►
From YouTube: OCI Weekly Discussion - 2023-03-02
A
C
A
E
A
Your
stuff
down
on
that
one
I
didn't
really
want
to
discuss
too
much
of
999
and
the
clamp
and
whatnot
I
want
to
just
put
them
out
there
as
we
need
to
make
a
decision
on
it
and
move
forward,
because
I
think
we've
discussed
it,
we
definitely
discussed
it
more
than
enough
would
help
us
move
those
up
to
the
actionable
agenda
items
instead
of
the
topics
we
want
to
discuss.
I.
A
A
Speaking
about
the
decision
part
I
feel
like
if
we
try
to
go
without
approving
999
I
think
we're
going
to
be
on
an
infinite
Deadlock
and
so
I'm
kind
of
looking
towards
you
Sashay
of
whether
or
not
you
would
be
willing
to
go,
not
just
with
an
abstain
but
to
an
approve
to
just
get
that
thing
worked
through
with
the
knowledge
that
we
would
still
Circle
back
and
try
to
get
this
thing
in
there
in
the
future.
In
some
other
way,.
F
G
I
think,
as
as
a
sort
of
plan
of
information,
does
this
mean
that
lowercase,
a
artifact
support
would
be
removed
from
image
spec.
A
A
F
G
But
we
want
the
interested
parties
who
want
to
upload
artifacts
or
s-bombs
and
other
things
would
like
the
spec
to
be
amended
before
we
remove
artifacts
back
right
like
if
we're
going
to
use
image,
manifest
I
think
there
are
steps
forward,
but
what
I'm
concerned
is
we
potentially
get
in
a
situation
where
we
merge
999,
because
it
has
received
a
consensus
and
then
we
fail
to
receive
consensus
on
lowercase
a
artifact
and
there's
no
guarantee
right.
A
Not
today
a
whole
nother
repo,
open,
container,
slash
artifact.
That
document
is
exactly
what
we're
going
to
move
into
image
spec,
so
that
already
exists.
It's
already
out
there
of
saying
here's
how
to
do
it.
That's
how
Helm
went
through
forward
with
adding
their
stuff
and
so
I
feel
like
that's
in
the
spec,
we're
just
going
to
move
it
from
one
side
to
another.
E
G
I
I
find
that
really
unsatisfying,
because
I
think
the
specification
allows
implementations
image
spec,
as
written
with
999
merged,
allows
implementations
to
deny
artifacts
from
being
published,
which
makes
it
not
a
valid
Target
for
conformance
right,
like
the
conformance
requirements,
are
not
there.
Unless
it's
part
of
image,
spec
saying
that
there's
another
repository
doesn't
doesn't
mean
it's
part
of
the
specification.
H
D
C
Yeah
this
is
it
as
even
Brandon
was
asking
me.
I
was
like
I'm
not
going
to
Green
Light
999
until
the
things
that,
basically
until
we
codify
all
the
warts
that
we're
accepting
with
using
the
image
manifest,
they
have
not
been
codified,
actually
get
merged
or
like
has
it
have
an
action
plan,
because
otherwise,
this
pressure,
that's
actually
bringing
people
to
talk
about.
It
instantly,
goes
stale
once
that
discussion
that
Crux
moves
on.
D
Michael,
you
have
your
hand
up
lucky
on
Saturday.
B
Yeah
I
was
just
going
to
piggyback
off
what
Aaron
was
saying.
The
little
a
artifact
repoers
actually
has
issues
open
to
say.
Should
we
delete
this
repo
and
there
are
plus
ones
on
the
deletion
of
the
little
a
artifacts
sitting
out
there
I
think
John
raised
it
in
issue
61
and
people
have
actually
jumped
on
that
as
well.
So
just
that's
worth
consideration.
If
you
were
to
do
this
and
then
delete
little
a
artifacts,
then
artifacts
would
be
gone.
B
D
The
if
we
remove
artifacts
and
1.1
gets
cut
without
codifying
it's
a
risk,
so
just
to
kind
of
make
sure
that
the
work
is
it's
holistic
right
we've.
We
agreed
that
we
don't
do
the
artifacts.
We
codify
the
issues
that
are
there
on
the
Milestone,
which
is
to
find
the
scratch
descriptor
at
least
the
that
way
we
make
sure
that
both
of
them
get
merged
together.
D
That's
what
my
I
think
I'm
leaning
towards
that
unless
there's
any
strong
pushback
or
anything
like
that.
That's
why
I
didn't
green
light,
the
other
one!
It's
not
like
you've
agreed
not
not
agreed
to
do
that,
but
we
just
haven't
completed
the
work
to
do
that,
and
just
keeping
everybody
honest
as
to
yes,
we
can't
just
push
one
PR
through
without
kind
of
addressing
the
other
feedback
is
the
way
I
would
like
to
go
again.
G
I
think
10
30
is
the
pr
that
I'm
most
interested
in
then,
if
we're
going
to
go
down
the
accepting
999
path,
which
is
to
sort
of
codify
the
change
to
descriptor,
to
config
media
type
and
to
layers
into
distinguish
I.
G
Think
CNN,
you
raise
a
point
on
1028,
which
is
that
the
language
around
implementations
didn't
sort
of
clarify
intent
like
runtime
implementations
should
be
able
to
error,
for
example,
when
they
receive
an
unexpected
manifest
so
10,
28
and
10
10
30
were
both
amended
to
have
language
that
distinguishes
between
a
implementation.
That's
involved
in
storing
and
copying
versus
an
implementation,
that's
processing
or
using
media
types,
and
that
the
storing
and
copying
would
cover
like
a
proxy
use
case
or
like
or
as
CLI
things
like
that.
C
Some
point,
including
1023
sure
I
I,
think
we
should
review
some
of
these
other
ones
in
light
of
not
really
deferring
on
999.
That's.
That
was
not
quite
what
was
just
said,
but
rather,
if
there's
General
agreement
literally
the
only
thing
holding
it
off
is
having
Clarity
on
what
the
action
plan
would
be
for
these
other
ones.
It
is
not
a
deferral
of
none
and
it's
not
a
strict
removal
of
artifact.
You
know
stuff,
it's
like
what
what
is
the
actual
action
plan,
because
I
think
right
now,
most
people
are
just
frustrated
and
tired.
C
G
Should
we
should
we
just
go
down
the
the
list,
I
think
all.
E
B
G
Looks
like
the
first
two
Tiaras
are:
either
making
players
optional
or
permitting
a
zero
length.
B
C
C
C
Okay,
I
think
both
of
these
should
be
in
the
light
of.
C
125.,
okay,
so
real
quickly,
zero
links
being
equal,
zero
or
empty
layer
set
and
clarify
that
layers
is
optional.
Are
these
two
columns
for
those
that
have
followed
this
issue?
C
I
will
say
succinctly:
this
has
been
a
great
effort
on
1025
and
I.
Think
that
what
we
ship
like
what
the
effort
was
done
with
the
conformance
test.
C
This
is
useful
to
say,
like
at
a
certain
point
in
time
like
what
things
were
not
just
like
a
living
issue
that
continues
to
get
updated
but
kind
of
like
here's,
why
certain
decisions
are
made
and
if
this
could
be
put
in
a
markdown
and
actually
like
tag
somewhere.
It's
very
useful
because
this
empty
layers
column
is
showing
that,
like,
while
a
lot
of
Registries
except
an
empty
array,
ECR
and
GCR,
don't
I'm
sorry
empty,
empty
layers.
C
The
main
meaning,
no,
no
layer,
okay,
so
these
are
the
empty
layers.
This
is
saying
just
the
layer
keyword.
Is
there
the
key,
but
their
array
is
empty,
that
GCR
and
ECR
bail
on
those
two.
So
it's
pretty
broad
consensus.
That
would
be
really
nice
if
we
could
say
this,
but
it
would
technically,
you
know
be
a
little
bit
breaking
if
we
said
it
may
be
zero
length.
C
I,
don't
I,
don't
know
like
I'm,
not
I
feel
like
it'd,
be
nice
to
have
this
kind
of
clarity,
but
it
does
seem
that
that
would
be
a
little
bit
of
a
departure
of
like
what's
expected
like
it
may
be,
but
some
some
at
least
two
Registries
so
far
have
said
they
require
it.
C
So
that's
a
bit
of
language
change.
I
feel
like
that's,
not
a
huge
hurdle.
If,
if
we
want
to
leave
the
language
ambiguous,
we
could
just
close
the
pr.
C
The
the
next
column
here
is
no
layers
saying
the
layers.
Object
is
not
even
present
in
the
structure.
More.
C
Registries
have
an
issue
with
that:
ECR
add
artifactory
and
quiet
event
list,
so
that
one
is
I.
Think
this
one
is
a
bolder
Bolder,
flavor,
PR
and
I'd
be
more
in
favor
of
closing
some,
even
if
the
fact
that
some
registries
actually
allow
you
to
not
ship
the
layers
object
either
layers
key
at
all.
This
is
probably
more
more
likely,
in
my
mind
to
close
opinions,
one
way
or
the
other.
D
C
A
C
G
Maybe
heavily
is
too
strong
of
a
word
but
you're,
leading
a
towards
one
of
these
being
more
viable
PR.
C
As
with
all
these
things,
osia
has
tended
to
be,
this
is
how
it
functions
broadly
in
the
world.
Let's
codify
it,
it's
a
trailing
spec
and
the
fact
that
it
was
ambiguous
and
there's
to
be
fair
as
the
exhaustion
and
frustration
that
you're
all
getting
to
feel
now
has
been
felt
before
and
at
some
point
you're.
Just
like.
Let's
merge
this
and
move
on,
and
then
you
realize
this
thing
is
ugly
and
it
could
have
been
clear.
C
This
is
one
of
those
moments
that
you
can
see
that
there
are
plenty
of
places
that
actually
accept
this
and
it's
like
cool.
So
at
what
point
do
we
say?
Let's
fix
it,
even
though
we
see
that
there's
a
lot
of
green
check
marks
here,
but
there's
a
couple
of
red
ones.
C
Should
we
just
make
the
decision
to
try
and
push
towards
uniformity
or
leave
it
ambiguous
and
that's
like
a
a
Breaking
Point
and
that
it
is
a
trailing
spec,
let's
codify
what
people
are
actually
using
and
you
can
see
them,
people,
maybe
not
using
it,
but
at
least
support
it.
So
we
could
swing
either
way.
I'm
in.
F
F
C
I
opened
these
two
PRS
literally
in
the
moment
of,
if
we're
removing,
if
we're
going
forward
with
999
and
we're
going
to
try
and
make
image,
you
know
codify
the
warts
on
image
spec
to
make
it
look
and
smell
enough,
like
artifact,
manifest
right
what
what?
What
are
the
kind
of
things
and
behaviors
that
we
can
kind
of
clean
up
and
there's
some
language.
That's
been
unclear
about
this
in
general,
you
can
add
fields
that
don't
exist
cool,
so
somebody
could
add
a
blobs
field
that
doesn't
exist
and
roll
with
it.
C
F
F
F
Okay,
my
concern
without
going
to
a
B2,
is
that
you
would
be
making
backwards
incompatible
change,
yeah,
older
Registries
would
no
longer
work
in
that
context.
F
Where
I
could
see
them
adding
code?
For
you
know
the
check
to
see
if
the
config
manifest
is
different
and
therefore
it's
an
artifact
or
some
kind
of
a
service
fix.
We
could
ask
them
to
do
that
on
the
older
Registries
I
can't
see
going
in
and
asking
them
to
say
hey.
Can
you
support?
You
know
nil
layers
now
on
the
older
images?
F
That's
what
I
was
expecting
to
see
in
the
context
here
and
again.
The
the
reason
is:
what
is:
you've
got
two
versions
of
of
your
registry
out
in
the
field
right,
one,
that's
being
used
private
and
one
that's
being
used
public
for
a
different
use
case.
It
works
in
private,
you
push
it
to
your
public
and
it
doesn't
work
because
it's
missing
you
know
the
change
to
allow
from
the
old
blobs,
for
example,
in
Quay.
A
A
C
So
that
sounds
I
mean
like
in
in
that
way.
It's
like
just
saying
cool:
let's
leave
it
ambiguous
and
codify
these
other.
Like
effective.
You
know
different
kind
of
thing
of
he
scratched
layers,
so
it
at
least
is
portable
and
and
that.
A
Way
the
only
ambiguous
in
one
place
of
the
spec,
and
so
maybe
you,
instead
of
changing
the
other
two
places
to
make
them
a
potential
breaking
change
on
places
to
codify
that.
No,
we
actually
do
require
one
at
least
one
layer
in
there.
C
C
G
Yeah
I
was
gonna,
say
I,
think
it
adds
ambiguity
for
artifacts
if
they
have
to
have
at
least
one
because
it
means
that
an
artifact
that
does
want
to
express
an
ordinal
list
of
things
that
could
be
zero
to
n
items.
If
there's
one
item,
it's
going
to
be
ambiguous,
like
they,
the
conditional
logic
moves
into
the
client
implementation
there.
In
terms
of,
is
that
first
item
a
scratch
blob,
or
is
it
a
actually
what
I
want
my
first
item
in
this
layers
manifest
or
layers
bloblist
to
be.
C
C
I
feel
like
it's
trying
to
meet
people
in
the
in
between
and
not
add,
there's
no
take
vaccies
in
this
stuff.
So
if
we
add
nonsense
that
works
around
the
fact
that
artifact
manifests
or
something
more
generic
isn't
there,
then
it's
now
codified,
and
so
this
is
like
a
bunch
of
people
trying
to
compromise,
including
myself,
and
then
it's
still
like
whatever
so
I'm,
fine
closing
them
honestly
yeah
it
just
it's
gonna,
literally
codify
warts
and
whatnot,
for
the
sake
of
trying
to
clean
things
up.
D
D
D
A
I
I
don't
want
to
call
it
Behavior,
good
or
bad,
because
I
think
people
are
trying
to
solve
different
problems,
and
so
for
you
for
a
problem
that
you're
solving
it
might
be
a
bad
behavior
for
problem.
Someone
else
is
solving.
It
might
be
a
good
behavior.
I
think
this
is
in
context
to
999
yeah
and
so
I'm
I'm.
A
Looking
at
this
in
terms
of
from
the
context
of
Registries
what
they're
supporting
today
the
clients
out
there
of
container
runtime
stuff
like
that,
what
they're
supporting
expecting
to
receive
today
I
think
all
the
tooling
out
there.
A
significant
chunk
of
it
has
this
dependency
in
there,
and
so
I
would
feel
like
it.
It
makes
sense
that
if
we're
inconsistent
and
the
most
supportive
thing
is
to
say
that
it
must
have
a
layers
field,
then
I
would
go
that
direction
rather
than
trying
to
force
a
change
to
support
a
different
use
case.
A
D
Would
What
Mike
Brown
what
he
was
suggesting
is
if
it
is
not
an
image,
config
clients
may
choose
to
have
the
layers
optional,
fair
enough.
F
A
I'm
comfortable
with
a
couple
of
those
statements,
but
not
all
of
them,
I'm
comfortable
with
saying
that
we
want
to
allow
it
to
be
non-ordinary
and
that
doesn't
necessarily
have
to
have
a
tar
file
in
the
lobby
and
whatever,
but
I,
wouldn't
say
the
whole
layer.
Skill
goes
away
just
knowing
that.
How
much
that's
been
codified
into
different
places.
F
A
The
way
I
would
differentiate.
The
two
is
knowing
that
whatever
change
we
make
here
also
applies
to
images,
and
so
if
we
say
that
we're
going
to
make
this
field
optional,
that
means
that
an
image
can
make
this
field
optional
and
that
can
be
chaotic
to
run
times
where.
If
a
registry
refuses,
because
you
upload
something
that
has
an
unknown
media
type,
well,
that's
only
affecting
that
one
media
type.
It's
not
breaking
all
the
images
and
other
content,
that's
being
pushed
to
registry,
not
good,
but
it's
limiting
what
we
broke.
A
F
In
the
context
that
this
is
a
container
that
makes
no
sense
unless
you're,
unless
you're,
unless
you're
promising
to
to
mount
an
additional
layer
that
didn't
come
with
the
image
and
run
that
in
the
command
right.
E
F
All
right,
Leah,
the
the
important
thing
is
when
you
get
over
to
the
distribution
perspect.
In
this
context,
you
need
to
be
able
to
tell
the
tell
the
the
race
provider
that
it's,
okay,
just
don't
make
it
backwards,
incompatible,
don't
change
images
to
Now
fail
if
they're
running
on
an
older
version
right
if
you've
made
your
decision,
you've
made
your
decision
and
now
here's
what
you
also
need
to
do
for
artifacts
right,
that's
what
I
meant
by
opso
optional,
but
not
breaking,
would
be
the
recommendation.
G
In
in
PR,
1030
I
had
to
double
check
what
I
had
written
down
in
that
PR,
like
your
PR
Brandon,
which
I
think
you
added
a
a
clause
above
the
layers
additional
requirements.
G
I
think
I
copied
language
very
similar
to
yours,
which
is
that
if
the
Manifest
is
into
the
way
I
wrote
it
in
10
30
is
if
the
Manifest
is
intended
for
use
by
oci
runtime
specification
implementations
so
specifically
just
calling
out
runtime
spec
implementers.
Then
these
additional
restrictions
apply,
which
is
that
there
has
to
be
a
certain
number
of
layers
that
they're
ordered
that
Etc.
Does
that
I
think
meet
your
your
needs
is
that
if,
if
this
manifest
is
intended
for
runtime
spec,
then
it
has
to
have
at
least
some
quantity
of
layers.
A
A
No,
but
that's
the
difference
between
the
two
and
so
just
knowing
that
something
like
GCR
is
not
going
to
be
updated
in
a
long
while
it's
just
Registries
like
that
I
I
hate,
to
put
something
in
the
spec
that
we
know
is
going
to
break
a
major
registry
like
that.
C
C
This
back
body
is
now
tripping
over
itself
at
the
fact
that
we're
trying
to
reach
consensus
beforehand
and
then
merge
stuff,
so
we're
literally
pivoting
an
entire
approach,
and
these
kind
of
circular
discussions
it's
this
is
this
is
a
little
bonkers.
This
is
why
everybody's
getting
confused
and
frustrated
is:
we've
never
sat
around
and
said
now,
everybody's
in
agreement.
Let's
merge
it,
that's
not
how
it's
worked
just
hasn't.
G
I
suppose,
then,
as
a
as
a
point
of
order,
would
it
be
sense
for
this
body
to
look
at
its
governance
rules
and
consider
codifying
something
from
another
like
w3c
or
what
WG
have
done
in
the
past?
Where
that
some
number
of
implementations
is
the
prerequisite
for
standardizing
something
right
like
you,
can't
standardize
something
out
of
The
Ether
right,
that's
not
how
HTML5
works.
G
So
that
way,
it
is
still
a
trailing
specification,
but
that
implementations
may
Advance
the
state
of
the
art
and
oci
may.
After
some
number
of
limitations,
Implement
some
spec
change,
we
could
standardize
it
I
I
think
that
may
be
a
way
to
breach
the
impasse.
A
And
there's
nothing
in
what
we're
doing
that
says:
a
registry
can't
go
through
and
Implement
whatever
they
want
to
create
a
new
media
type
new,
manifest
type
they've
done
that
already
and
Implement
and
deploy
this
stuff
out
there.
It's
the
whole
point
of
us
is
to
try
to
come
forward
and
say:
let's
make
stuff,
that's
compatible
with
each
other.
D
A
A
C
G
I
I
think
it's
worth
asking.
You
know
we're
we're,
adding
the
refers
API
and
subject
Fields,
we're
choosing
to
add
features
to
oci,
1.1
and
I.
Think
I
think
it's
clear
that
it's
okay
for
a
registry
to
stay
on
oci
1.0
right!
It's
like
that!
That
is
fine,
I
I,
don't
see
why
we
should
let
that
in
Paris
from
clarifying
the
spec
or
branded
to
your
point,
you
said
that
we
wanted
to
copy
over
the
artifacts
repository
into
this
repo
before
we
merge
999.
G
G
F
A
In
the
artifact
markdown
file.
A
C
Okay,
this
one
was
on
the
agenda
again,
one
that
I
opened.
It
just
needs
another
lgbtm
and
sign
off.
This
is
fixing
stuff
that
was
missing
from
the
Json
schema.
C
C
Yeah-
and
this
is
seems
like
best
effort-
it
really
is
just
a
type
of
a
string,
but
there's
no
straightforward
way.
The
field
is
called
okay,
so
I'm
defining
an
object,
called
base64.
E
C
Those
I
I,
don't
like
Jason
schema
and
it's
complicated
enough
that
I
think
most
people
here
are
not
using
it.
This
is
just
describing
an
object
which
is
referred
to
here
to
say,
use
the
definition
for
this
thing
called
base64
that
we're
defining
literally
right
there
with
that
string,
which
is
a
type
of
string.
So
largely
this
is
just
checking
that
this
here
is
base64
is
in
fact,
a
string
technically
through
this
Json
schema
validation
test.
C
I
I
did
a
little
bit
of
Google
Foo
and
it
apparently
certain
certain
specific
Json
schema.
Frameworks
will
also
check
this
media
type
as
well
to
see
that
it
has
a
certain
content,
encoding
I,
don't
think
ours
does
that
I'm
not
going
to
get
into
the
conversation
of
changing
chasing
schema,
validation,
Frameworks.
C
So
if
somebody
were
to
validate,
then
it
might
also
validate
that
it
is
in
fact
correct
base64.,
but
in
the
meantime,
the
best
that
we
can
do
with
the
the
most
common
thing
is
just.
Is
it
a
string?
Not
an
integer,
not
an
array,
not
a
list,
not
a
dictionary,
it
is
a
string
and
this
is
in
fact
a
string.
C
E
E
C
The
the
conversation
kind
of
sidewinded
a
good
deal,
and
then
it
started
off
as
just
being
a
defining
of
a
digest
to
be
usable
and
then
to
make
it
easy
for
folks
who
are
probably
already
implementing
importing
this
specs.go
for
some
of
the
data
structures
or
image
media
type.
C
You
know
Concepts
that
this
could
be
an
easy
one-liner
for
folks
who
are
putting
stuff
in
and
then
whether
or
not
to
have
the
media
type
as
a
argument
or
just
have
something
that's
default
like
this
one
and
I
I
can
go
back
and
change
this
to
being
back
as
an
argument
that
just
seems
kind
of
like
whatever
I
get
it,
that
there's
a
conflict
in
the
media
type
field
and
archetype
artifact
type
field.
C
It's
just
kind
of
like
easy.
If
people
wanted
to
have
a
one-liner
of
like
here's,
this
descriptor
for
an
object
that
I
actually
don't
care
about,
and
then
they
could
manipulate
it
after
the
fact
if
their
workflow
is
to
set
that
media
type
as
a
field
I
mean
obviously
the
object
it
returns
here
is
not
immutable.
You
could
call
this
function
and
then
change
the
field.
C
You
know
that
immediately
no
media
type
field
and
add
your
archetype
failed
if
needed
be,
but
at
least
in
the
discussion
of
having
something
that
is
valid
Json
every
time-
and
this
is
the
agreed
upon
object,
blob
whatever,
then
it's
something
that
could
be
usable
for
either
the
config
and
or
the
single
singular
thing
in
the
list
of
layers.
C
That's
all
so
it's
kind
of
fish
tailed
around
a
little
bit
and
there's
some
comments
here.
Like
I,
agree.
There's
it's
this!
This
all
feels
a
little
silly
for
me,
but
at
least,
if
I
made
this
media
type
a
config
image
config,
then
it's
then
this
is
now
an
invalid
data
structure
and
the
only
thing
that
we
really
were
originate.
This
whole
conversation
originated
with
is
just
having
an
agreed
upon
Shaw
that
if
anybody
saw
this-
and
it
is
a
const
on
our
end-
hey
somebody
sees
this
shot.
256,
some
and
I'm
sure.
C
C
A
A
Yeah
my
own
take
on
this
one
is
that
we're
mixing
input
and
output,
and
so
you're
your
input
is
usually
going
to
be.
Hey.
I've
got
this
sha.
What
is
the?
What
is
the
actual
data
that
we're
going
to
get
back
from
the
registry
and
vice
versa
and
I?
Don't
think
you're
going
to
be
any
places
where
you
can
use
the
descriptor
as
a
comparator?
A
Maybe
it's
useful
if
someone
wants
to
have
something
that
automatically
generates
it
for
your
code,
but
I
think
as
you
just
subscribed,
you
still
have
to
go
through
and
customize
the
media
type
for
each
one,
because
it's
going
to
be
different
for
every
implementation
out
there,
because
this
is
going
to
be
the
media
type
for
the
artifact
type.
Everybody's
got
a
different
artifact
type
so
for
for
the
implementation,
I
threw
up
there
as
made
in
birth.
I
C
Feelings
about
this,
this
is
just
another
compromise
of
like
trying
to
enable
people
to
not
be
completely
hung
up.
E
F
F
D
So
it's
just
forgotten:
the
disagreement
is
about
having
it
in
the
go
files
versus
because
in
the
top
we
do
follow
that
the
blob
digest
off
open,
girly
braces.
Is
there
and
that's
the
shot?
The
specification
calls
it
out,
as
that
is
what
a
sketch
layout
is.
So
it's
just
a
matter
of
having
that
either
the
function
or
the
constants
is
the.
C
I
I
F
A
A
C
So
as
a
as
a
quick
nod
because
I
like
some
of
the
examples
that
you're
doing
there
and
there's
basic
overlap
with
this
one
yeah
we're
basically
saying
drop
the
function,
keep
the
consts
and
if
anything
else,
maybe
a
word
clarifying
that
this
is
not
actually
const,
not
actually
not
to
be
confused
with
the
scratch
layer.
A
From
scratch,
rather
I'm
going
to
pick
a
different
name
instead
of
scratch,
I've
already
decided,
but
okay
yeah,
not
not
because
I
I
dislike
what
you
name
it
just
because
I'm
I've
seen
too
many
people
confuse
swarm
mode
with
swarm
classic
and
I.
Have
a
I.
Don't
want
to
relive
that
one.
But
yeah!
Look
looking
through
this!
The
other
big
thing
I
did
was
you
you
define
what
a
scratch
config
is,
but
I
wanted
to
define
the
entire
artifact.
C
C
A
C
Zero
yeah,
so
I
I
have
not
actually
I
just
know
that
in
the
two
weeks
of
the
1023
was
open,
that
there
was
some
quick
iterations
on
it.
It
went
through
three
different
changes,
exactly
what
you
just
described
of
why
you
went
with
what
you
went
with
we
already
passed
through
that
and
went
back
to
the
function.
C
A
B
A
B
C
C
And
if
the
removal
of
the
if
99
goes
through,
then
is
there
a
working,
Branch
I
think
that's
the
only
other
thing
like
what
are
these
cleanups?
That
would
actually
suffice
for
the
absence
of
the
artifact
manifest
and
then,
if
artifact
manifests,
is
a
strict
note.
You
know
like
if
it's
not
a
strict.
No,
if
it's
a
next
then
like
actually
see
it
in
a
branch.
A
I'm
pushing
for
that,
it
sounds
like
you
agree
with
that
too
Vincent
I've
heard
from
others.
There
go
with
that.
Yep,
okay,.
B
My
I
mean
my
only
reservation
is
just
you
know
the
nickel
and
diming
for
years.
I
don't
know
if
I
want
to
come
back
and
subject
myself
to
you
know
Life's
too
short.
So
if
people
want
to
carry
that
forward,
maybe
they
could
do
it,
but
like
just
judging
on
this
conversation
today,
can't
imagine,
there's
going
to
be
any
less
nickel
and
dining
on
everything
there.