►
From YouTube: OCI Weekly Discussion 2020-03-04
Description
https://hackmd.io/El8Dd2xrTlCaCG59ns5cwg?view#March-4-2020
Chat:
```
00:28:47 Nisha Kumar: Link to backlog?
00:28:58 Nisha Kumar: Or is that not public?
00:32:20 Phil Estes: Nisha: Vincent just put links to the general issues tab for runtime and dist. spec
00:32:22 Phil Estes: In the notes
00:33:08 Phil Estes: https://github.com/opencontainers/distribution-spec/issues
00:33:20 Phil Estes: https://github.com/opencontainers/runtime-spec/issues
00:33:32 Phil Estes: https://github.com/opencontainers/image-spec/issues
00:34:00 Phil Estes: Meeting agenda/notes if anyone needs them: https://hackmd.io/El8Dd2xrTlCaCG59ns5cwg
00:39:35 Steve Lasker: https://github.com/opencontainers/distribution-spec/issues/71 - label for `vNext`
01:02:48 Josh Dolitsky: Gotta run - peace y’all
```
A
C
B
D
D
Yeah
and
there's
there's
it's
not
to
be
confused
with
the
containers
that
we
all
work
with,
that
there's
a
couple
of
plug-ins
for
Firefox
for
having
all
your
tabs
and
containers
now
and
I've
been
using
them
for
the
last
two
and
a
half
months
for
something
and
well
one
of
them.
I've
been
using
for
a
while
with
one
of
them
some
couple
of
extra
ones,
but
only
been
plus
two
and
a
half
months,
and
they
have
revolutionized.
My
browser
works.
Oh
I
love.
It.
A
D
Individual
account,
even
that's
nice,
yeah
I've
got
a
few
of
them
that
are
like,
like
LinkedIn
any
time,
I
open
LinkedIn.
It
automatically
jumps
into
a
container.
That's
only
sandbox
for
that
one
and
github,
and
so
forth
particular
accounts
and
be
neat
too,
but
every
every
every
time,
I
control
every
time,
I
open
a
new
tab.
It
just
it
starts
a
temporary
like
a
clean
slate
container
and.
D
E
B
So
I
will
keep
this
brief,
but
if
people
want
to
ask
more
detailed
questions,
there's
a
range
of
us
from
the
TOB
on
this
call
already.
So
you
know
we
can
field
specifics.
If
you
look
at
today's
call,
I
have
a
link
to
another
hack
MD,
which
has
notes
from
our
Monday
late
day.
Tob
call
the
purpose
of
that
call
was
to
effectively
move
toward.
B
We
have
two
project
proposals,
so
I
think
people
are
aware
of
oroz
an
emoji
proposed
to
be
added
to
the
OCI.
The
tlb
call
was
not
to
jump
directly
to
those
proposals,
but
given
we
have
new
seats,
just
filled
on
the
TOB
to
talk
a
little
bit
of
history,
a
little
bit
about
voting
process
and
proposals,
and
then
I'd
say
80%
of
that
call
was
really
discussing
that
we
need
to
develop
a
framework
to
handle
proposed
project
additions
to
the
OCI
and
the
the
reason
being
for
those
that
haven't
been
around
for
four
plus
years.
B
You
know
the
early
days
of
the
OCI,
you
know
if
you
read
through
the
governance,
I
think
it
becomes
fairly
clear
that
the
expectation
is
that
it
would
really
stay.
Quite
tightly
scoped
around
the
specs
and
again,
the
initial
specs
were
runtime
and
image
and
run
see
as
that
default
implementation
of
the
runtime
spec.
B
As
far
as
having
kind
of
a
framework
for
interacting
with
the
specs,
and
then
we
have
selinux
and
go
digest,
which
are
more
helper,
very,
very
simple,
helper
libraries
to
help
you
fill
out
or
use.
You
know
very
specific
parts
of
the
spec
like
assembling
digests
and
assembling
the
proper
SELinux
pieces
of
the
spec.
B
So,
given
that
it
was
important
for
us
to
to
take
a
step
back-
and
you
know
make
sure
that
that
proposers
are
aware
that
the
OCI
has
a
very
different
vision
and
scope,
the
CNC
f,
and
we
want
to
make
sure
that
we
don't
make
these
decisions
based
on
popularity
based
on
hey.
This
is
really
nice
code,
but
really
have
a
very
specific
set
of
criteria.
B
Given
we
have
nowhere
near
the
staff,
we
don't
have
the
marketing,
we
don't
have
the
you
know,
organization
of
the
CNC
F
to
actually
do
project
management
and
that
may
sound
silly
for
something
small
or
something
that
seems
like
just
a
helper
tool,
but
you
have
to
think
forward
a
year
to
what,
if
there
are
no
more
maintained,
errs
for
that
project
now
the
OCI
TOB
has
to
figure
out.
How
do
we
archive
this?
What
do
we
do
with
it?
B
Do
we
give
it
back
to
someone
else
so
anyway,
we're
trying
to
be
careful
before
we
step
in
to
just
making
sort
of
off-the-cuff
decisions
on
projects,
so
hopefully
no
one's
deeply
offended.
You
know
that
we're
taking
a
little
bit
of
caution
here.
First,
if
you
go
read
those
notes
and
scroll
down,
there's
a
start
of
an
attempt
to
think
about
project
types.
B
B
C
Think
the
there
is
a
yeah.
Obviously,
registries
want
to
be
able
to
support
deleting
tags.
So
I
don't
think
we
want
to
say
the
conformance
says
you
must
fail
a
tag
delete
which,
but
I
think
there
was
a
good
topic
of.
Can
we
break
this
up
and
see
some?
You
know
a
specific
list.
I
was
kind
of
using
the
ski
trail
list,
kind
of
idea
of
yeah.
G
G
Yeah
so
I
mean
I,
I.
Guess
there's
that
much
to
discuss
here,
but
I
think
we
do
have
a
proposed
solution
for
that
specific
in
the
what
Steve
is
mentioning
is
what's
being
talked
about
on
the
mailing
list
topic
that
Derrick
and
Romney
attached.
She
commented
out
and
again
deferring
to
Derrick's
comments.
I
think
he
had
one
that
was
broke,
the
distribution
spec
into
four
main
categories,
the
most
important
one
being
the
tool
operation
and
so
I
guess
what
I'm?
D
Are
you
talking
about
like
the
breaking
down
on
the
categories
a
bit
yeah
so
right
now,
it's
seven
I
think
I
think
that,
oh,
that
was
how
it
was
already
kind
of
broken
down,
but
it
is
interesting
to
like,
rather
than
just
doing,
an
API
level
check
taxi
grouping
them
into
workflows.
I,
don't
know
how
about
right.
D
C
I
C
It
the
workflows
as
a
higher
level
concept,
which
I
think
is
good
like
when
we
first
looked
at
the
conformance
test,
there's
just
this
red
and
green
and
then,
if
you
look
a
little
closer,
there's
a
quantity
and
you
know
it's
not
really
qualitative,
it's
just
an
arbitrary
arbitrates,
obviously
a
finite
number,
and
what
does
it
really
mean?
So
the
thought
was:
can
we
break
them
down
into
a
set
of
here's,
the
api's
that
pass
and
fail?
Because
and
then
we
could
discuss
whether
something
should
actually
fail
or
like
the
delete?
C
Api
should
support
different
responses,
because
I
think
I
put
it
in
this
one
where
it
basically
said,
look
for
a
public
registry.
We
probably
don't
want
delete
to
happen
because
then
we
wind
up
in
this
NPM
kind
of
scenario,
where
everybody's
tools
break
and
because
something
got
deleted.
The
so
certain
registries
might
have
different
behavior,
but
the
fact
that
a
registry
doesn't
page,
for
instance
or
I,
don't
know
pick
another
arbitrary
thing
in
there.
That's
not
the
end
of
the
world,
I'm
sure
the
register,
you
look
red.
Where
should
it
just
be
like
hey?
C
G
I
think
there's
to
me
the
reasons
are
clear:
why
a
certain
registry
wouldn't
or
certain
points
and
I
think
that
that's
fair,
so
yeah.
So
what
I?
What
I
think
a
good
suggestion
is.
Is
that
there's
four
tests
there's
for
greener
Reds
if
you
pass
the
first
one
which
is
poor,
your
overall
status
is
green,
but
people
could
dive
in
and
see
just
how
much
of
the
spec
your
your
support
method,
but
all
registry
should
learn
a
cool
operation.
Have
a
nice
guy.
J
Well,
just
one
there's
only
one
tactic,
another
tactic
is
is
just
to
you
know,
clearly
define
where
may
language
is
going
to
be
used
in
the
specification,
rather
than
must
language
right,
I
mean,
in
other
words,
if
we're
gonna
allow
some,
you
know,
delete
not
to
be
used,
and
it's
okay
for
it
to
only
you
know.
Maybe
it's
not
you
don't
have
to
right.
So
the
pool
is
the
problem
right.
We're
doing
here.
G
D
D
D
J
D
J
C
D
C
I
do
think
it's
important
for
us
to
call
out
what
registries
support.
What
API
is
so
that
people
that
are
writing
tools
are
trying
to
do
a
certain
behavior
on
a
certain
registry
and
maybe
in
the
test
results.
A
registry
can
specify
to
put
a
comment:
hey.
We
don't
support
this.
For
this
reason
you
know
delete,
for
instance,
on
docker
hub
or
paging,
but
we
support
this
other
API
because
we
believe
it's
better
I,
don't
know
whatever,
but
at
least
give
registries
and
opportunity
to
explain
why
or
where
something
like
might
be
somewhere
else.
I.
A
Think
if
it's
an
optional
behavior
by
definition,
isn't
required,
so
a
registry
that
doesn't
implement
an
optional
behavior
should
still
show
up
as
conformant.
It
should
still
show
up
as
green
and
yeah.
There
may
need
to
be
something
that
says
this
optional
feature
is
not
supported,
but
that's
not
something
that
says
it's
not
conforming
with
the
specification
agreed.
G
G
C
G
I
mean
I
just
think
if
your
environment
tables
clean
of
any
OC,
I
underscore
extra
test
one.
It
just
runs
the
baseline,
otherwise
Egypt
it'll
eat
yourself,
you
know
and
that
those
we
can
break
you
again,
yeah
I,
don't
want
to
yeah
I,
don't
want
to
spend
way
too
much
time
and
I'm
happy
to
have
one
oxen
AAPIs.
The
only
the
only
other
topic
I
had
was
regarding
performances
did
want
to
release
the
binaries
as
a
container
image
and
I
was
curious.
G
D
J
E
D
G
C
A
C
We've
been
having
a
couple
of
conversations
around
like
when
would
you
put
something
on
github
package
registry
versus
in
your
registry?
That's
close
to
your
deployment,
and
so
I
was
trying
to
find
some
opportunities
where
actually
it
does
make
sense
kind
of
pitching
against
them.
So
I,
just
I
thought
this
was
a
good
example
and
if
they
are
not
doing
something,
that'd
be
great
for
this
team.
That
knows
more
about
containers
and
stuff
to
be
able
to
help
them
along.
G
B
D
D
D
So
this
the
my
anchor
for
housekeeping
things
is
really
not
very
exciting
and
I'm
sure
a
lot
of
people
are
welcome
to
drop
off
or
participate.
It's
something
that
we
we
used
to
do
more
often
as
we
were
approaching
release
times
for
any
one
of
the
releases
and
it
seemed
to
work
best
when
we
just
have
like
a
burn
down
list
of
issues
or
what
what's
what's
the
final
things
to
get
out
of
the
way
for
like
a
one
dot.
C
C
D
Literally,
giving
like
these
these
as
markers
of
prices,
that
everybody
on
the
call
please
take
a
minute
and
go
to
one
of
these
things
that
you
know
you
might
see
something
that
is
of
relevance,
and
this
is
kind
of
an
you
know.
This
is
not
a
observe
observers
activity
here.
This
is
like
an
invitation
to
get
involved
at
this
point,
no
spectators.
D
D
D
D
So
like
right
here,
nobody's
added
anything
yet
but
like
right
here,
the
runtime
specs
I
just
added
a
link
to
an
issue
comment
where
it
was
fairly
nicely.
Itemized
of
on
a
issue
of
like
please
tag,
a
new
release
that
here's
a
few
things
that
just
need
to
be
reviewed
and
or
pushed
through
and
he's
keeping
a
pretty
neat
list,
how
they're
keeping
up
being
a
list.
D
Would
be
good
to
get
into
a
respect
release
so
while
the
distribution
spec
is
ready
for
a10
I,
think
some
of
these
like
this
is
this-
is
we're
now
on
the
period
of
cleanup
of
like
what
are
these
things
that
either
need
to
be
just
closed
or
if
I
need
to
create
a
label
for
like
this
needs,
a
back
burner
label
like
post,
1.0
or
pre
1.0.
I
can
create
labels
right
now.
If
people
will
look
through
those
issues
lists
right
now
like
actively
on
this
call
and
comment
and
then
paste
a
link
to
it.
D
C
D
D
So
what
what
do
you
think
would
be
most
helpful,
like
maybe
V
next
to
say
what?
As
far
as
what
labels
yeah
like
because
right
now,
if
you
go
to
distributions
backslash
labels,
there's
only
like
eight,
but
none
of
them
are
kind
of
the
post.
One
auto
consider
like
conversation
for
post
one
dot.
Oh
yeah,.
C
D
There's
the
ones
that
we
know
well
leaders,
so
for
that
we
could
just
do
add
things
to
milestones.
Oh
I,
say
house,
we
have
two
milestones:
we
have
a
v1
Oh
next
and
we
have
d1
o
RC
one
which
is
I,
guess
kind
of
squishy
with
that's
that's.
That's
like
this
is
used
to
be
a
release,
consideration
candidate.
B
B
A
B
J
J
J
E
B
B
D
C
D
C
D
I
pasted
a
link
to
the
image
I
was
asked
to
reviewed
last
year,
sometime
yeah,
honestly
kind
of
waited
until
the
dust
settled
and
I
understand
the
use
case
and
understand
some
of
the
like
secure,
Enclave
computing
and
all
that
kind
of
stuff,
but
as
far
as
like,
what's
what's
needed,
as
far
as
like
special
accommodation
for
this
one
particular
you
say
a
use
case,
or
is
it
like?
You
just
mentioned
Phil
something
more
generic
for.
B
C
D
D
D
C
D
So,
okay,
so
alright,
yeah
right
so
there's
two
things:
one
is
it's
fundamentally
just
introducing
a
different
mime
type
that,
in
essence,
is
delivering
a
runnable
layer
that
container
images
a
container
and
tons
should
know
what
to
do
with
with
that.
But
the
truth
is
that
only
certain
specialized
and
newer
runtimes
would
even
be
able
to
recognize
these
these
mime
types
and
know
what
to
do
with
them
right
and.
D
And
so
in
in
essence,
it
is
like.
Yes,
it
is
kind
of
an
image
spec,
an
image
thing,
it's
delivering
a
container
image,
but
in
another
sense
it
is
very
much
so
like
what
the
artifacts
propose.
You
know,
artefacts
is
doing,
have
like
here's,
a
mime
types
and
here's
a
readme.
What
a
client
should
do
if
they
are
looking
for
or
want
to
handle
this
mind
type.
E
B
So
to
me
the
the
discussion
point
where
that's
the
eye
is
now
the
artifact
showed
up.
Do
we
need
this
for
requests,
changing
the
image
back
or
as
an
entry
and
the
artifacts
database?
That's
the
sufficient,
because
work
is
already
long.
You
know
nothing
game,
yeah
I
mean
krios
or
it's.
The
container
D
supports
that
they're
kubernetes
caps
is
getting
some
traction
so,
like
the
work,
continue
your
on
claims
and
all
that
stuff
yeah.
So.
B
C
Well,
so
what
I
was
working
on
as
you
to
have
these
well-known
types
that
you
would
submit
and
there
would
be
one
specifically
there
for
like
the
OCI
runnable
image,
so
in
that
one,
you
would
express
that
here
are
some
of
the
layers
that
you
would
support
or
and
much
like
this
other
conversation
we
had
there's
some
layers.
You
would
optionally
support.
D
D
E
D
D
Yeah
there
is
something
about
this
going
through
and
cleaning
house,
and
sometimes
like
discussions,
do
get
a
little
wild
and
sometimes
it
just
needs.
I
think
this
is
something
that
even
I
is
a
maintainer.
We'll
look
at
issues
sometimes
and
be
like
I
feel
bad
for
just
closing
some
of
these
things,
because
we
do
have
such
a
policy
for
open
discussion
and
if
it
doesn't,
if
somebody
doesn't
show
up
in
these
calls
to
discuss
it,
then
it's
like
I'm,
not
closing
it,
because
I
feel
like
these.
D
Things
need
to
have
a
little
bit
of
discussion,
but
it
by
the
same
token
there's
nobody
who's
standing
up
champion
championing
it
so.
Consequently,
some
of
these
issues
sit
open
for
a
long
time,
and
we
just
occasionally
need
to
prune
a
bit
because,
like
run
times
back
and
run
see
particularly
I've
got
a
lot
of
open
issues
and
a
lot
of
open
pull
requests
and
it's
an
exercise
that
is
really
best
if
people
just
open
up
and
just
start
clicking
through
the
issues.
D
When
we
do
this
and
it
helps
clean
house-
and
even
just
even
just
you,
you
know,
everybody
is
the
community.
Commenting
on
something
of
like
I
think
that's
gonna
be
closed
or
I.
Think
this
you
know
it's
still
a
valid
conversation
or
I
think
this
has
gone
completely.
Stale
helps
kind
of
sort.
Those
things
out
so
feel
free
to
do
that
anytime.
You
look
at
stuff
and.
D
D
D
D
Still
requires
two
other
maintainer
x'
to
lgt
Emmett,
so
sometimes,
as
as
the
maintainer
says,
we're
talking
about
things,
it's
harder
for
us
to
make
changes
than
the
community
like.
If
the
community
makes
a
PR,
then
it's
pretty
quick
to
two
fairly
active
maintainer
jump
on
it.
But
if
you
only
have
two
fairly
active,
maintainer
x'
and
one
of
them
makes
the
PR,
then
it
languages,
yeah.
C
C
C
D
C
Actually,
Vincent
you
still
there
I'm
looking
at
this
week,
because
we
actually
did
this
no
dissent
january
of
nineteen,
so
this
was
actually
after
it.
I
think
this
is
adding
some
content
to
the
image
spec
that
says
it
reference.
It's
like
one
of
the
types
of
artifacts
from
the
artifacts
back,
isn't
so
it's
kind
of
like
a
back
pointer
to
the
work
in
flight.