►
From YouTube: OCI Weekly Discussion - 2023-05-11
A
Foreign
there's
chatter
and
slack
about
canceling
since
there's
a
bunch
of
open
source
Summit.
Is
there
anything
anyone
wants
to
talk
about?
Well,
we
are
here.
B
Yeah
I
was
considering
bringing
up
some
stuff
around
wasm.
B
Sure
I
can
give
it
a
go.
So
today
we
have
a
lot
of
stuff
going
on
around
getting
wasm
into
the
Container
runtime
stack,
some
of
that
like
we
have
project
Ron
Wazi
that
is
basically
containerdi
shim,
that
executes
wasm
from
container
images
and
right
now.
Basically,
what
we
do
is
we
look
at
the
configured
entry
point
and
assume
that
that's
a
was
a
module
and
compile
it
and
execute
it
and
passing
whatever
args
or
whatever
were
given
to
it.
B
This
is
this
has
a
lot
of
things
that
are
not
ideal,
mainly
well,
not
even
just
mainly
there's
several
really
big
things.
One
is
that
these
images
ideally
would
be
platform
agnostic.
However,
because
of
the
way
things
need
to
execute
and
like
how
you
can
run
something
on
Windows
like
it
needs
a
specific
file
system
layout
and
on
Linux,
you
need
might
need
things
like
ca
certificates
and
all
this
kind
of
stuff.
B
So
we
basically
have
this
module,
that
is
platform
agnostic,
but
the
tooling
around
it
is
to
be
able
to
execute.
B
It
is
not
so
we
end
up
we're
kind
of
in
a
situation
where
we
need
to
have
platform
specific
images
for
platform,
agnostic
executables,
and
we
would
definitely
like
to
be
able
to
change
that
and
with
that
also
be
able
to
do
things
a
little
more
efficiently
than
what
you
would
normally
do
in
a
traditional
image
where
we
know
so
with
wasm
you're
going
to
have
a
module
or
multiple
modules
and
what
I
think
what
we've
been
discussing.
B
B
So
you
can
imagine
something
like
a
change
to
the
image
spec
that,
instead
of
so,
it
could
have
layers
like
it
has
now.
But
then
a
new
field-
I,
don't
know
if
it
would
just
be
for
wasm
or,
if
there'd
be
some
other
use
case,
but
something
like
a
modules
fields
that
that
has
all
the
similar
to
the
way
the
root
of
s
layers
are
set
up
in
the
image
config,
but
for
modules.
It's
just
okay.
B
So
kind
of
just
wanted
to
get
people's
take
on
that
and
if
it
makes
sense
to
work
on
a
proposal
for
updating
the
spec
to
support
that
or
potentially.
A
B
A
It
looks
like
the
OS
field.
There
was
like
a
special
text
in
here
here:
I'll
just
send
what
I'm
looking
at
right
now
about
the
OS
field,
being
special
paced
for
Wazi?
B
I
would
say
sort
of
in
the
sense
that
you
would
choose
the
behavior
based
on
what
the
OS
is.
So
if
it's
Wazi,
then
okay,
let's
look
at
what
modules
are
are
in
the
spec
and
do
it
do
what's
needed
for
that,
whereas
if
it
was
Linux
like
it,
probably
doesn't
even
make
sense
to
have
that
extra
field.
So
it
would
just
kind
of
be
ignored
or
validated
that.
C
B
B
Yes
exactly
and
then
in
that
case
like,
if
we're
able
to
do
this,
then
we
can
just
have
a
a
Wazi
image
that
is
just
Wazi.
It
can
run
anywhere
versus
today.
You
kind
of
need
to
do
some
funny,
funky
things
at
runtime
like
infer
certain
things
about
the
spec
and
do
different
things
based
on
platform
or
push
different
platform,
specific
images
with
your
wasm
in
it,
so
that
the
file
system
layouts,
are
correct
and
has
whatever
linuxisms
or
whatever
are
needed.
B
Yeah
so.
B
In
in
terms
of
like
an
old
run
time
being
able
to
so
so
today,
so
that
would
you
would
have
a
a
spec
that
says:
oh,
this
is
Wazi
and
you
can
either
support
Wazi
or
not
support
Wazi
and
of
course,
today
everything
is
still
moving
and
whatever
we
do
like
it's
going
to
break
some
prior
implementation
that,
but
it's
all
kind
of
proof
of
concept
at
the
moment
so
yeah.
Basically
it
would
just
be.
You
can't
run
if
you
can't
run
Wazi,
then
don't
try
to
run
an
awazi
image.
C
My
off
the
cuff
gut
take
would
be
that
you
should
try
it
out
and
see
how
well
it
works,
and
then
we
look
at
standardizing
it,
especially
because
unknown
fields
are
defined
by
the
spec,
as
should
be
ignored.
Essentially
so
yep
yep
or
you
could
try
something
like
an
annotation
to
just
get
basically
I'm
trying
to
help
navigate
into.
We
have
proof
that
this
this
works
and
is
stable
and
does
solve
the
problem,
mostly
comprehensively,
so
that
we
don't
have
a
lot
of
spect
churn.
While
we
try
to
figure
out
the
solution
to
the
problem,.
C
B
That
makes
sense
yeah.
A
All
right
shall
we
call
it
unless
there's
anything
else.