►
From YouTube: OCI Weekly Discussion - 2021-04-14
Description
Recording of the weekly OCI developer's call from 14 Apr 2021; agenda/notes here: https://hackmd.io/El8Dd2xrTlCaCG59ns5cwg?view#April-14-2021
B
A
A
What's
the
phil.
F
E
Just
because
the
comment
a
few
minutes
ago,
I
mean
it's
not
big
numbers,
but
I
do.
I
just
pulled
up
the
oci
video
uploads,
like
yep,
we're
averaging
10
or
more
views
for
a
lot
of
weeks
for
the
meetings,
which
is
not
a
lot,
but
that
means
somebody's
using
it
to
catch
up.
B
A
couple
of
folks
from
china
or
japan,
yeah.
A
C
C
I
think
five
minutes
is
a
pretty
good.
You
know
amount
of
time
to
wait.
I
I
have
the
first
kind
of
you
know
couple
agenda
items,
so
I'm
happy
to
kind
of
kick
things
off
on
on
that
front
if
you're,
if
everyone's
okay
with
that.
C
Cool,
so
the
the
first
thing
is
on
the
distribution
spec.
We
have
that
vote
out
for
vono
final.
Thank
you
so
much
josh
for
getting
that
done.
Just
a
heads
up
when
that
is
closed.
C
We
kind
of
have
a
small
process
thing
as
part
of
oci
to
notify
all
the
members
that
the
release
is
final
and
there's
kind
of
this
little
weird
ip
commitment
thing
where
they
basically
have
30
days
to
kind
of
potentially
back
out
of
the
oci,
if
they
so
choose
to
do
so,
based
on
the
ip
ipip
ipr
commitments
that
they
made.
So
this
is
just
a
reminder
and
just
kind
of
wide
notification
for
others
that
I
will
go.
Take
care
of
that
once
the
vote
is
finalized.
C
So
just
a
small
small
heads
up
there,
it's
been
a
while,
since
we've
had
to
do
it
so
on
the
maintainer
activity
and
audit
front
is
so
amy
and
I
put
together
a
basic
little
kind
of
you
know,
there's
many
ways
to
kind
of
gauge
activity,
because
you
know
there's
some
folks.
You
know
that
may
not
do
active,
commits
but
may
vote
on
things
as
they
arrive.
C
So
we
put
together
a
bit
of
a
spreadsheet
that
went
across
kind
of
the
different
projects
and
you
know
pulled
pull
information
and
kind
of
shared
it
with
everyone,
while
simultaneously
also
doing
a
call
for
you
know
active,
you
know,
maintainers
that
are
maybe
outside
of
the
you
know,
traditional
set
of
maintainers
that
we
had
to
see
if
anyone
wants
the
volunteer
to
step
up
and
good
news
is,
we've
had
a
lot
of
folks
that
have
actually
came
up,
especially
for
both
image
and
distribution
spec.
C
So,
on
my
end,
based
on
kind
of
the
way
our
process
works.
Is
you
know
I
collated
all
the
volunteers
for
at
least
distribution
and
image
spec?
You
know
image
spec
generally,
being
kind
of,
I
think
the
problem
child
of
of
all
those
where
the
other
specs
seem
to
have
a
lot
more
active
folks
based
when
we
looked
on
the
data
we
have.
I
think
we
have
about
a
handful
of
volunteers
for
both
the
image
and
distribution
spec.
C
The
way
our
process
works
in
oci
is
the
existing
set
of
maintainers
basically
have
to
go
vote.
Those
folks
in
and
the
idea
was
to
go
individually
for
all
the
folks
that
have
volunteered,
based
on
the
existing
maintainers
viewpoint,
get
the
votes
done
and
hopefully,
within
a
week
or
two
get
the
folks
that
get
approved
added
and
then
simultaneously.
C
I'm
also
asking
you
know,
folks
that
may
have
not
been
active
in
the
last
year
to
potentially
to
step
down,
but
I
prefer
to
do
that
after
you
know
the
initial
initial
call
for
adding
maintainers.
B
With
the
vote
with
the
existing
maintainers
voting
on
these
set
of
you
know,
nominated
or
however,
the
the
list
of
folks
that
are
stepping
up
is
this:
is
there
going
to
be
some
amount
of
activity
of
reaching
out
and
finding
those
paywall
existing.
C
Groups,
I've
already
reached
I've
already
reached
out
to
some
of
the
folks
that
are
inactive
right.
The
active
folks
have
stated
already
like
they're
interested,
and
you
know
some
were
like
hey
I'm
a
maintainer
from
trow
or
I
do
this.
You
know
hostess
registry
things,
so
it
it's
going
to
be
up
to
the
maintainers.
C
It
shouldn't
be
an
issue
for
distribution
spec
at
all
image.
Spec,
we'll
see
how
it
goes
if,
if,
if
not,
but
I
I
think
we
could
get
enough-
I
think
we'll
be
able
to
get
quorum
to
meet
the
minimum
minimum
vote.
If
not
we'll
we'll
may
remove
people
first
and
then
do
the
vote,
but
I
think
we'll
be
able
to
to
get
those
folks.
I've
been
reaching
out
to
some
of
the
enact.
I've
been
reaching
out
to
some
of
the
inactive
folks
already
and
getting
some
responses
in
summer.
B
C
So
yeah,
for
from
this
perspective,
if
everyone's
okay,
we'll
just
go,
you
know
essentially
implement
this
and
roll
this
out
and
you
know
start
start
start
the
voting.
You
know
if
there's
any
particular
concerns
about
this
approach.
Let
me
know
I
think
this
is
kind
of
the
fairest
way
to
do
it,
based
on
all
the
great
folks
that
kind
of
stepped
up.
C
G
Chris,
is
there
any
like
restrictions
on
people
having
too
many
people
from
one
organization.
C
Maintainers
generally
are
people
just
who
show
up
and
and
kind
of
do
the
work
where
tob
has
a
little
bit
more
kind
of
overall
oversight
for
the
organization,
and
we
limit
that
to
basically,
I
think,
two
max
for
two
people
from
the
same
company.
If
I
recall.
C
That,
before
official
or
after
after
basically,
if
you
look
at
the
link,
I
put
you
like
I've
done
this
for
all
the
other
ones.
C
I
have
to
basically
a
update
the
website
and
send
an
email
to
all
of
our
members,
giving
them
30
days
potentially
to
back
out
of
their
commitments
on
that,
if
they're,
if
they
had
some
issue
realistically,
no
one's
done
this,
but
this
is
just
part
of
the
roles
and
chartered
process
that
we
have
based
on
kind
of
the
initial
agreement
we
had,
where
everyone
basically
kind
of
agreed
to
not
sue
each
other
regarding
container
related
paths
in
the
beginning.
A
C
Yep
every
every
year,
someone
asks:
why
aren't
they
the
same
thing
and
I
try
to
remind
people
for
oci
all
the
cloud
providers
were
involved
and
happy
to
support
this
initiative
where
cncf
the
beginning,
not
so
much
so
different
different
reasons.
B
C
H
C
It's
a
back
office
thing
that
we
kind
of
do
and
it's
a
little
bit
different
than
it's
completely
actually
completely
different
than
say:
cnn
cf,
how
it
works
cool.
Let
me
go
through.
Let
me
get
the
agenda
back
up,
so
the
other,
so
that
that's
it.
So
if
everyone's
okay
on
the
on
the
voting,
we'll
essentially
kick
those
those
those
those
processes
off
over
the
next
week
or
so,
hopefully
it
won't
take
too
long
and
then
we'll
kind
of
go
from
there.
C
The
next
kind
of
thing,
I
I
think
I
I
pundited
you
to
phil
in
the
email
you
sent
out
today
on
kind
of
work
streams
and
working
groups
and
thoughts,
thoughts.
There.
E
Yeah,
so
hopefully
everyone
had
a
chance
to
to
see
the
email.
It's
not
a
ton
of
information,
but
there
is
a
link
in
there
to
a
different
hackmd
document.
That
already
has
a
few
comments
from
alexa,
who
must
have
stayed
up
late
his
time,
but
it's
effectively
a
draft
proposal
of
the
idea
we
talked
about
two
weeks
ago,
thanks
to
derek
for
writing
some
of
that
down.
E
So
obviously
it's
open
for
anyone
to
comment,
but
the
goal
would
be
that,
first
of
all,
it
re
kind
of
triggered
the
awareness
that
we
had
some
in-flight
charter
cleanup
that
we
started
last
spring,
that
we
all
ended
up
in
pandemic,
funk
and
never
finished
it.
I
guess
we
should
have
said
we
had
extra
time
to
work
on
it,
but
we
didn't
so
hopefully
we'll
get
that
cleaned
up.
E
But
in
alongside
that,
I
think
this
working
group
ideas
is
effectively
in
addition
to
that
charter,
which
means
we
would
want,
we
would
have
to
have
kind
of
a
tob
vote
once
we're
all
happy
with
the
language.
E
I
think
we've
all
decided,
there's
like
one
or
two
hours
of
the
day
that
may
that
we
could
actually
have
a
call
so
I'll
send
out
a
due
poll
for
a
couple
options
late
next
week
and
early
the
following
week
after
that
to
get
that
on
the
calendar
for
the
tob
to
actually
kind
of,
I
think
it'd
be
good.
If
it's
on
the
calendar,
then
it
it
forces
us
to
actually
do
the
review
work
between
now
and
then
and
kind
of
have
it
time.
Boxed.
E
H
Yeah,
I
think,
at
a
high
level
we're
just
trying
to
scale
some
of
these
different
conversations
and
make
sure
like
the
right
folks
and
interested
folks
are
involved
in
at
least
the
development
of
some
of
these
newer
specifications
or
some
of
these.
These
these
newer
features,
so
it
it
formalizes
that
a
little
bit
in
that
it
creates
a
working
group
that
the
tob
votes
on
to
establish
and
it
defines
what
that
working
group
is
trying
to
accomplish.
H
So
it's
not
completely
open-ended.
It
actually
has
an
end
goal
that
everybody
has
agreed
to
work
towards.
There's
a
there's,
a
certain
aspect
to
it
around,
like
including
projects
and
organizations
that
that
want
to
actually
work
on
an
early
implementation
of
it.
That's
one
thing:
various
people
on
oc
I've
been
concerned
about
with
newer
specifications
is
traditionally
we've
taken
working,
something
that's
already
working
and
improving
the
industry,
and
we've
we've
standardized
it.
H
So
there
wasn't
a
lot
of
debate
over
whether
or
not
this
was
useful
and
used
because
it
was
already
it
had
already
reached
that
level.
So
the
working
groups,
the
idea-
is
to
kind
of
follow
along
almost
what
we
saw
with
like
the
http
http
process.
You
see,
there's
like
a
working
group,
they
start
developing
drafts
and
then
those
get
voted
on
and
not
that
it's
the
best
example
of
like
working,
but
at
least
like
that
that
flow.
H
A
I
just
said
it
was
good
to
say
I
was
it
kind
of
really
fostered
people
that
wanted
to
work
on
stuff
like
we've
had.
Obviously
we
have
the
recent
stuff
we've
been
talking
about,
but
even
the
things
like
the
multi-arc
work
that
you
know
phil
was,
you
know
we
did
some
work
around
and
some
others
were
interested.
There
wasn't
any
real
obvious
ways
to
what
the
next
steps
was.
So
you
don't
have
to
be
a
maintainer.
You
don't
have
to
be
a
to
b.
A
H
F
C
Yeah,
I
think
we
would
avoid.
We
basically
would
treat
them
as
draft
specs.
They
would
not
be
subject
to
the
ipr
unless
they
were
final
and
it
would
be
tob
would
have
to
bless
bless.
That,
I
think,
is
how
we'd
go
about
it.
H
H
A
H
Yeah,
I
think
it's
it's
pretty
it's
pretty
early
on,
like
it
started
off
as
just
kind
of
notes
from
last
week,
and
then
I
tried
to
use
language
that
corresponds
more
with
like
the
existing
charter
and
and
and
lay
it
out
like
that.
But
it's
it's
it's
so
pretty
pretty
early
on
so
like
in
any
level
of
feedback
is
fine.
Like
sure
you
can,
you
can
nitpick
the
wording,
but
also
just
like
the
high
level.
What
it's
trying
to
accomplish,
I
think
is
this
is
fair.
E
And
that's,
I
think
the
goal
is
that
you
know
we
have
these
in-flight
charter
cleanups
as
apr
that
derek
asks
alexa
to
split
into
separate
prs,
which
he
said
he's
going
to
do
this
week.
Then
this
would
be
you
know
effectively
a
pr
when,
when
we
kind
of
are
all
close
to
feeling
like
the
language
is,
is
there
that
you
know
final
kind
of
edits
or
comments
could
happen
the
normal
pr
process?
At
that
point,.
E
C
Yeah
any
other
agenda
items
while
we
have
everyone,
I
I.
I
Have
a
question
about
working
groups?
Sorry,
I
was
trying
to
read
and
listen
and
failed
at
both.
I
think
is
this:
the
a
required
mechanism
for
changes
to
specifications,
or
is
this
just
a
new
way
of
organizing
work,
a
ladder.
H
Okay,
my
thought
is
that
this
would
apply
for
would
say
like
major
revisions.
We
talked
about
like
image,
spec
b2.
I
think
it's
better,
that
it
would
go
through
this
process
than
just
like.
Have
the
image
spec
v1
existing
maintainers
try
to
like
all
those
like
smaller
iterations.
I
don't
think
it
makes
sense
to
go
through
this
process,
but
it's
like
what
everything
we
talked
about
like
in
oci
v2
or
amspec
b2.
It's
it's
not
that
related
to
v1,
really
in
many
of
those
cases.
H
I
I
can
imagine
anywhere
where
intentionally
making
a
breaking
change
would
require
a
lot
of
coordination
amongst
stakeholders
and
agreement
that
it
is
a
good
thing
to
change
and
that
we'll
all
implement
things
and
work
around
it.
But
for
you
know,
non-breaking
additive
changes
yeah.
I
think
the
current
process
works
pretty
well
as
soon
as
we
get
some
maintainers.
A
I
don't
know
if
maybe
this
is
part
of
what
we
can
do
with
maintainers,
but
I
think
that's
one
of
the
questions
that
we
have
is
what
does
backwards
compatibility
define
when
we're
adding-
and
this
has
been
the
root
of
the
conversation
we
would
have
it
recently
is
what
does
adding
behavior
mean
and
just
because
it's
optional
doesn't
mean,
there's
an
expectation
on
it,
so
I
think
trying
to
figure
out
like
then
the
obvious
example
is
adding
an
annotation
right.
Annotations
by
definition,
are
just
strings
they're,
you
know,
there's
really.
A
The
only
point
of
the
spec
is
here's
a
well-known
name
for
it
other
than
that
registries.
Don't
really
care
and
the
clients
necessarily
don't
even
care,
adding
behavioral
things
like,
I
think,
that's
the
question
we've
been
struggling
with.
What
exactly
does
it
mean
to
add
something
to
a
released
version
and
having
some
structure
around
that
will
really
help,
because
if
somebody
says
they
adhere
to
version
one
of
a
spec,
but
something
was
added
to
that
after
what
is
what
does
that
mean
for
a
conformance
to
that.
A
B
B
That's
kind
of
why
the
the
use
of
optional
should
or
whatever
is
that
they
don't
implement
it,
that
you
can
still
get
basic
functionality
and,
like
all
the
conformance
tests
that
josh
and
other
others
worked
on
is
like
you
could
still
pull.
You
could
still
push
and
whatever
like
basic
categories
of
workflows,
but
in
utilizing
some
some
nuanced
behavior
is
completely
optional.
H
I
think
the
z
standard
one
was
a
good
example
of
like
what
are
what's
the
expectations
here
in
terms
of
compatibility,
I
don't
think
we
could
ever
expect
like
if
new
media
types
and
stuff
are
introduced,
that
older
clients
would
necessarily
understand
them,
but
we
need
to
make
sure
that,
like,
if
there's
a
set
of
like
everything
we've
defined
like
if,
if
you
come
across,
anything
that's
outside
of
what
you
know
to
understand
like
failing,
should
happen
gracefully
and
for
other
things
like
I,
I
think
the
the
data
field.
H
A
Types
I
mean,
I
think
this
is
the
the
coupling
of
the
distribution
which
you
know
turns
into
registries
and
image
spec,
which
you
know
has
been
used
for
multiple
things
as
well,
and
I'm
going
to
try
to
solve
it
here.
I
think
this
is
just
one
of
the.
The
conversations
for
working
group
is
to
figure
out.
What
does
that
overlap
mean,
because
adding
something
to
one
actually
impacts
both
and
what
could
be
ignored
on
a
client
may
not
necessarily
can
be
ignored.
A
You
know
in
a
registry-
and
I
think
that's
the
the
nuance
that
we
have
to
just
sort
through
what
that
means,
because
I
think
we
all
agree.
We
want
some
of
these
behaviors
we're
just
trying
to
figure
out
how
to
have
it
done.
Besides,
when
there's
a
conversation
about
how
the
implementation
might
be,
but
I
put
that
aside
in
either
case,
there's
implications
that
I
don't
know
if
we've
thought
through
all
the
way
to
understand
what
does
it
mean
to
add
and
what
is
the
expectations,
customers
and
users
would
have
on
those.
A
So
just
I
think
that's
the
piece
that,
when
they're
tangled
together
the
way
they
are,
and
I've
heard
that
run
c
and
the
image
spec
or
tag
were
or
still
are
to
some
extent
tangled
and
somehow
that
got
detangled
having
that
structure,
I
think,
will
help
us
be
clear,
like
yeah.
This
makes
perfect
sense.
We
can
add
this
here
and
here's
the
semantics
around
it.
Oh,
this
is
has
this
kind
of
behavior,
so
we
need
to
treat
it
differently
in
some
way.
B
I
I
think,
honestly,
a
lot
of
that's
already
there
even
like
derek
said,
with
the
z
standard,
like
expecting
old
clients
to
still
be
able
to
behave
with
it
properly.
That's
kind
of
like
one
of
the
figuring
out
points
is
that
if
you
wanted
to
enable
some
new,
you
know
either
hashing
hashing
algorithm,
that's
not
sha,
256
or
compression.
B
That's
not
gzip
that
the
old
clients
would
gracefully
fail
because
they
don't
know
what
to
do
with
it
and
that's
expected
behavior,
but
just
figuring
out
what
what
is
an
acceptable
level
of
that
is
kind
of
the
industry
conversation.
That's
the
ongoing
piece
of
it.
I
Yeah
I
just
put
in
the
hackmd
a
couple
links
to
similar
ideas.
I
think
it's
a
very
hard
problem
and
I
don't
think
it's
easy
to
even
describe
what
a
breaking
change
is,
because
it
relies
on
historical
context
and
all
kinds
of
things
that
are
impossible
to
really
say
without
you
know,
having
a
set
of
maintainers
that
have
been
around
for
years
and
no
no
behaviors
exist
in
the
wild.
So
I
linked
to
the
kubernetes
api
change
guidelines
and
the
golang
1.0
compatibility
guarantee.
H
I
J
I
mean
I
think,
that
the
go
example
is
interesting
because
they've
broken
the
run
time,
a
number
of
times
for
backwards,
compatible
changes
and
like,
for
example,
locking
os
threads
is,
is
one
preemption
is
another
one,
but
the
source
code
has
always
been
compatible,
and
I
think
that
separating
out
the
image
and
runtime
has
some
value
in
that,
but
then
they
also
have
this
whole
thing
of
like
you,
you
can
ask
for
certain
features
in
the
runtime
or
you
know
at
compile
a
time.
J
A
I
mean
like
the
impression
is
that
if
you
think
about
that,
when,
like
from
a
a
registry
and
client
like
and
the
compression
stuff
registry,
don't
care
just
a
blob
right,
so
just
send
it
on
in
and
don't
care
if
it's
in
a
manifest
as
long
as
we
can
track,
it
know
how
to
clean
up
afterwards
life's
good,
but
the
clients
could
get
totally
hooked
if
they
don't
know
how
to
deal
with
that.
Where
are
the
things
you
know?
A
We've
talked
about
the
references
lately
like
sure
I
might
be
able
to
push
it
in,
but
does
it
wind
up
being
these?
You
know
zombie
pieces
that
are
left
around,
that
we
don't
know
how
to
clean
up.
So
I
think
there's
just
a
couple:
it's
that
cross
issue
that
we
need
to
figure
out
is
how
do
we
have
capability
in
one
and
how
does
it
affect
the
other?
There's
lots
of
ideas
like
the
the
phone
platforms
have
figured
this
out.
They
don't
version
phones
for
everything
and
every
phone
has
different
capabilities.
A
So
there's
a
capabilities
check
right.
So
at
run
time
you
can
say:
does
the
phone
support
a
camera?
Does?
What's
the
depth
of
the
camera,
does
it
have
a
front
camera,
a
rear,
camera,
a
microphone
so
on
and
so
forth?
So
there's
lots
of
ways
to
kind
of
do
lots
of
capability
checks,
but
those
were
built
in
as
part
of
the
the
phone
standards,
the
specs
for
the
various
phone
platforms.
So
maybe
that's
some
of
the
flexibility
that
we
can
start
thinking
about.
A
I
guess
to
your
point:
phil
and
derek:
it's
like
getting
getting
the
working
group,
piece,
kind
of
discussed
and
merged
doesn't
really
stop
working
groups
from
starting
to
form
and
then
that
kind
of
helped
figure
out
like
hey.
Does
this
model
actually
work
because
we're
actually
trying
to
test
it?
We've
actually
done
this
a
couple
times
right,
the
you
could
argue
the
multi-arc
stuff
didn't
turn
in
turn
out
very
well
like
we
never
really
got
traction
on
it
because
we
discussed
it
a
couple
times
and
I
think
everybody
was
hoping.
A
Somebody
else
would
do
something.
You
know
the
artifact
stuff
we
did.
You
know
continue
to
work
at
tickets
over
a
year
to
finally
get
kind
of
settled
on
which
direction,
and
there
was
a
working
group
that
formed
around
that,
and
then
there
was
a
culmination
of
it.
So
that's
what
I
liked,
because
I
wasn't
one
of
the
maintainers.
It
wasn't
anything
to
the
group
at
first,
but
I
was
able
to
come
in
and
find
people
that
were
able
to
help
me
figure
out
how
to
scope
that.
A
So
if
we
can
do
that,
for
others
to
bring
in
new
features,
that'd
be
great
as
well.
Okay,
I
know
that
chris,
who
was
at
the
one
of
the
was
it
the
alibaba
folks.
They
were
trying
to
do
stuff
around
scanners,
and
you
know
they
had
some
ideas
and
you
know
kind
of
gave
them
a
list
of
stuff
with
some
of
the
things
we've
been
struggling
with.
E
E
G
I
just
I
just
added
at
the
last
minute
here
yeah,
so
in
part
of
getting
the
distribution
spec
out
the
door.
There
was
some
vincent
had
some
old
broken
image
on
docker
hub.
Oh.
G
So
we
had
to
find
we
had
to
find
a
home
for
it.
So
there's
this
repo
that
is
now
under
the
organization
that
has
his
source
code
for
that
image
updated.
G
So
that's
pretty
much
it
if,
if
any
other,
it's
mostly
for
use
by
like
the
ci
and
the
different
projects,
if
there's
any
images
that
you're
using
there
that
are
customized
in
any
way
put
it
there.
I
also
one
of
the
images
for
go.
Linting
is
actually
not
custom,
but
it's
mirrored
from
docker
hub.
So
github
is
not
it
hubs.
Registry
is
not
rate
limiting,
as
of
yet
so.
G
I
I
pretty
much
want
to
stay
out
of
that
conversation
they're
in
github
right
now,
because
that's
we
had
a
bot
to
get
to
github,
but
I
don't
see,
I
don't
necessarily
see
the
value
in
it,
but
I
don't
see
any
reason
why
we
can't
mirror
that
in
quay
doctor
hub
or
clouds
public
registries,
and
then
I
think
steven
day
actually
had
the
idea
what
if
we
actually
hosted
our
own
registry
at
like
opencontainers.org,
which
I
thought
was
an
interesting
idea.
G
I
don't
know
if
that's
like
a
white
list
of
one
of
the
clouds
donating
some
space
or
docker
hub.
I
just
wanted
to
keep
people
posted
that
that
exists.
No.
B
I
think
that's
great
and
it's
it's
kind
of
it's
funny-
that
one
of
the
other
conversations
that
from
the
past,
that
we
never
slowed
down
really
to
to
do
that
with
was
even
talking
about
like
image
image
format,
like
conformance
and
even
like,
with
the
conformance
tools
that
for
the
distribution,
spec
and
otherwise
that
you
could
just
like
pull
an
image
that
could
run
against
something
rather
than
having
to
go,
get
it
or
build
it
or
otherwise.
B
Just
having
those
kind
of
images
built
every
every
commit
to
you
know
to
the
trunk,
I
think
it's
actually
pretty
useful
and
interesting
to
have
it.
G
Yeah
it'd
be
cool
to
kind
of
dog
food.
Our
stuff
too,
like
it's,
it's
just
using
docker
and,
as
we
all
know
like
docker,
is
its
own
tool
with
its
own
stuff
and
it
would
be
cool
if,
like,
we
could
really
do
our
like
really
use
the
open
standards
around
all
the
different
specs,
and
maybe
we
can
yeah
we
can
like
maybe
host
a
registry.
I
don't
know
the
overhead
of
that,
but.
G
But
yeah
we
just
we
just
needed
somewhere
for
this
image,
so
it's
there
right
now
and
it's
not
really
like
an
official
project,
but
if
people
want
to
help
maintain
that
or
add
their
images,
just
let
me
know.
G
For
distribution
spec,
it's
two
there's
one
for
the
pandoc
project,
which
will
convert
the
spec
and
mark
down
to
pdf
and
html,
and
I
think
I'm
guessing
that's
also
on
image,
spec
and
runtime
spec
somewhere.
If
vince
was
involved,
and
then
we
also
have
something
that
links
go
source
files,
that's
that's
not
modified
that
we
were
just
our
travis
was
failing
because
docker
hub
was
was
rate.
G
Limiting
us
like
every
other
time,
and
I
know
justin
had
said
we
could
get
that
like
kind
of
on
an
allow
list,
but
it
probably.
A
G
B
Yeah,
I
don't
think
anybody
in
the
world
used
that
vbats
pandock
image,
except
for
the
oci,
build
process
or
release
process
and
I'm
completely
okay,
not
putting
it
on
clay,
because
I'd
created
that
board
to
put
it
in
a
while
ago.
But
that's
it
just
always
seemed
too
complicated
to
have
yet
another
org
for
folks,
like
chris
and
amy
to
participate
and
manage
so
having
it.
All
under
github
seems
nicer
in.
A
General,
I
mean
I'm
happy
to
offer
a
custom
domain.
You
know
registry,
if
we
want
one,
it
just
seems
like
having
one
alongside
the
source
with
github
that
makes
sense
like
it
doesn't
preclude
it
being
on
quay
and
dockerhub
and
others.
You
know
as
we
in
ecr
and
so
forth,
but
there
is
overhead
and
just
keeping
it
simple
with
the
github
integration
seems
like
a
logical.
A
G
Yeah
I
it's
really
I
mean
all
of
this
is
not
worth
more
than
five
minutes
of
of
mentioning.
Maybe
it's
something
like
going
forward
with
this
type
of
thing.
It's
something
if
we
do
create
a
working
group
around,
I
don't
know,
maybe
there's
a
testing
conformance
working
group,
and
we
can
talk
about
interesting
things
to
do
around
this.
B
F
Yep
just
some
formalities.
There.
C
Yep,
cool
yeah
I'll
send
a
note
out
kind
of
the
list
to
remind
folks
and
then
we'll
just
kind
of
kick
those
off
and
and
see
where
things
land
and
we'll
go
from
there.