►
From YouTube: Carvel Community Meeting - May 24, 2021
Description
Carvel Community Meeting - May 24, 2021
We meet every Monday at 11:30am PT / 2:30pm ET. We'd love for you to join us live!
Full details and notes from the meeting here: https://hackmd.io/F7g3RT2hR3OcIh-Iznk2hw?view#May-24-2021-Agenda
A
All
right,
hi
everyone
welcome
to
this
week's
edition
of
the
carmel
community
meeting.
Today's
date
is
may
24th
2021
just
a
reminder.
Please
read
and
abide
by
our
code
of
conduct,
which
is
listed
out
here
in
the
agenda.
While
you
are
attending
these
meetings,
this
is
being
recorded
and
will
be
uploaded
to
our
youtube
playlist.
A
A
Pacific
time,
2
30
p.m,
eastern
time
and
that's
just
an
opportunity
for
the
maintainers
to
go
even
more
in
depth
with
any
discussion
items
that
you
have
or
anything
you
need
help
with
from
the
carnival
tool
suite
when
you
are
attending
these
meetings.
We
do
ask
that
you
put
in
your
name
and
any
organization
that
you're
representing
so
that
we
can
keep
track
of
who
from
the
community
is
joining
us
live
and
we
can
keep
in
in
touch
with
you.
All.
A
B
C
Yeah,
just
top
line
we've
got
these
next
two
items
that
pr's
are
sorry,
this
item,
that
pr
is
eminent.
So
like
it's
there's
just
like
one
final
review,
we
think,
and
that
will
be
we'll
be
merging
that
in.
D
Yeah,
adding
a
spec
for
schemas
and
then
also
sort
of
an
introduction
document
currently
in
progress.
B
Awesome
anything
else
on
schemas
from
the
past
week
or
so
cool
up
to
the
next
one.
Then
cap
controller
that
package
manager
api
work
eli
want
to
provide
any
updates
from
the
past
week.
E
D
F
Said
last
week
we
released
version
0,
8
0
of
image
package
and
basically
this
this
new
version
contains
it
was
released
because
it
contained
two
bug
fixes
one
that
was
related
to
a
copy
of
a
bundle
when
the
bundle
contained
images
like
if
the
bundle
was
not
copied
to
a
place,
and
there
were
images
in
in
like
in
a
pro
private
and
a
public
repository.
F
So
that
was
fixed
on
this
on
this
version
on
this
0
8
0
version
and
another
issue
that
we
that
we
solved
was
related
to
that
that
we
only
saw
in
arbor
where,
if
you
did
a
head
request
to
try
to
find
out
if
an
image
was
present
or
not,
arbor
was
just
failing
and
just
getting
us
like
a
401
as
a
reply,
and
so
now
we
fixed
it
and
the
way
we
fixed
it.
F
We
do
a
head
request
and
if,
for
some
reason
the
head
request
fails
to
see
if
the
image
is
there,
we
just
do
again.
So
that
has
some
problems.
Some.
We
are
incurring
in
some
in
some
time
penalties
here,
because
the
get
request
takes
much
more
time
than
I
had,
but
at
the
same
time
we're
still
keeping
the
head
request,
because
that
will
allow
us
not
to
to
get
into
problems
with
docker
hub
and
so
on,
because
we
can
just
check
if
the
image
is
there
without
game
being
rate
limited.
F
So
that's
those
are
like
the
two
big
issues
that
were
bugs
that
were
fixed.
Also,
there
were
two
deprecations
that
happened
one
I
talked
about
it
last
week.
That
is
now.
We
preserve
the
execution
bit
on
any
file
that
you
have
in
your
bundle
and
so
that
that
creates
a
breaking
change,
because
if
you
try
to
create
a
bundle
with
version,
070
or
prior
you'll
get
a
shot
from
that
bundle.
But
if
you
try
to
do
exactly
the
same
bundle,
but
with
0
8
0
you'll
get
a
different
cha.
F
The
only
difference
is
that
there's
a
bit
now
that
is
saved
the
other
one
is
that
we
changed
the
format
of
the
default
tag.
So
if
you
were
relying
on
that,
be
aware
that
we
changed
that
and
then
the
rest
just
added
some
ux
improvements,
and
you
would
like
to
call
out
carlesia
that
she
made
one
of
the
changes
there
on
on
our
on
our
change
log
there
so
and
are
pretty
good.
Just
pretty
good.
Thank
you
carlesi
for
your
work
and
we
hope
to
see
more
people
helping
us
out
on
these
releases.
F
I
can
so
this
one.
This
one
was
so
like
the
two
new.
The
two
features
that
are
there
have
been
developed
for
some
time,
so
we
cut
a
new
release
because
of
that
bug
that
if
you
were,
if
you
had
like
your
configuration
for,
if
you
had
your
configuration,
that
you
passed
to
k,
build
and
for
some
reason
in
that
configuration,
the
word
image
was
present
and
you
were
trying
to
replace
an
image
somewhere,
and
it
was
underneath
that
key
in
yaml
it
would
never
find
it.
F
So
it
would
find
the
key
image
and
it
would
stop
searching
down.
So
that
was
a
bug.
It
was
like
just
somewhat
of
like
an
important
bug,
because
some
some
people
could
have
like
their
configuration
where
they
have
like
image
and
then
the
name
of
the
image
that
they
want
to
replace
and
then
the
image
so
like
a
list
of
images
and
that
could
hinder
their
their
yammo.
So
this
is
a
very
big
change
that
we
did
here.
F
Another
thing
we
started
deprecating
everything
that
is
related,
that
of
the
features
that
were
moved
from
k
building
to
image
package.
So
now
they
have
like
an
application
warning.
It
just
tells
you
like.
Oh,
this
is
not
going
to
be
here
and
for
much
longer.
So
if
you
want
to
use
them
just
start
using
image
package-
and
I
believe
that
was
it.
B
C
Yeah
we
have
an
initial
round
of
internal
exploratory
testing,
which
yielded
out
a
couple
items.
We
still
plan
on
getting
some
usability
testing
into
as
we're
getting
to
the
point
where
we're
gonna
lift
the
experimental
flag,
but
there's
a
few
things.
We
we
already
see
right
now
that
we
want
to
make
improvements,
quick
question
folks,
do
we
already
start
that
first
story.
C
Just
a
quick
peek
at
that
just
to
let
folks
know
what
that's
about
it's
that
today,
if
a
value
is
being
contributed
to
a
data
value
through
an
overlay
and
it's
and
it's
a
starlark
value
that
is
here.
This
is
a
starlark
expression
that
equals
four
that
that
that
the
starlight
value
actually
gets
inserted
in
that
comes
up
as
a
64-bit
integer,
rather
than
a
general
integer.
So
this
is
about
making
those
two
equivalent.
C
Seems
like
it
might
be
a
little
bit
of
a
weird
use
case,
so
actually
there's
here's
a
here's.
What
came
from
the
exploratory
testing
there's
that
example
here
where
there's
a
existing
project?
That's!
Oh!
No,
that
that's
what
the
output
is.
Currently
that's.
The
context
section
is
describing
hey
here's
an
existing
user
of
ytt,
where
this
would
be
a
blocker
for
them
to
migrate
to
schema.
So
that's
why
it's
important
for
us
to
do
it.
Yeah.
B
C
So
the
overall
symptom
here
is
straightforward.
If
you
scroll
down
to
the
undesired
result,
real
quick,
we'll
come
back
up
but
down
here.
What
you'll
notice
is
that
there's
no
found
in
this
type
mismatch?
We
would
like
to
show
what
type
was
found
to
be
able
to
at
least
compare
to
what
what
was
expected
by
the
schema
this
one's
particularly
gnarly,
because
it
comes
from
a
location.
It
comes
from
a
piece
of
code
that
doesn't
have
a
known
location,
so
you
don't
even
know
the
type.
C
Nor
do
you
know
where
it's
coming
from
that
just
made
it
worse.
C
So
what
we'd
like
to
do
is
at
least
show
the
type,
even
if
we
can't
show
the
exact
location
right
now,
so
it
is
out
of
scope
to
try
and
like
somehow
resolve
providing
some
direct
output
here
for
the
for,
like
the
source
line.
That's
contributing
this!
That's
actually
quite
difficult
to
do
at
this
point.
C
So
we're
just
just
for
this
story.
We're
just
saying
hey,
let's
at
least
include
the
found
that
that
should
be
present.
C
C
So
careful
this
may
feel
like
like
a
really
very
pinpoint
type
thing:
don't
don't
be
fooled
by
the
precision
of
the
way
the
store
is
pointing
at
what
is
probably
the
source.
Please
consider
like
what
why
why
was
that?
If
statement
in
there
like
is
there
where
there
will
be
ramifications
of
removing
it,
hopefully,
if
you
did
a
test
would
fail,
but
it
is
definitely
something
to
for
us
to
like
make
sure
we
consider
we're
going
to
inadvertently
create
a
regression.
So
just
keep
that
in
mind,
please.
C
G
When
we
pick
up
the
story
similar
to
the
story
we
pointed
above,
we
have
to
work
up
should
do
we
have
to
work
off
the
overlay
schemas
branch?
Is
it
dependent
on
any
of
those
changes,
or
can
we
actually
work
off
develop
for
this?
I
don't.
C
F
C
F
I
just
pointed
it
to
like
this.
This
looks
like
a
pretty
simple
removal
of
an
if
statement,
but
I
think,
as
john
said,
I
gave
it
one
more
point
just
because
I
have
no
not
sure
why
the
if
statement
was
there-
and
I
don't
know
if
that
will
cause
any
problem
if
we
just
remove
it
so
there's
like
uncertainty
on
that
part,
we
might
have
to
do
something
more
like
we
might
have
to
fix
that
question
mark
problem.
In
order
to
solve
this
story,
I
don't
know
so.
C
C
C
C
C
Okay,
so
the
intent
here's
there
are
four
messages,
not
some
open-ended
thing
about
n
number
of
messages
and
that
there
is
like
real
intention
behind
the
shift.
We're
not
just
like
moving
things
around
in
order
to
sort
of
seemingly
solve
a
rubik's
cube
randomly
like
there
there's
intention
here
so
the
details.
C
So
if
we
scroll
down
just
a
hair,
so
this
kind
of
sets
up
trying
to
set
up
the
motivation.
What's
the
gap
here,
so
we're
inspired
by
another
project,
the
rust
project
that
has
been
doing
lots
of
work
for
some
time
on
improving
the
how
approachable
their
error
messages
are.
They've
done.
Some
of
that
work
very
publicly.
There's
a
link
to
that
article.
C
You
can
go
look
at
that
and
and
and
see
a
whole
bunch
of
art
that
has
been
stolen
in
quotes
and-
and
then
here
are
some
to
when,
when
you
go
to
pick
this
up,
some
very
specific
factors
that
went
into
the
the
formatting
of
these
messages.
This
is
not
to
say
that
what's
present
is
sacred.
This
is
to
say,
like
these
are
the
reasons
for
these
shifts
in
in
movements
trying
to
convey
the
purpose
and
the?
C
Why
we're
doing
what
we're
doing,
if
you
find
that,
like
there's
a
there's
another
adjustment
to
make
like
you're,
the
implementer
think
through
like
how
that's
going
to
improve
the
ux,
and
if
you
see
that,
like
that's,
the
point
of
the
story,
is
to
improve
the
ux.
So
this
is
trying
to
help
like
give
some
of
that
mindset
like
help
load
up
some
of
that
mindset.
C
So
if
you're
in
there
and
you
see
something
even
better
like,
I
encourage
you
to
use
your
creativity
in
your
in
your
aspiring
designer
eye
to
try
and
make
it
better.
If
that
is
true
and
feasible,.
C
C
C
So
here's
here's
an
example
of
where
we
have
invalid
schema.
This
is
our
current
error
message:
format
you
can
see
where
yelling
at
the
user,
the
hints
but
up
against,
there's
no
white
space
here
and
it's
not
that
anybody
ought
to
be
confused
but
we're
using
the
same
delimiter
to
show
contents
of
a
file.
C
There's
your
line
number
see.
The
commentary
has
a
different
delimiter,
hoping
that
it's
doing
the
same
thing
to
your
eye,
where
that
reads
a
little
bit
more
like
this
is
stuff,
that's
not
in
the
file,
but
more
about
comments.
Notice.
The
there's
actually
like
a
series
of
hints
in
this
particular
case
each
one
having
it's
sort
of
like
a
little
bit
of
a
list.
C
When
dennis
looked
at
this,
he
had
a
great
suggestion
of
you
know.
If
you
put
in
new
lines,
you
can
split
on
that,
maybe
format.
So
that's
an
example:
let's
go
to
the
next
one
and
there's
going
to
be
four
total
here
same
story,
less
yelling,
the
the
message.
Also,
the
wording
here
is
a
little
bit
more
about
what
is
it
that
hopefully
could
stick
in
your
mind
about
what
was
exactly
not
expected
without
repeating
exactly
what
you
see
down
here
in
the
expectation
some
of
the
stuff
was
great.
C
We
kept
a
lot
of
the
the
elements
here.
Those
were
fine
again
suggesting
that
if
we
had
space
for
having
multiple
hints,
you
can
actually
sort
of
cover
multiple
situations
as
long
as
they're
worded,
in
a
way
that
doesn't
tell
the
user
what
to
do
but
then
says
instead
sort
of
either
explicitly
or
implies.
Well,
if
this
is
what
you
were
trying
to
accomplish,
this
other
approach
might
be
better,
is
sort
of
the
is
the
intent
there.
C
G
Now
I
see
the
last
two:
those
are
the
the
type
mismatch
and
undefined
map
key
like
the
titles
in
the
after
block,
like
so
in
scenario.
Three
after
it's
like
a
ytt
error,
value
is
of
wrong
type
and
then
scenario,
four,
it's
an
undefined
map
key.
But
here
in
the
title,
it's
like
it's
talking
about
overlaying
again
like
in
scenario.
Three,
like
the
original
message,
was
about
an
overlaying
message
error.
I
thought
like
if
you.
G
C
A
typo,
that's
that's
an
inadvertent
omission.
Okay,
the
the
everything
up
to
the
error
that
could
catch
up
up
to
the
the
scheme
of
error
message
should
be
identical,
left
alone,
cool
yeah
and
the
intent
here
like.
So
as
we
were
going
through
this,
and
in
doing
this
design,
the
intent
was
to
not
expand,
expand
any
need
for
significant,
extremely
feasible,
like
if
there's
anything
in
here
that
that
isn't
data
that's
already
sitting
there.
That
is
not
matching
up
with
what
we're
hoping
with
this
story.
C
We're
hoping
that
we
take
the
information
we
have
available
to
us
and
maybe,
like
I
said,
would
like
the
hints,
maybe
a
little
bit
of
play
with
that.
Maybe
you
decide
as
implementary
like
no
I'll
go
ahead
and
make
that
a
slice,
or
maybe
you
decide
to
interpret
new
lines
as
as
a
prefix
on
his
whatever,
but
not
nothing
any
more
than
that
was
intended.
If
it
gets
any
gnarlier
than
that,
then
then
we
missed
something.
We
inadvertently
asked
for
something
that's
less
feasible
than
we
intended.
F
Ux
wise
I,
when
I'm
looking
to
that
error
message,
I
am
waiting
to
see
the
contents
of
line
two
and
four
as
well.
It
must
be
just
me
and
my
div
head
because
that's
what
I
expect
to
see
where
you
have
the
pipes
without
anything,
so
I
don't
know,
I
don't
know
if
that
might
so.
Like
my
the
first
reaction
when
I
saw
this
was
like
oh
interesting,
the
file
has
nothing
on
line
two
or
line
five.
That
was
like
my
first
thought,
but
not
sure
if
this
like
is
helpful
or
not,.
C
Yep
there's
a
lot
of
stuff
in
here
that's
subjective,
and
that
means
that
it
lands
differently
for
different
people.
I'd
encourage
you
to
like
play
with
that
so
like
if
there's
something
that
that
can
be
done.
That,
like
would
improve
that
I'd,
say
that's
exactly
what
I
was
talking
about
when
I
said:
don't
take
these
as
sacred
per
se.
C
I
did
I'll
say
this:
I
did
play
around
with
trying
to
have
more
more
lines
in
there
and
it
actually
gets
it.
So
a
it's!
It's
not
that
it's
infeasible,
but
it's
quite
hard
because
you
have
to
actually
probably
access
your
parent.
You
have
to
know
whether
or
not
the
meta
that
you're
looking
at
is
on
that
line,
or
you
have
to
somehow
pipe
in
original
source,
which
we
do
not
have
at
that
moment.
So
there.
F
Yeah
I
I
was
not
saying
that
we
need
to
have
it
there.
I'm
saying
that
the
way
it
is
displayed
is
telling
me
that
it
is
going
to
display
me
line
two
and
five,
not
saying
that
this
should
be
there.
I'm
just
saying
like
the
fact
that
you
have
like
the
they
are
all
the
same.
I
would
expect
more
lines
to
see
to
expect
to
see
more
lines,
I'm
not
saying
that
it
should
show
more
lights.
Gotcha.
D
The
complexity
of
implementing
these
changes
in
the
error
message
seems
rather
low
to
me,
but
thinking
about
the
topics
that
are
in
the
summary
there
like,
how
can
this
be
better
like
the
thing
that
joao
mentioned,
those
things,
I
think,
are
more
complex.
That's
why
I
added
a
point
to
my
pointing.
H
And
just
to
add
on
to
that
working
with
error
messages
in
the
past,
there
can
be
a
lot
of
like
nuanced
ways
that
the
error
message
is
communicating
to
the
user
and
to
try
to
tease
those
out
and
try
to
get
it
just
right.
Just
perfect,
almost
from
like,
with
all
all
your
current
context
of
what
feels
like
what
should
be
perfect
can
take
a
while
to
get
there,
and
so
I
want
to
give
the
space
for
that.
If
I
was
the
one
working
on
that,
I
would
want
this
kind
of
space.
F
As
a
suggestion
like
I
suggest
to
implement,
as
is
or
any
tweak
that
you
see
like
okay,
this
looks
better
like
this
way
or
that
way,
and
then
have
people
look
at
this.
Just
like
have
people
look
at
it,
and
this
can
be
part
of
the
story.
You
can
be
like
after
story
after
this
story,
but
I
think
that
we
should
try
to
do
here
like
the
the
messages
look
good
as
they
are
right
now,
so
whoever
did
the
work
before
good
job
there.
F
So
personally,
I
think
we
should
just
implement
this,
as
is
in
the
story,
and
maybe
a
little
tweak
in
here
or
there.
If
we
feel
like
oh
this,
this
feels
a
little
bit
strange
or
I
prefer
it
in
like
a
little
bit
of
a
different
way
and
then,
after
that,
just
get
people's
opinion
about
it,
because
we'll
never
craft
a
perfect
message.
G
It's
like
the
only
place
from
from
an
engineering
code
wise
perspective
that
we
need
to
worry
about
and,
like
maybe
just
like
test
drive
I'll,
add
some
more
tests
around
that.
If
there
aren't
any
so
I
thought
most
of
the
work
there
and
iterating
on
that
function.
I
don't
know
if
that
function
even
exists,
but
that's
that's
how
I
assumed
it
worked.
C
Cool
all
right,
well,
we're
basically
an
even
split.
So
let's
do
one
more
one
more
pointing
real,
quick.
C
Okay,
rough
consensus
is
a
one.
I
I
like
to
make
sure
folks
know
that,
like
they
have
the
latitude
to
put
in
the
effort
they
want
in
order
to
like
get
it
to
what
they
want.
I
think
I
think
we've
talked
through
the
parameters
there
and
there's
there's
options
so
use
your
best
judgment.
C
Three,
so
today,
if
you
have
turn
on
the
schema
function-
and
you
like
put
in
the
flag-
and
you
have
a
private
library
that
doesn't
have
schema,
but
it
does
have
data
values,
you
get
an
error
like
you
can't
go
forward.
C
C
C
We're
talking
about
even
once,
we
release
it
when
we
take
the
flag
off.
This
would
still
end
up
being
true,
I
think,
let's.
H
C
C
Null
and
so
we
don't
warn,
we
just
downshift
to
whatever
you
get
with
data
values,
is
what
you
get:
okay,
okay,
yeah,
let's
defer
this!
Then
I
didn't.
I
didn't
see
that
I'll
pull
this
down.
C
C
Story:
okay,
yeah,
the
title
isn't
great
on
this
one,
but
it
is
baked
this
one
actually
is
very
specifically
about
improving
the
error
message
you
get
when
you
include
a
a
schema
and
you're
targeting
it
to
a
private
library,
but
you
accidentally
name
like
gave
the
wrong
name
for
the
library,
let's
say,
and
so
it
never
gets
used
because
the
library
ref
points
to
some,
a
library
that
doesn't
exist
so
there's
actually
a
check
there.
C
Today,
data
values
has
been
doing
this
from
for
some
time
now,
and
we
now
have
this
for
for
schema,
but
this
is
about
improving
the
the
way
we
message
it.
So
this
is
an
error
message,
improvement.
C
So
this
is
this
gets
to
what
garrett
was
talking
about.
It's
like,
I
think
you
were
saying,
touch
well
like
it's
hard,
it's
hard
to
get
to
the
right
words.
This
is
saying.
Please
take
the
time
to
get
to
the
right
words
in
this
case.
G
C
It's
a
good
question.
I
hadn't
considered
it
personally
when,
when
we're
working
on
this
story
like
putting
the
story
together,
it
felt
more
about
just
straight
up.
That
sentence
was
jargon-laden
and
hard
to
hard
to
really
make
sense
of
what
it
was
really
trying
to
say,
which
was
you
included
a
schema,
and
this
probably
be
the
same
for
data
values.
C
It
was
meant
clearly
meant
for
some
library,
some
dependency
and,
and
it
didn't
make
it
to
that
dependency,
so
something's,
not
right
whether
or
not
that
is
like
easy
to
do
and
and
write
that,
as
like
I'd
say,
go
explore
that
see
what
that
looks
like
see
what
you
think,
like
joao
said,
grab
others
to
like
get
a
second
opinion,
see
if
that
works.
If
you
want
to
go
down
that
route,
yeah.
C
D
F
We
can
discuss
the
image
package
story.
That's
there
about
the
locations.
H
F
F
So
the
the
problem
that
we
were
facing
here
was
the
fact
that
there
isn't
really
like
a
nice
way
for
you
to
know
if
a
bundle
was
copied
from
one
registry
to
another
or
if
just
a
single
image
push
was
was
done
and
because
of
that,
there's
also
a
second
problem
that
is,
when
you're
trying
to
execute
an
image
package
pool.
We
don't
know
if
the
images
that
are
part
of
your
bundle
if
they
are
bundles
themselves
or
if
they
are
just
single
images.
F
So
the
problem
here
is,
if
you,
if
you'd,
never
relocated.
If
you
never
copied
your
bundle,
then
there
might
be
some
images
that
are
part
of
the
the
bundle
repository
like
the
final
bundle
repository.
There
might
be
some
other
images
that
are
not,
and
you
might
not,
with
your
keys
with
your
authentication,
you
might
not
have
access
to
the
location
that
these
images
were
to
start
off.
F
You
don't
need
to
check
the
other
images
to
see
if
they
are
there,
if,
if
they
are
not
there,
so
this
is.
This
is
roughly
the
problem
that
we
are
trying
to
address
here
and
so
another
problem
that
this
causes
is
that,
let's
imagine
that
you
have
10
images
on
your
images
lock
for
a
particular
bundle,
the
operation
of
pulling
a
bundle.
It
requires
you
to
go
to
the
registry
11
times,
one
for
the
bundle
and
one
for
each
image.
F
F
So
how
are
we
planning
on
addressing
this?
So
the
idea
is
on
on
the
copy
with
rename
proposal
there
was
the
concept
of
this
location,
oci
image
that
contained
like
a
supermarket
list,
with
all
the
images
that
are
part
of
a
particular
bundle
right
and
if-
and
we
decided
like
why-
why
don't?
We
borrow
this
particular
feature
that
is
like
isolated-
that's
not
related
to
this
proposal,
but
it
will
be
used
by
this
proposal.
We
can
just
implement
it
before
and
we
can
solve
this
problem
that
we
are
seeing
with
pool.
F
So
this
is
exactly
what
we
are
planning
to
do.
So
if
you
scroll
a
little
bit
down,
there's
a
little
bit
of
information
there,
but
basically
this
oci
image,
this
locations,
oci
image,
will
live
inside
the
bundle
repository
and
it
will
be
of
the
that
format
that
you
see
there
annotations
this
format
here
right,
so
it
will
be
of
the
kind
it
will
be
a
file
on
the
root
of
the
image.
The
file
will
be
called
images,
lock,
location.
Sorry,
I
don't,
I
don't
see
it
there.
F
I
think
it
might
be
a
little
bit
up,
but
it
will
be
on
the
root
of
the
filesys
of
the
of
the
oci
image
and
it
will
contain
a
list
of
all
the
images
and
the
values
that
it
will
contain
are
the
image
itself
so
where
the
image
like
the
identifier
that
is,
will
be
the
same
as
you
have
it
in
the
images
log.
F
The
repository
where
you
can
find
this
image
and
then,
if
it
is
a
bundle
or
not,
so
this
is
relevant
information
for
the
pool,
because
when
you
pull
recursively,
you
want
to
pull
all
the
bundles,
but
you
don't
want
to
find
if
it
is
a
bundle
or
not.
So
you
already
have
this
information
here,
so
this
file
will
contain
all
that
and
the
question
now
is:
how
might
we
find
this
oci
image
right
because
it
will
have
his
own
chai
to
have
like
everything
together.
F
F
So
there
are
some
caveats
for
this
idea
right.
So
if
you
scroll
a
little
bit
down,
so
one
of
them
is
that
this
one
we
are
we're
going
to
use,
tags
and
tags
are
mutable
and
because
tags
are
mutable,
you
might
have
someone
replace
the
image
that
we
place.
That
looks
all
nice
that
contains
that
one
file
it
might
be
replaced
with
something
else.
So
this
is.
This
is
a
problem.
This
is
like
a
week,
it's
a
caveat,
but
for
now
we
are
thinking
that
this
is
okay
and
we'll
see
in
the
future.
F
If
this
starts
causing
us
some
problems,
I
know
that
there
are
some
registries
that
have
immutable
tags
for
these.
This
would
be.
This
would
be
like
this
would
be
a
cabinet
that
doesn't
exist
right.
So,
for
example,
I
think
aws
doesn't
have.
Has
the
ability
for
you
to
say
that
the
repository
have
immutable
tags
and
you
cannot
change
a
tag
after
you
create
it.
So
this
is
one
of
the
caveats.
F
The
other
caveat
is
related
to
replication.
So
if
you
have
two
registries
and
you're
replicating
repos
from
one
registry
to
another,
something
might
happen,
you
might
not
replicate
all
the
images.
So
this
is
a
problem
and
for
now
we're
gonna
say
okay.
So
if
you
did
not
replicate
everything,
it
is
normal
that
things
will
fail.
F
So
these
are
this.
Is
it's
not
like
the
greatest
solution,
but
it's
the
solution.
We
have
for
right
now
and
we're
going
to
see
if
we
can
move
us
forward
with
this.
With
these
two
caveats
that
we
have
so
given
all
this,
how
is
image
package
going
to
start
searching
for
the
images
that
are
part
of
the
button?
So
the
first
thing
that
it's
going
to
look
for
is
for
this
image
to
see
if
it
can
find
the
locations,
and
if
you
could
find
the
locations,
then
it
doesn't
need
to
do
anything
else.
F
It
found
the
images
it
has
a
list
of
all
the
images,
so
we're
good
what
happens
if
it
doesn't
find
it.
So
we
are
going
to
go
back
to
how
things
were
before
right.
We
check
if
the
image
exists
in
the
particular
location
that
it's
the
bundle
registry
bundle
repository
plus
the
shot
to
56
of
the
image,
and
if
we
find
it
awesome,
but
if
we
cannot
find
it,
we
are
going
to
assume
that
the
image
is
in
the
original
location
of
that
image
and
we're
gonna
stop
searching
so
this.
F
Okay,
but
this
could
cause
our
problems
if
we
start
getting
back
to
the
original
image
location.
So
this
this
is
more
or
less
like.
This
is
feels
like
a
very
big
thing,
that's
going
to
happen
and
it
is
going
to
be
a
very
big
thing.
It's
not
going
to
be
back,
it's
going
to
be
backwards
compatible
because
the
schema,
if
it
tries
to
find
the
oci
image
and
if
it
finds
it,
that's
awesome
if
it
not,
it
just
goes
back
to
to
everything
that
to
the
old
ways
right.
F
F
That
contains
more
or
less
the
acceptance
criteria
that
we
were
thinking
for
this
story.
So
in
the
first
scenario
you
create
a
bundle
and
then
you
copy
the
bundle
to
this
new
location,
and
it
should
succeed
any
when
you
pull
the
locations
image.
You
should
see
one
file
called
images
locations,
not
yaml
on
that
where
you
pulled-
and
it
should
look
like
this,
so
it
should
contain
this
particular
image
where
it
is
located
and
that
this
is
not
the
bundle,
so
this
is
always
piggybacking
on
our
examples.
F
F
F
When
we
copy
it
to
a
new
location,
there
should
be
only
one
image,
one
locations
image,
and
when
we
pull
that
location's
image
now
it
should
have
two
images
associated
to
it
with
it.
One
is
the
initial
image
the
same
one
as
we
had
in
the
scenario
one,
but
now
we'll
also
have
a
location
for
the
bundle.
That
was
a
nested
bundle
and
it
should
say
true
there
cool
so
third
scenario:
copy
bundle
twice
with
image,
creates
a
location
image
pointing
to
the
late
ladder
location.
F
F
F
F
And
fifth,
then
I
believe
it's
the
last
it
when
you
copy
the
bundle
twice
first,
with
an
older
version
of
image
package,
it
creates
a
images
on
the
second
copy
and
the
images
location
images.
So,
let's
imagine
you
have
an
older
version
of
the
image
package.
You
copy
the
bundle
first
from
place
a
to
place
b
and
then,
if
you
copy
again
image
package
with
their
newer
version
of
image
package
that
contains
this
feature
when
you
copy
it
from
b
to
c
it
creates
the
images
log
file
and
it
doesn't.
G
F
F
C
G
And
just
to
be
clear,
like
this
story
is
just
to
build
that
new,
like
the
oci
image
that
contains
the
locations
like
at
the
end
of
the
story
having
you
know
not
having
all
the
authentication
credentials
for
the
other
registries,
that
problem
still
exists
and
at
the
end
of
the
story.
Well,
the
performance
gets
a
little
bit
worse,
but
the
next
story,
once
we
start
using
pool
like
those
two
problems,
go
away.
F
G
B
B
A
All
right
thanks
everyone
for
all
the
great
discussions
we
had
today
if
you're
watching
this
from
home.
We
hope
that
you
are
able
to
attend
our
next
community
meeting,
which
will
be
june
7th
at
11.
30
am
pacific
time,
as
our
may
31st
meeting,
which
is
next
week,
is
cancelled
due
to
international
holiday.
Otherwise
we
do
meet
every
monday
for
these
community
meetings
at
11,
30
am
pacific
time.
We
also
meet
every
second
and
fourth
thursday
at
11.
A
We
will
be
meeting
for
office
hours
this
week
on
may
27th,
so
please
feel
free
to
join
us
then.
Otherwise,
if
you
aren't
able
to
at
any
of
the
meetings
you
can
find
us
on
the
kubernetes
public
slack
course
workspace
and
the
carville
channel
and
on
twitter
at
carvilleunderscoredevdev,
and
with
that
we
hope
you
have
a
great
week.
Thank
you.