
and
they
kind
of
like
stepped
away
because
they
couldn't
do
it.
I
think
this
is
an
opportunity
for
us
to
at
least
make
those
use.
Cases
happen
going
forward
and
not
leave
the
index
out
this
time.
Right.
D
We've
done
that
multiple
times
where
the
images
kind
of
like
looking
but
there's
quite
a
lot
of
value
in
the
index
is
what
I
feel.
A
C
G
D
Wanted
to
thank
you,
I
I,
know
that
you
have
reservations
on
this,
but
I
really
appreciate
it.
The
thing
is
that
we
should
call
out
the
concurrency
use
case,
because
we
hit
that
for
the
tag
schema,
so
I
didn't
want
to
talk
about
it
till
you
took
it
off,
but
the
tax
scheme
has
the
same
problem
when
you
try
to
copy
over
a
bunch
of
things
to
update
manifest
in
Google
manifest.
So
it's
not
like
the
concurrency
problem
can
be
solved
just
because
yeah.
A
F
A
A
A
D
A
Json
schema
and
yeah.
A
D
This
is
just
to
confirm,
we
can
merge
the
docs
and
then
I'll,
follow
up
by
end
of
before
the
next
call
for
the
scheme
appear.
C
The
only
I'd
have
to
go
to
on
the
the
comment
that.
C
As
a
registry
operator,
if
it
imposes
things
like
what
soju's
comment,
was
I
reviewed,
the
changes
before
I
read
all
the
comments,
but
this
this
one
line
was
the
only
one
that
was
like
that
must
not
alter
like
I
agree
with
it.
But
if,
if
the
intent
was
the
was
just
not
to
generate
errors
or
like,
if
things
are
not.
C
Never
yeah
yeah,
which
is
like
I,
said
completely
agree
with
it,
but
if
it's
just
as
far
as
like
supporting
use
of
artifact,
you
know
you
use
as
an
artifact
slightly
different
than
the
topic
of
the
pr
bit.
I
don't
disagree
with
it,
so
maybe
it's
fun,
but
as
far
as
the.
H
C
B
D
I,
wasn't
a
customer
call
last
week
where
they
said
they
want
to
do
disable,
pushing
register
like
image
types
and
they
wanted
to
use
like
a
specific
file
type.
Only
a
bicep
module
has
something
that
Azure
stores
and
they
said
they
want
to
use
that
to
store
like
NSG
rules
and
things
like
that
and
not
allow
any
other
artifacts
in
there,
because
they
wanted
to
use
that
as
a
sharing
mechanism.
But
they
don't
want
that.
D
Images
and
you're
like
no.
We
don't
support
that
right
now,
because
you
can
masculate
anything
as
an
image,
the
and
but
they
were
ready
to
kind
of
do
the
work
of
kind
of
like
blocking
stuff,
cleaning,
stuff
and
stuff
like
that,
and.
C
You
know
whether
it's
like
a
HTTP
accepts
header
or
something
like
that
or
like
these
are
the
things
that
are
configured
for
this
endpoint
to
be
known,
are
accepted
or
whatever,
like
that.
I
think
it's
interesting,
maybe
maybe
so
if
this
wording
allows
for
that,
then
that's
that's
fine
or
if
there's
a
small
tweak
that
we
can
make
to
allow
for
that.
But.
D
No
I
think
in
general
the
Voting
is
okay.
It
doesn't
confuse
anybody,
just
kind
of
adds
over
what
an
artifact
type
is
so
me,
it
doesn't
actually
make
any
like
significant
changes
to
how
we
interpret
perspective.
It's
fine.
A
I
know
we're
going
back
and
forth
over
this
a
while
back.
It
was
thinking
that
there
were
those
Registries
that
wanted
to
only
allow
album,
charts
and
a
few
other
things
they
knew
about,
but
not
allow
any
arbitrary
and
yeah
I.
Think
like
Sasha
is
saying
one
of
the
comments
that
came
back
from
that
was
well.
You
can
just
impersonate
anything
you
want
to.
You
can
pretend
you're
a
Helm
chart,
even
though
you're
not
really
a
home
chart
and
get
allowed
onto
a
repository.
A
H
Good
looks
like
there's
quite
a
stack
on
the
1.1.
Did
you
guys
want
to
try
to
close
the
image
first
and
then
close
the
1.1
stuff
on
distribution
or.
A
A
lot
of
stuff,
that's
out
sound
distribution
is
conformance
and
ROM.
Kumar
has
been
doing
an
awesome
job
on
that.
So
we
reviewed
a
bunch
of
those
to
start
with,
and
the
only
reason
I
haven't
put
my
own
green
checks
on
is
I
haven't
had
time
to
actually
spin
it
up
and
run
against
some
Registries
and
make
sure
it
looks
good.
But
it
looks
good
to
me
in
general.
A
H
Was
like
wait
a
minute,
yeah
question:
if
whether
423
should
be
in
there
or
not,
I
did
a
review
and
I
think
it
should
be
in.
A
If
there's
anything,
I've
missed
feel
free
to
tag
it
for
the
milestone.
A
G
A
Okay
and
yeah
that
that
one
I
think
we'll
get
that
one
knocked
out
here,
pretty
quick.
H
A
F
D
Oh
yeah,
do
you
want
to
a
random
deal
for
Quantum
large
1030?
It's
been
sitting
for
quite
some
time
so
and
nobody's
decides.
We
I
mean
you
or
me
have
kind
of
commented
on
that.
So
there's
no
big
issue
that
we
want
to
block
on
then
it
might
be
worth
just
want
to
be
just
like
sensitive
to
Adam
and
also
he's
being
on
the
call
to
like
so
many
so
many
weeks,
yeah.
A
Can't
think
of
the
best
word
for
that,
but
very
accepting
of
our
difficulties
in
trying
to
get
anything
through.
There's
the
eye,
spec
and
so
I
took
any
of
the
blocks.
I
had
off
on
their
moderately
comments.
So
if
someone
wants
it
merge,
I
won't
I
won't
raise
the
stink
at
all.
G
A
I
think
most
everything
in
there
was
just
saying:
hey,
let's
make
sure
if
it's
for
a
registry
versus
for
a
client
and
something
else
not
think
he
went
through
a
bunch
of
them
and
got
it
cleaned
up.
If
there's
something
really
bad
and
we
need
to
clean
up
again
later
on
with
the
follow
on,
we
can
always
do
another
PR
later
on
if
we
missed
something
yep
cool
thanks
for
your
bets,
all.