►
From YouTube: Carvel Community Meeting - May 3, 2021
Description
Carvel Community Meeting - May 3, 2021
We meet every Monday at 11:30am PT / 2:30pm ET. We'd love for you to join us!
Details and notes from this meeting can be found here: https://hackmd.io/F7g3RT2hR3OcIh-Iznk2hw#May-3-2021-Agenda
A
A
If
you
have
anything
you
wish
to
discuss
with
a
team,
you
can
add
that
to
the
agenda
here
down
at
the
bottom,
with
a
discussion
topic,
so
the
triage
help
for
anything
that
we
do
not
get
to.
Today
we
will
move
those
items
to
our
office
hours
or
to
the
next
community
meeting
our
office
hours
are
held
every
second
and
fourth
thursday
of
the
month
at
11,
30
am
pacific
time,
2
30
p.m.
Eastern
time
and
our
carnival
community
meetings
are
every
monday
at
that
same
time,
11
30
a.m.
A
A
This
week
is
actually
a
really
special
week
as
kubecon
clown
native
con
europe
virtual
begins
tomorrow
and
carville
is
participating
in
a
few
different
ways.
First,
we
do
have
a
session
tomorrow
may
4th
at
5
40
a.m,
pacific
time,
and
that
is
over
the
kubernetes
package
management
using
unix
philosophy
with
carvel
and
then
on
may
6.
We
have
live
office
hours
which
are
going
to
be
held
at
the
vmware
virtual
booth
at
5
a.m,
to
5
30
a.m.
B
C
Yeah
we're
still
charging
along
here
there's
a
couple
of
aspects
that
we're
tackling
right
now.
The
biggest
chunk
right
now
is
being
able
to
include
multiple
schemas
so
that
you
can
either
break
them
apart,
as
you
need
to
as
an
author
or
as
a
someone
who's
consuming
someone
else's
ytt
library
being
able
to
add
your
own
if
you're
going
to
include
additional
templating.
C
So
that's
that's
a
chunk
and
then
another
part
is
in
how
we're
expanding.
We
have
one
more
type
that
we
want
to
include
in
the
overall
type
system
to
get
to
sort
of
an
mvp,
which
is
the
ability
to
say
well,
allow
any
type
so
we're
working
on
how
to
make
that
work.
Well
with
the
rest
of
the
typing
work
that
we're
doing
so,
there's
quite
a
bit
of
progress.
C
B
C
Yeah
we're
starting
to
around
the
last
lap
here
on
the
plane
yaml
as
data
values,
pretty
exciting
there
there's
a
few
things
to
nip
and
tuck
on
the
proposal
itself.
We're
actually
even
getting
started
helen
and
I
were
just
working
on
starting
to
write
some
stories
that
will
be
there,
regardless
of
how
some
of
these
details
iron
out
so
we're
getting
ready
to
to
be
able
to
pull
the
trigger
on
on
that
work
as
well.
C
First
portion
of
that,
we
still
have
more
work
that
we're
doing
more
research
that
we're
doing
in
terms
of
how
do
we
help
support
the
integrators,
who
are
using
schema
information
to
generate
their
ui
they'll
need
more
than
just
types,
and
so
we're
working
on
how
to
nail
down
exactly
what
that
is
in.
In
what
priority
should
we
tackle
this.
B
B
C
B
D
I
think
it
depends
on
like
how
familiar
folks
are
with
the
like
proposal.
Listed
here
is
anyone.
I
know
I
just
sent
a
message
about
it.
Last
friday,
when
most
people
were
probably
gone,
so
I
think
we
decided
to
add
this
as
a
discussion
topic
a
little
late.
If
folks
don't
really
have
context
or
have
thoughts
to
air
about
it.
Maybe
we
could
delay
until
next
week
or
some
other.
C
E
F
B
F
H
I
agree
with
that.
I
think
that's
like
the
simplest
way
to
do
go
about
it.
One
thing
we
would
have
to
modify,
though,
is
to
be
explicit,
which
one
out
of
all
the
list
of
items
here
we're
displaying
in
the
bottom,
because
it
loses
that
connection
between
what
you're,
what
what
you're
seeing
versus,
which
item
is
selected
above
so
highlighting,
could
be
an
easy
way
to
solve
that.
C
I
C
I
remember
having
discussions
with
some
folks
again.
This
is
like,
without
you
know,
like
user
testing
or
anything
in
the
room
of
like
wanting
to
be
able
to
potentially
have
this
view
like
there
might
be
somebody
who's
like
okay.
I
want
to
see
everything
so
I
can
kind
of
visually
scan
through
here,
but
also
not
wanting
to
overwhelm
the
user
as
well.
So
if
it
were
just
tabs
alone,
then
I
don't
know
if
that
would
preclude
being
able
to
sort
of
open
everything.
I
I
If
they
had
like
an
area,
we
would
know
that
the
tabs
are
on
top
right,
maybe
that
the
design
that
we
have
there,
it's
not
really
helpful,
and
if
you
really
want
to
expand
them
all
to
be
fair,
I
don't
really
know,
but
maybe
the
buttons
should
be
vertical
and
be
underneath
each
section
like
the
basics,
opens
the
basics.
If
they
are
all
closed,
they
are
like
three
buttons
and
when
you
open
it,
it
just
moves
the
two
to
bottom,
some
sort
of
like
an
accordion
or
something
that
way
you
could
expand
them.
C
E
I
think
another
piece
of
this
that
maybe
I
always
struggle
with
sometimes,
is
that
I
feel
like
a
lot
of
the
times.
The
titles
in
here
don't
always
like
fully
capture
like
what
is
shown
in
an
example
sometime
and
if
there
was
some
type
of
search
functionality
by
which
you
could
kind
of
see.
Sub-Topics
of
the
examples
would
be
kind
of
interesting,
for
instance,
like
the
overlay
example,
I
think
really
focuses
on
like
the
remove
aspects
of
overlays.
E
Or
maybe
maybe
it's
a
different
one,
but
like
there,
there
are
always
these
like
yeah.
There
are
always
like
topics
like
within
it
that
I
feel
like
are
very
often
not
captured
by
a
title
itself,
and
I'm
wondering
if
there
would
be
a
way
to
add
like
some
like
tags
or
something
that
someone
could
like
search
through
to
like
find
examples,
kind
of
a
little
bit
more
easy.
That's
not
like
the
perfect
way
to
do
it,
but
that's
something
I
personally
like
struggle
as
far
as
like
finding
useful
examples
through
the
site.
F
B
D
So
this
is
basically
a
proposal
that
proposes
that
we
take
the
existing
package
cr,
which
is
a
cr
that
sort
of
represents
a
generic
app
and
allows
consumers
to
like
easily
create
an
app
cr
without
knowing
about
an
app
cr.
So
they
can
see
that
fluid
bit
is
available
in
their
cluster
and
that's
available
as
a
package
and
then
install
it.
D
And
currently
the
package.
Cr
is
a
cr
that
contains
the
entire
spec
of
how
to
do
that.
So
informing
cap
controller,
how
to
fetch,
template
and
deploy
an
app,
as
well
as
like
a
lot
of
metadata
around
the
package
itself.
So
like
a
logo
or
a
short
long
description,
some
of
the
like
higher
level
stuff
about
the
package.
D
D
So
due
to
those
two
sort
of
poor
experiences,
we're
proposing
to
like
split
out
the
package
cr
into
two
new
crs
one
being
the
package
cr,
which
will
just
sort
of
contain
the
high
level
information
of
the
package
stuff,
that's
likely
not
versioned,
so
short
long
description,
logo
that
sort
of
info
here
and
then
to
go
along
with
that
a
package
version
cr.
So
this
is
where
you'll
put
all
the
versioned
information.
So
what
are
we
actually
fetching?
How
are
we
templating?
D
D
So
now,
in
this
world,
discovering
packages
becomes
a
lot
easier
because
there's
just
one
entry
for
each
unique
package,
so
you
can
cube
ctl
get
packages
and
you'll
get
a
list
of
just
a
single
entry
per
package,
then
taking
that
to
package
versions,
you're
able
to
filter
by
a
specific
package
name
and
get
a
list
of
all
the
versions
that
are
available.
So
it's
a
little
bit
more
natural.
D
D
D
So
that's
an
open
question,
but
that's
also
a
question
that
exists
in
today's
world
of
packaging.
So
if
you
have
a
package
with
the
same
name
and
version
in
today's
world,
but
a
slightly
different
description
or
spec,
you
run
into
the
same
issue
of
how
do
you?
How
do
you
pick
one
repo
or
one
package
from
a
repo
over
the
other,
so
we're
punting
on
that?
Answering
that
question
for
now
and
sort
of
calling
that
other
scope
of
this
proposal.
J
D
D
Any
versions
of
any
package,
so,
if
you
have
you
know
40
packages,
each
with
10
versions,
you're
already
at
400
list
entries,
so
it
doesn't
take
much
to
sort
of
blow
up
in
that
sense
and
if
you're
packaging
repos,
I
mean
three
or
four
repos
could
result
in
lots
of
packages,
especially
as
time
goes
on
and
new
versions
are
released
and
the
assumption
I
think
we're
making
is
that
repos
aren't
trimming
out
versions
as
they
go.
So
every
new
version
just
increases
the
number.
C
It
does,
it
does
also
conceptually
make
sense
to
me
that
there'll
be
algorithms
we'd
like
to
write
against
a
package
and
that's
harder
to
do
now,
because
it's
got
the
version
baked
into
it,
be
sort
of
easier
to
talk
about
things
that
I
don't
care.
What
version
like
this
thing
goes
from
here
to
here?
I
don't
know
or
expunge
everything
that
sort
of
easier
to
reason
about.
F
I
This
is
interesting
because
I
did
not
expect
this
to
be
a
problem
at
least
the
way
I
was
looking
at
the
full
package,
cr
api
and,
like
the
usage
of
it,
I
was
expecting
these
repositories
to
be
catered
repositories
that
contained
only
one
or
two
versions
per
package,
and
why
is
that?
Because
I
assume
that
if
we
want
to
allow
people
to
have
like
if
you're
a
if
you're
a
cluster
operator
and
you're
managing
a
cluster,
you
want
to
sanction
just
one
or
two
versions
of
a
particular
software
for
people
to
use.
D
D
I
I
General
concept,
I
think
I
think
it's
an
interesting
concept
like
in
this
separation.
I
see
some
problems
like
you
mentioned
the
problems
of
like
like
competition
between
two
two
repositories
that
can
have
the
same
package
definition
or
like
a
little
bit
of
a
different
package.
Definition
what
you
do
about
it
about
that.
I
I
think
this
could
be
like
a
big
problem
or
even
if
for
some
reason,
as
you
mentioned
like
if
the
logo
changes,
which
is
not
something
that's
unheard
of,
then
what
logo
are
going
to
going
to
display,
because
one
some
versions
have
a
particular
logo.
Some
other
versions
have
a
different
logo
or
some
sort
of
like
description
of
a
particular
version.
C
D
But
that
is
sort
of
like
a
high-level
description
and
then
you
can
go
into
the
release,
notes
and
see
like
okay,
here's
what's
changed
in
all
the
releases,
but
the
readme
is
sort
of
like
here's
at
a
high
level.
What
this
tool
does
so
we
sort
of
thinking
of
the
package
crs
a
place
for
that
information,
and
then
you
can
drill
further
into
versions
and
what
each
version
does.
I
Like
always
a
version
of
this
package
cr,
and
this
could
hype
even
more
the
problem
of
having
like
different
versions,
different
repositories
that
have
different
versions
of
this
package,
cr
that
have
a
little
bit
of
a
different
things.
So,
for
example
like
if
you
add
a
maintainer
to
a
particular
package
right
like
in
one
side,
you
have
it
in
the
other
side,
you
don't,
and
if
you
do
not
provide
a
package
cr,
then
you
cannot
update
that
information.
I
I
D
D
B
I
think
another
benefit
to-
I
guess
like
actually
implementing
this
soon.
Is
that
we're
we're
still
under
the
alpha
release
umbrella,
and
we
can
make
these
changes,
so
we
could
always
change
back
if,
if
we
hear
from
package
out
there
saying
like,
oh,
like
I
created,
you
know
packages
in
the
old
way
and
it
was
much
better.
I
It's
like
an
idea
like
you,
don't
need
to
have
a
lot
of
packages
built
in
order
to
test
out
the
current
behavior.
You
can
just
mock
out
something
and
get
people
into
a
room
and
just
ask
them
like
just
use
kubernetes
and
their
cue
cuddle
here
and
find
out
what
packages
you
have
installed
and
then
you
just
just
give
like
a
lot
and
see
how
they
navigate
that
without
having
to
incur
in
the
in
the
cost
of
having
to
build
something
new
that
people
might
not
like
might
not
work
for
people.
So
maybe
ins.
B
D
B
B
I
Really
not
discuss
discuss,
but
we
we
are
in
the
midst
of
releasing
a
new
version
of
image
package
that
contains
same
performance
improvement.
So
I
think
by
today
we
should
have
a
new
version
out.
I
think
the
release
is
already
in
github,
but
not
all
the
channels
are
updated
yet
so
we're
in
the
midst
of
doing
that.
I
Another
thing
is
about
the
the
rename
copy
rename
proposal
for
image
package.
I
know
that
we've
been
in
a
little
bit
of
a
like
a
silence
period
because,
like
nobody
was
saying
anything
and
we're
still
it's
still
going
on,
I
wrote
something
into
the
karma
channel.
So
just
take
a
look
there.
There
was
like
some
other
proposal
for
the
strategies
and
eventually
I'd
like
to
bring
this
up,
maybe
on
the
next
community,
not
maybe
the
next
community
meeting
or
the
office
hours
try
to
go
through
it.
I
C
I
More
to
it
so
like
I,
I
link
the
comment
from
jorge.
I
think
that
was
proposing
a
different
way
of
providing
strategies
instead
of
having
like
this
long
name.
That
contains
like
all
the
things
that
happen
on
that
strategy.
You
get
like
a
shorter
names,
for
example
like
relative
to
to
bundle,
and
then
you
have
other
fields
that
you
can
set
say
like
I
want.
I
want
also
to
include
this
prefix.
I
I
want
to
include
that
suffix
on
the
on
the
on
the
strategy,
so
have
like
more
broad
strategies,
but
then
that
can
have
like
little
thing
little
fields
that
you
can
change
and
make
it
a
little
bit
better.
So,
like
that's,
not
a
proposal
of
how
we
might
do
strategies
here
and
I'd
like
also
like
to
know
a
little
bit
more
to
see
if
like
people
would
be
okay
with
completely
removing
the
overrides
section.
I
The
problem
is
that
you
might
not
have
credentials
to
search
on
the
original
registry
right
so
like
we
are,
we,
we
wrote
up
an
issue
that
is
going
to
try
to
address
that.
That
is
part
of
what
existed
on.
That
was
part
of
the
proposal
for
the
rename,
and
that
is
like
the
location
file.
So
we
are
also
in
the
midst
of
doing
that.
So,
if
you
run
into
this
problem
in
image
package
know
that
we're
looking
into
it.
I
Basically,
previously
we
were
so
the
first
step
that
image
package
does
when
it
copies
the
images
from
one
register
to
another
or
from
one
registry
to
a
tar.
It
tries
to
find
all
the
images
that
needs
to
copy
first,
that's
like
the
first
step
and
that
was
being
done
in
a
sequential
fashion.
So
now
what
we
did
we
parallelize
all
of
that,
so
we
are
instead
of
checking
one
image
at
a
time.
We're
checking
five
images
at
a
time
to
see
if
they
exist
and
where
they
exist,
so
they
can
be
copied.
I
So
that
was
the
improvement
and
afterwards
we
also
did
something
else
that
we
to
select
to
see
where
the
images
are
like
this
this.
This
part
that
I
said
that
we
that
had
a
problem
right
now,
so
it
was
also
done
sequentially.
So
it
first
goes
to
the
new
registry,
see
if
the
image
is
there,
then
it
goes
to
the
old
regis
in
the
original
register
and
see
if
the
image
is
there-
and
this
was
done
sequentially
for
all
the
images.
I
So
two
two
calls
to
the
registries,
so
this
is
done
in
parallel
as
well.
So
we
were
able
to
in
I,
oh
I
don't
remember
the
timings,
but
we
we
had
an.
We
had
a
bundle
that
had
30
images
on
it
and
we
were
able
to
go
from
a
two-minute
marker
from
copying
to
disk
to
roughly
45
seconds
or
something
like
that.
So
it
was
like
a
huge
increase,
just
the
fact
that
we're
parallelizing
the
check
to
see
if
the
images
are
present
well.
Thank.