►
From YouTube: OCI Weekly Discussion - 2020-03-25
Description
https://hackmd.io/El8Dd2xrTlCaCG59ns5cwg?view#March-25-2020
chat log:
```
00:09:03 Vincent Batts: https://hackmd.io/El8Dd2xrTlCaCG59ns5cwg?edit
00:22:28 Josh Dolitsky: Here’s the table w/ new html reports: https://github.com/bloodorangeio/oci-distribution-conformance-results/blob/split-tests/README.md
00:23:41 Josh Dolitsky: I would very much like to see a diagram of these axises :)
00:24:01 Samuel Karp: ECR seems to have more tests enabled when you click through to the test report, so what's in the table right now seems incorrect.
00:24:43 Josh Dolitsky: @sam this report? https://oci-distribution-conformance-results.s3.amazonaws.com/ecr-split-tests.html
00:25:32 Samuel Karp: @josh sorry, I somehow clicked on a different link and was confused.
00:25:40 Josh Dolitsky: Master has the old one yea
00:25:53 Josh Dolitsky: Or “current” one
00:30:27 Josh Dolitsky: https://github.com/bloodorangeio/distribution-spec/blob/split-tests/conformance/README.md
00:30:36 Josh Dolitsky: https://github.com/bloodorangeio/distribution-spec/blob/split-tests/conformance/README.md#pull
00:30:42 Josh Dolitsky: https://github.com/bloodorangeio/distribution-spec/blob/split-tests/conformance/README.md#push
00:30:50 Josh Dolitsky: https://github.com/bloodorangeio/distribution-spec/blob/split-tests/conformance/README.md#discovery
00:30:57 Josh Dolitsky: https://github.com/bloodorangeio/distribution-spec/blob/split-tests/conformance/README.md#management-delete-etc
00:40:36 Steve Lasker: https://github.com/opencontainers/distribution-spec/issues/62
00:41:20 Mike Brown: https://github.com/opencontainers/distribution-spec/pull/111
00:51:50 Josh Dolitsky: Can we have a Steve Lasker icon
00:53:27 Josh Dolitsky: Link to icons pdf: https://github.com/opencontainers/tob/files/4383085/opencontainerinitiative_icons_V2.pdf
```
B
B
A
A
F
F
A
Good
to
see
everybody
on
the
call
I
know
a
lot
more
hey
some
ways.
This
everybody
staying
at
home
has
been
a
little
bit
more
status,
a
little
bit
of
status
quo,
but
it's
been
kind
of
nice
to
have
like
social
social
chat.
Some
involved
interaction,
so
everybody
here
and
honestly
I
probably
am
pretty
open
to.
A
If
we
just
have
like
open
hours
for
people,
if
they
want
to
be
able
to
chat,
there
is
points
for
what,
when
were
working
through
things,
I'm
keen
for
that
I
know,
I've
done
it
for
some
of
the
teams
internally
and
some
of
communities.
Externally,
I've
been
working
with
over
the
last
couple
weeks,
just
to
have
like
an
open,
open
line
like
a
video
chat
like
this.
A
If
folks
want
background
noise
or
peer
program,
pair
programming
or
whatever
I'm
game,
so
it's
an
open
invitation
just
reach
out,
so
we'll
go
ahead
and
jump
into
things
Josh.
You
are
the
first
item
on
the
agenda
again
that
Morgan
stuff
lots
of
lots
of
good
reviews
that
name
no
and
you're
not
audible
either
so.
D
But
what
we've
done
now
is
we've
changed
it
to
run
four
major
categories,
polish
discovery
and
management,
and
so
now
the
tests
have
been-
and
this
is
still
not
up
for
a
review.
Yet
we're
still
working
out
some
of
the
kinks.
But
the
higher
level
of
the
test
is
now
those
those
four
categories
and
it's
the
the
objective
is
that
if
you
pass
pool
you're
considered
conformant
with
the
speck,
and
so
maybe
we
need
to
update
the
speck.
If,
if
everyone
agrees,
we'd
update
the
speck
to
reflect
that.
D
But
this
is
now
so
I
shared
this
in
the
HEC
nd
under
the
topic.
But
this
is
a
branch
of
that
OCI
blood-orange
URL
called
split
tests
and
that's
the
same
name
as
the
branch
for
the
new
test
source.
If
you
want
to
check
that
out,
but
I've
added
a
column
here,
tests
enabled
so
explicitly,
you
can
now
say
that
you
want
to
run
specific
tests.
So
here's
here's,
how
that
looks.
D
But
we
know
like
this
is
what
at
least
what
I
understood.
Derek
to
that
docker
hub.
Would
support
so
I
put
this
year
and
I
know
from
talking
to
Steve
they're
targeting
support
for
everything,
but
everything
else
I
put
pool
and
if
you
open
up
here's
an
example
of
what
the
report
updated
report
looks
like
so
now
we
have
a
skipped.
D
We
have
a
categorization
for
skip
tests
and
that
can
be
no
matter
how
high
that
number
is.
The
test
could
still
pass
as
long
as
there's
zero
failing
and
now
all
the
tests
have
been
kind
of
segmented
into
their
own
sections
and
if
they're
not
run
everything's,
just
gray,
and
so
still
all
the
registries
are
not
passing
the
test.
So
maybe
we
need
to
update
the
tests,
but
yeah
I'm,
just
trying
to
kind
of
like
get
feedback
on
this
approach
and.
D
There's
there's
some
other
idiosyncrasies
too,
like
for
Poole,
like
if
you're
only
enabling
Poole,
we
have
this
setup
section
and
the
setup
section
tries
to
populate
their
registry
with
a
single
blob
and
a
single
manifest.
But
if
you
only
if
you
really
only
let
get
and
Head
request
your
registry,
this
is
not
gonna
pass,
and
so
there's
a
few
other
extra
environment
variables
that
you
can
set
like
OCI
tests,
blob
manifest
or
blob
sha,
and
you
can
give
a
registry
to
the
test.
D
D
Well
so,
let's,
let's
say
like
there's
a
registry:
that's
completely
read-only
and
we'd
consider
that
compliant,
but
the
tests
are
sort
of
like
destructive,
so
they
the
setup
tries
to
push
stuff
into
it.
But
we
have
a
thing
that
checks.
If
you
provide
the
sha
of
the
blob
and
provide
the
tag
name
of
the
manifest
or
whatever
it
is
Peter
Peters
on
the
call
to
he
can
kind
of
walk
through
this
more.
F
I'm
asking
a
based
on
that
I'm
asking
a
different
question
like
there's.
Certainly
some
registries
that
are
pull
only
pretty
sure.
Ibm
has
one
that
we
have
the
Microsoft
container
registry,
where
no
customer
is
ever
going
to
push
to
it.
I
don't
know
if
it
even
needs
to
be
tested
per
se,
because
it's
not
really
a
customer
registry.
F
So
but
there's
like
docker
hub.
Are
there
any
registries
that
would
be
tested
that
fall
in
the
category
that
you
wouldn't
be
able
to
push
to?
Not
the
public?
Obviously,
but
what
I'm
getting
it
is.
Is
the
test
really
that
I
shouldn't
I
don't
enable
push
but
I
enable
push,
but
the
repo
has
to
be
created
and.
H
I
think
there's
configurations
that
you
might
want
to
test
here.
I
mean
you're,
just
testing
all
the
registries
and
assuming
that
you
have
full
control
over
configuring,
the
registries
but
yeah
the
the
categories
you
you
mentioned.
They're
kind
of
the
registries
are
acting
more
as
they
kind
of
read.
Only
cash
or
kind
of
some
sort
of
proxy,
so
I
mean
we
won't
be
able
to
test
those
configurations
without
having
to
rely
on.
You
know
what
the
push
stuff
up,
but
I,
don't
know
of
any
like
software.
That
explicitly
only
supports
it.
That's.
F
D
One
of
the
things
we
ran
into
is
that
there's
I
think
three
different
types
of
pushes
pushes
and
so
I,
don't
think
all
registries
have
full
support
for
all
three
of
those
pushes
and
maybe
and
and
I
think
so
we're
considering,
if
you
don't
have
support
for
all
three
of
those.
You
don't
have
support
for
push,
at
least
in
the
current
stage.
C
F
F
Okay,
I
hadn't
really
thought
about
that
level
of
detail,
but
is
there
what
I'm
getting
at
without
going
into
much
detail?
Anyone
is
like
we
talked
about
these
high
level
categories
and
like
sub
features
underneath
them
so
there'd
be
push
and
then
to
be
like
three
bullets
underneath
in
this
case
and.
F
B
F
Up
the
speck
too
and
find
the
gaps
that
guy
just
I
clicked
on
the
ACR.
One
I
noticed
that
where
we
were
returning
a
slightly
like
a
plus
one
on
the
response
code,
I'm
gonna
look
at
and
figure
out
why
we
can
go
fix.
Those
like
those
are
should
be
easy
things
if
there's
other
things
that
are
bigger
than
let's
figure
out
what
those
are
yeah.
B
H
Reality
is
there's
there's
a
large
grey
area
of.
Is
this
if
I
have
a
client
that
is
the
OCI
compliant,
can
I
talk
to
this
registry
or
not?
And
if
we're
just
trying
to
you
know
give
registries
like
a
gold
star
to
say
like
you're
compliant,
then
you
know
that's
one
thing:
we
need
to
strictly
defined
what
compliance
means,
but
in
terms
of
like
whether
or
not
this
registry
is
going
to
be
able
to
work
as
an
OCI
registry,
that's
another
thing
like
the
three
different
types
of
pushes,
like
many
clients
can
handle.
F
There
was
something
we
discussed
in
the
data
we
may
or
may
not
want
to
support
and
we'll
leave
it
up
to
a
registry
and
that
you
know
I
like
using
the
repo
creation
as
an
example,
because
it's
just
I
think
everybody
agrees.
It's
a
perfectly
valid
thing,
but
I
like
I'm
sure
ECR
doesn't
want
to
be
listed
for
poll
only
like
they
want
to
have
the
other
ones
listed.
F
So
what
how
do
we
do
that
and
even
to
the
point
of
the
status
rating
I'm
pivoting
a
little
bit,
but
it's
an
exam
of
how
I'm
thinking
that
those
four
categories
are
listed?
Is
each
one
of
those
four
has
its
own
pass
or
fail.
So
it's
a
little
more
interesting
that
have
you
know
four
pass
or
fails
individually
for
push-pull
discovery
management?
Then
there's
one
thing
missing
in
the
whole
thing,
and
the
whole
thing
is
a
big
red
X.
E
2,
this
is
Romney
so
to
expand
on
what
direct
said
right
so
I
think
a
good
summary
would
be
on
the
x-axis.
We
have
all
the
registries
on
the
y-axis.
We
have
these
work
flow
tests,
but
there
is
a
z
axis
on
which
there
are
clients
and-
and
real
work
gets
done
when
we
find
some
sort
of
an
intersection
on
all
these
three
axes.
F
Is
it
really
a
client
like,
wouldn't
the
clients
be
have
their
own
matrix
on
how
they
test,
how
they
compare
to
the
OCI
spec
like
be
kind
of
cool,
to
have
a
three-dimensional
printout,
and
we
all
give
everybody
goggles
to
do
this,
but
is
it?
Is
that
really
what
we're
trying
to
do
here?
We're
saying
this
is
the
registry
compliance
list
and
that
there's
a
registry
client
test,
then
we
could
do
that
matrix
as
well.
Well,.
H
I
guess
I'd
like
to
separate
a
little
bit
like
these
are
just
conformance
tests.
These
are
telling
what
works
and
what
doesn't
work
on
a
registry,
but
it's
not
trying
to
define
whether
or
not
a
registry
is
compliant
like
if
we
want
to
have
conformance
test
and
go
through
and
have
like
some
list
of
compliance.
Like
honestly,
I'm
not
too
concerned
about
the
compliance
I
want
these
tests
to
be
as
kind
of
detailed
as
possible
so
that
we
can
push
it
against
or
we
can
point
it
at
the
registry.
H
F
F
Not
have
to
tweak
the
test
so
much
so
they
can
figure
how
to
get
it
to
green,
but
the
reality
is
a
client.
That's
trying
to
interact
with
this
thing.
It
doesn't
support
some
particular
operation,
so
it's
gonna
fail
like
I'm,
hoping
that
we'll
all
get
to
green,
not
just
because
we're
trying
to
get
to
green,
but
because
the
east
everything
I,
see
here
with
maybe
one
or
two
differences
or
all
people
that
are
actively
involved
in
this
and
we're
not
just
trying
to
get
our
eyes
to
green
for
marketing
reasons.
F
H
Guess
it's
more
like
Green
is
different
than
just
like.
Is
it
working
and
yes,
I
think
in
the
future?
We
love
it
if
every
Registry's
OCI
its
passing
all
the
conformance
tests,
but
like
it's
just
not
the
kind
of
a
fantasy
to
get
to
that
point,
I
mean
we've.
We've
had
this
specification
out
for
a
long
time.
We
don't.
This
is
the
first
time
we're
actually
writing
a
performance
test
and
pushing
it
out.
H
But
what
we
want
I
think
for
from
this
perspective
is
like
previously
the
standard
that
registry
operators
were
coding
against
was
docker,
which
itself
wasn't
fully
implemented
in
the
spec
and
not
always
doing
it
correctly,
so
that
we
have
to
have
a
conformance
test.
That's
fully
utilizing
the
spec
correctly,
so
that
this
itself
is
something
that
registry
operators
can
go
and
test
against
and
having
as
much
data
about
you
know,
what
doesn't
work
is
I.
Think
really
useful
here,
like
in
terms
of
like
green
and
red,
and
you
know
some
things
having
all
these.
H
F
B
F
Delete
the
till
I
don't
know
if
delete
falls
in
management,
but
like
the
delete.
One
is
a
great
conversation
for
us.
We
should
have
if
it
hasn't
already
happened,
a
good
breakout
on
how
do
we
want
to
handle
tags
and
manifests
what
what
do
we
think
the
spec
should
actually
say,
regardless
of
what
we've
each
implemented?
What
do
we
think
it
should
if
there's
work
for
us
to
do,
then
we'll
go
off
and
do
that
work
and
if
there
isn't
there
isn't
but
having
it
just
be
this
big
open
bag?
F
F
H
D
D
D
D
C
F
D
F
Just
maybe
okay,
then
just
pop
it
onto
this
page.
So
if
I,
here's
what
each
one
of
these
mean
and
because
I
wouldn't
have
thought
that
pole
is
three
different
categories
and
management.
I
can
guess
what
that
means,
but
and
I
think
by
continuing
to
iterate.
Through
that
definition,
we
can
figure
out
what
are
the
gaps
and
I.
Think
there's
the
point
that
I'm
trying
to
push
is.
We
know
we
want
to
make
changes
to
the
spec.
D
A
D
Know
which
peelers
been
open-
oh
no,
I'm,
sorry
I,
might
have
misled.
I
meant
we're
happy
to
open
a
PR
against
the
spec
MD
that
we
think
would
be
sort
of
like
an
overhaul
and
categorization
of
these
things
and
making
certain
endpoints
required
versus
optional.
But
what
I
was
saying
is
I
have
a
feeling
when
we
open
that,
if
we
don't
discuss
it
first,
it
will
be
open
for
a
long
long
time.
Oh
I
just
know.
A
That's
one
of
those
things
that
I'm
actually
been
itching
for
some
kind
of
a
working
session:
I'm,
okay,
sometimes
so
a
TD
that
I
have
already.
It
helps
me
just
to
have
folks
working
on
stuff
at
the
same
time
and
can
get
out
the
axes
and
that's
yeah.
What
kind
of
leads
into
the
other
bullet
point
I
have
of
like
wanting
to
get
a
release
candidate
to
out
the
door,
because
I
mean,
like
the
other
day,
I,
never
look
at
the
rebate
in
Dion.
A
B
F
F
D
So
not
to
cut
you
off
see,
but
how
about
how
about
this
will
we'll
take
a
pass
at
the
markdown
file
open,
a
PR
that
I
don't
I?
Think
from
there?
Maybe
events
we
can
have
the
working
session
about
the
overhaul
I
like
to
have
no
but
I'm
just
saying
to
have
that
first
pass
out
there
versus
going
into
a
working
session
or
we
or
we
can
do
a
working
session
to
open
the
PR,
whichever,
if
you
think
that
you.
A
Can
make
a
first
pass
on
it
I?
Just
it's
sometimes
a
little
bit
daunting,
because
it's
especially
when
I
look
at
like
the
return
codes
and
how
its
formatted
and
Derek
can
vouch
to
this
of
like
effectively
the
current
content,
that
is,
there
has
been
through
several
different
iterations
of
types
of
markdown
and
HTML
and
embedded
tables
and
whatnot.
So
sometimes
even
just.
A
Aldo
generation-
that's
true,
so
some
of
what
you're
seeing
it's
just
like
you
know
is
it
invented
as
a
which
layer
you
know
like.
Would
it
make
more
sense
to
just
have
like
an
error
code
section
and
you
know
I
mean
just
have
you
know
any
given
method
say
it
uses
just
this
this
and
it
just
like
link
down
to
one
unified
section.
Would
that
be
cleaner?
A
Probably,
like
you
can
definitely
a
more
than
one
file
you
know
like
we
could
have
like
an
Aires
MD
and,
like
everything,
would
just
link
over
to
that
one.
The
the
benefits
and
drawbacks
are
that
right
now
every
section
has
kind
of
a
clunky
like
we
have
an
overview
and
then
a
detail
and
the
detail
has
a
lot
of
copypasta
and
so
breaking
it
out
be
nice.
I
am
two
ways
about
trying
to
generate
a
PDF.
A
A
F
So
I'm
just
trying
to
break
this
up
into
something
manageable
like
there
is
the
format
which
is
what
we're
talking
about
here,
which
is
always
the
it
doesn't
change
the
content
per
se,
but
the
format
we
definitely
want
to
clean
that
up.
But
are
you
finding
that
there's
just
information
missing
like
this
API
is
not
clear
about
what
response
codes.
It
should
have
that's
what
I'm
more
trying
to
drive
towards
well.
H
Some
cases
the
formatting
is
actually
led
to
inconsistencies
because
I
said
there's
like
the
overview
in
the
detail.
We've
had
many
cases
where
those
weren't
this
so
every
time
we
split
something
up.
If
it's
not
linking
to
one
source
of
truth,
you
we
end
up
with,
like
the
speck
saying
two
different
things
based
on
what
section
you're
at
so.
H
H
F
A
Identified
some
of
what
you
know,
I
got
Josh
and
Peter
right.
You
know
that's
what
they're
seeing
like
trying
to
group
like
seeing
so
much
red.
Are
they
are
they
generally
consistent
and
where
they're
failing
so
that's
that's
to
me
is
like,
if
they're,
if
they
wrote
the
test
of
the
spec
and
everybody's
failing
it
consistently.
You
know
consistent
ways.
Then
we
need
to
update
this
back.
Yeah.
C
E
F
D
So
me
and
Peter
will
sit
down
virtually
and
we
will
yeah
we'll
come
up
with
a
bullet
list
of
everything
that
was
kind
of
hard
to
grok
and
from
there
proposed
so
we'll
come
up
with
the
issues
and
then
a
proposal
also
of
how
we
would
address
it
and
I
think
it's
done.
It
I
think
like
Vince's
idea
of
like
breaking
it
into
multiple
files
and
doing
this
type
of
thing,
I
think
is
the
answer
and
I
really
don't
think
it's
too
far
off
like
it's.
D
H
A
Sure
cuz
it's
it's
like,
okay,
so
the
for
those
watching
at
home.
It's
the
pull
request,
one
one,
one
on
the
distributions
back
it's
it's
obvious.
This
is
clearly
not
ready
for
being
merged.
Just
a
discussion
piece,
but
the
topic
here
is,
like
we've
talked
about
so
far
having
some
way
that
we
could
have
additive
things
to
the
distribution
or
registry
API
to
not
mess
with
all
the
stuff.
A
That's
currently
being
done
a
part
of
like
core
conformance
but
might
want
to
add
later
like
it's
come
up
in
the
notary,
v2
stuff
about
being
able
to
say,
show
me
all
the
signatures
that
have
you
know,
signed
or
attested.
This
particular
containers
digest,
or
something
like
that.
So
you
could
have
like
various
extensions
added
to
the
registry.
A
A
So
the
current
approach
that
is
up
for
discussion
here
is
to
have
like
an
extension
directory
where
there's
a
readme
with
basic
layout
information,
namely
having
a
table
that
would
enumerate
the
extensions
that
have
been
merged
into
this
OCI
repo.
This
kind
of
falls
outside
the
versioning.
So,
as
these
things
might
get
added
later,
it
doesn't
bump
the
verdict,
OCI
version
per
se
and.
A
It
would
adding
a
new
extension
would
be
adding
a
markdown
document.
You
know
having
a
PR
with
the
markdown
document
in
a
certain
pattern
so
that
it
would
be
would
describe
the
name.
So
you
could
see
it
here
in
this
table
of
its
key
and
the
documentation
explaining
that
key.
So
when
you
hit
that
API
on
the
server,
you
could
get
a
list
of
those
key
names
and
know
correspondingly
what
documentation
it
is
about.
H
So,
do
you
see
a
client
in
this
case
say
your
say:
you're
a
notary,
YouTube
client,
and
you
want
to
see
what
the
registry
support
signatures.
The
first
thing
you
would
do
is
do
ahead
requests
on
this
MD
file
for
the
signature
extension
and
see
if
the
registry
supports
that,
what
do
you
see
that
so
I
guess.
A
A
But
effectively
they
would
do
a
head
a
head
or
a
get
I
guess
on
that
/v.
So
whatever
your
URI
prefix
path,
prefix
is
up
to
that
point,
but
/b
and
then
you,
you
Derrick,
preferred
something
like
underscore
ext,
slash
and
so
on
that
route
path
of
that
ext
would
return.
Maybe
a
dictionary
or
an
array.
I,
don't
know
what
it
would
enumerate
the
names
of
these
things.
It
could
be
a
dictionary
with
more
information.
H
A
You
know
at
some
point
that
maintainer
of
that
thing
says
we
screwed
up.
You
know,
then
they
could
make
a
SIG's
dot,
v1
or
whatever
six
dot,
v2,
and
so
that's
kind
of
one
of
those
like
versioning
is
hard,
but
it
would
all
be
that
prefix,
if
it's
just
officially
deprecated
and
they
move
on,
then
registry
maintainer
could
choose
to
like
not
have
to
keep
around
the
old
one.
I
don't
know
that
that
is
a
commitment
and
a
decision
to
be
made,
but
I
don't
think
that's
anything
new
to
anybody.
H
So
my
comment
related
to
the
underscore
that
was
just
I
mean
that
was
kind
of
a
something
we
decided
early
on.
We
knew
that
was
going
to
be
a
problem,
so
when
we
added
other
registry
wide
end
points,
they
had
to
start
with
that
underscore
because
that's
something
that's
disallowed
in
the
in
the
namespace
of
the
deco
repositories,
but
I
mean
I.
Think
it's
fine
with
what
you
described.
I
I,
think
that's
perfectly
fine
to
do.
H
H
A
All
right
so
is
it.
Does
everybody
have
any
pressing
questions
on
it?
Otherwise,
so
it's
a
good
a
go
to
the
PR
I
had
that
commentary,
and
probably
within
the
next
few
days,
I'll
polish-
that
up
into
something
that's
actually
because
I've
got
like
X
X,
X
markers
for
to
be
to
be
determined
stuff,
but
mainly
it's
just
a
conversation
piece
yeah.
A
Looks
like
the
next
topic
is
really
Chris.
Insects
topic,
but
Lasker
you've
had
I
think
it
was
actually
originally
your
ask.
Chris
a
had
some
graphics
created
for
various
use
and
places
and
there's
a
link
there.
The
TOB
issues,
number
65,
that's
open
for
comment,
I
think
we're
on
version
2.
Now
Lasker,
you
have
anything
you
want
to
say
about
them,
not.
F
I
was
just
that
and
I
think
just
a
matter
of.
Did
we
capture
all
the
is
two
things
one?
How
do
people
feel
about
the
individual
images?
I
I
think
this
round
two
looks
awesome.
I
had
the
same
question
around
the
pipeline.
One
was
the
only
one
that
I
really
thought
was
something
to
tweak
a
little
bit
more
source
artifact,
but,
more
importantly,
because
we
can
continue
to
tweak
on
individual
ones.
F
Are
there
any
ones
that
people
feel
that
we
should
add,
because
once
you
it's
kind
of
like
having
a
contractor
in
the
house,
you
there's
somebody
actively
working
on
this.
It's
easy
for
them
to
crank
out
another
one.
Once
the
project
ramps
down
and
I
wish
I
had
one
more
icon,
it's
a
lot
more
to
get
it
spun
up.
There
is
a
PR
there,
they've
been
super
helpful
in
coming
back.
F
A
G
You
know
the
idea
of
official
ones
that
our
designers
have
grown
the
problems,
gonna
drop,
something
for
the
OCR
and
that
since
I
think
it
makes
sense,
and
hopefully
I,
don't
think
this
worked,
its
price
go
for
the
OC
ice
and
we
can
use
this
for
CN,
CF,
stuff
or
whatever
you
know
it
could
be
viable
for
a
lot
of
future
uses
where
we
need
generic.
You
know
imagery
for
architecture
diagrams
or
whatever
in
the
future,
so
cool.
A
In
the
meantime,
we
actually
had
a
few
other
good
cleanup
and
fixes
on
like
a
few
more
PRS
merged
for
the
runtime
spec
that
were
now
that
people
were
like.
Oh
we're
really
still
gonna
happen.
Have
some
good
cleanup
for
revisiting
on
things
so
put
up
a
vote
for
the
runtime
spec
this
week
to
close
next
week,
but
already
we
have
four
out
of
six
thumbs
up
on
it.
So
that's
two-thirds!
So
I'll,
probably
you
just
go
ahead
and
tag
that
release
this.
We
can
get
that
out,
notably
for
all
those
keeping
score.
A
This
one
version
102
includes
the
stuff
about
hooks
the
cleanup
of
hooks
conversation,
which
was
needed
for
the
Nvidia
runtimes
and
Runcie
Forks
and
stuff
like
that,
because
that
that
was
one
of
the
last.
You
know
unresolved
conversations
for
Runcie,
while
they
were
holding
off
on
a
run
c
version.
100
everybody's
been
running
a
release
candidate
of
run
c
for
untold
number
of
years.
At
this
point,
so
we
might
actually
see
a
one
c
one
dot
o--
this
year,
which
would
be
curious.
A
So
that's
that
so
nothing,
no
real
action
needed
for
anybody.
Just
letting
you
know
the
other
piece
like
alluded
to
is
when
josh
talked
about
earlier
I'd
like
to
see
a
distribution,
spec
release
candidate
happen,
we
were
sitting
at
a
bar,
C
1,
which
is
just
something
a
milestone
for
people
to
iterate
at
a
lot
has
happened
since
then,
and
I
think
we
should
start
the
March
now
towards
like
cleanup
of
knowing
that
that
this
is
the
trajectory
we're
on.
A
So
it
kind
of
clean
up
the
working
session
that
josh
has
talked
about
when
you
hit
the
web's.
You
know
when
you
hit
the
page,
and
you
see
the
readme,
that
this
looks
like
a
skeleton
copy
feel
free
to
open
PRS
for
that
kind
of
stuff,
that's
kind
of
what
I
started
expecting
to
see
and
release
candidates.
It's
just
like
the
polish
and
feedback
at
polish
type.
A
A
A
A
D
A
A
Cool
again
I'll
maintain
opening
by
for
folks
it
if
you
want
to
have
life
at
a
working
session
or
even
just
get
the
background
noise.
If
we
have
some
kind
of
working
session,
oh
I'll,
post
it
to
the
dev
at
list,
and
so
even
if
you're,
just
wanting
to
hear,
hear
a
nice
job
eavesdrop
on
the
conversation
that's
time,
but
I
know
that
a
lot
of
people
are
working
from
home
that
are
not
used
to
working
in
remote,
so
feel
free
to
reach
out.