►
From YouTube: CNCF Serverless Working Group 2021-04-08
Description
CNCF Serverless Working Group 2021-04-08
C
Yeah
I've
been
I've
been
really
busy
so.
B
B
C
B
B
B
B
B
B
B
D
D
B
E
B
So
lance
I
did
ping
grant.
I
don't
think
I
actually
did
it
till
yesterday
or
maybe
the
day
before.
I
came
here
for
sure
to
remind
him
that
he
agreed
to
sort
of
talk
to
the
the
commit
tagging
stuff
or
whatever
you
guys
call
it
if,
for
some
reason
he
doesn't
show
up,
do
you
think
you
could
talk
to
that
to
give
a
quick
overview
of
how
it
all
works
and
what
we're
supposed
to
do,
or
would
you
prefer
to
maybe
wait
till
next
week
since
I
catching
you
off
guard,
I
can
do
that.
B
Okay,
kristoff.
B
E
F
B
All
right,
three,
after
let's
do
this
thing:
okay,
we
got
20
people
all
right,
ai's,
clemens,
not
here,
okay,
community
time,
anything
from
the
community.
People
want
to
bring
up.
That's
not
on
the
agenda
me.
E
F
Yeah,
very,
very
quick,
as
you
already
know,
I'm
working
on
the
implementation
for
the
expression,
language
and
I'm
going
to
extract
from
it
a
kind
of
tck,
so
adjacent
containing
several
cases
and
the
expected
results.
So
what
I
was
wondering
is:
where
should
I
place
it
so
is?
Should
I
place
it
like
in
the
spec
pr
in
the
spec,
repo
or
or
where.
F
F
B
Let
me
go
ahead
and
stop
sharing
there,
you
go,
you
should
be
listening.
B
E
F
Choose
whatever
you
want?
Okay,
just
pick
the
last
one:
okay
and
yeah.
These
are
test
cases.
Each
json
is
the
test
case.
It
contains
the
schema
and
then
it
contains
the
test
with
the
expected
results.
I
want
to
create
something
serial,
so
description
of
the
test
case,
the
expression,
the
input
event
and
then
the
test
result.
B
Interesting
looks
kind
of
cool.
My
initial
reaction
is
this
sounds
very
similar
to
like
the
conformance
work
that
I
know
scott's
trying
to
get
going.
So
I'm
wondering
whether
this
would
be
best
suited
for
a
brand
new
repo.
That
way
you
can
do
more
than
or
you
could
basically
do
whatever
you
wanted
in
there
and
and
have
complete
freedom.
F
F
That
could
be
fine
too
well.
The
thing
is
that
we
already
have.
You
know
the
grammar.
I
I
already
pr
the
grammar
in
the
spec
repo.
F
Maybe
we
should
have
a
folder
like
in
the
spec
reaper,
which
contains
the
spec
and
the
test,
the
spec,
the
grammar
and
the
test
cases
like
like
a
bit
of
the
bitter
reorganization
of
the
repo,
because
in
the
reaper
we
also
have
like
for
each
subspec.
We
have
everything
in
the
root
of
the
repo
so
like,
for
example,
for
the
subscription
spec.
We
have
the
everything
of
the
repo
I
mean
it
might.
Maybe
maybe
it's
time
to
to
order
a
bit.
The
repo.
B
Yeah,
I
know
that
that's
been
in
my
to
do
this
for
a
while.
I
actually
was
hoping
to
get
to
it
this
week,
but
I
never
did
so.
Okay
yeah
tell
you
what,
since,
since
creating
a
whole
new
repo,
does
have
some
overhead
associated
with
it.
Let's
do
this,
let's
let
me
go
ahead
and
take
the
action
item
to
create
a
pr
to
reorganize
the
repo
and
then,
if
it
turns
out
a
folder,
is,
is
annoying
or
too
big
or
whatever.
Then
we
could
look
at
splitting
out
onto
a
separate
repo
later.
F
And
I
think
it
makes
sense
to
have
this
in
to
fit
inside
the
the
process
that
we
use
to
merge
things
inside
this
repo,
because
you
know
it's
good
that
people
takes
a
look
at
the
tck
and
check
if
there's
something
that
doesn't
make
sense.
So.
D
B
All
right
this
week
we
do
have
the
sdk
on
call
after
this
one.
A
Oh,
yes,
hi
scott.
What's
up
hey,
so
I
did
some
work
on
the
conformance
tool
this
week
and
you
can
now
send
binary
and
structure
events
and
there's
a
rod.
Dump
mode
so
go
check
out
the
cli.
It's
it's.
It
mostly
works.
Give
it
a
try.
C
B
A
A
A
There
are
a
couple
breaking
changes,
sort
of
we
we've
had
to
move
some
packages,
they're
technically,
not
breaking
they're,
technically
breaking
okay,
but
we
chose
to
not
bump
it
to
three
v3,
because
they're
they're
very
specific
niche
usages.
B
D
Well,
it
you
know
it's
it's
the
initial
input
into
initial
commit
into
the
repo,
so
your
mileage
may
vary,
but
the
team's
going
to
be
updating
it
and
should
be
useful
for
people
that,
like
to
use
powershell
yep.
B
B
B
There
you
go,
there's
a
link
to
it
cool
all
right,
moving
forward.
We
do
have
an
sc
case
called
schedule
for
this
week.
As
of
right
now,
I
think
there's
nothing
on
the
agenda.
So
please
add
an
item.
If
you
want
to
discuss
something
otherwise,
we'll
cancel
the
call
interop.
We
didn't
really
have
much
of
a
call.
Last
week
there
has
been
a
whole
lot
of
progress.
B
So
let
me
just
take
this
opportunity
to
nag
people
if
you've
promised
that
you
or
if
you
said
you
were
interested
in
participating,
we
did
say
we're
going
to
start
beginning
of
april.
So
please
put
up
your
endpoints
in
that
document,
so
people
can
start
doing
some
testing.
Let
us
know
your
status,
which
things
may
or
may
not
work,
so
we
know
what
we
can
hit
and
what
we
can't
so
just
consider
this
being
nagged
appropriately.
B
All
right
moving
forward
just
reminder:
I
did
sign
up
for
two
45-minute
sessions.
There's
the
times.
Please
add
your
name
here.
If
you
can
join
remy
did
upload
his
video,
so
I
think
it's
probably
too
late
to
make
any
changes
to
it.
But
if
you're
interested
in
seeing
the
video
or
the
slide,
that
did
include
the
links
here
and
he
so
he's
doing
our
cloud
event
session
for
us,
okay
and
tmr.
Anything
you
want
to
mention
from
the
workflow
stuff.
G
Just
more
sdk
stuff,
we
added
a
net
sdk
and
a
typescript
sdk
and
both
were
community
contributions,
so
that
was
really
cool
and
other
than
that
yeah.
We
I
picked
our
sessions
at
kubecon
very
late,
so
we
got
those
two
in
the
so
yeah
we're
gonna
work
on
that
and
make
sure
we
have
all
the
content
for
that
ready,
cool.
C
All
right
any
question:
oh
okay,
yeah
does:
does
the
net
workflow
sdk
depend
on
the
cloud
events
sdk
dot
net
sdk,
because
if
so,
in
fact,
I
was
just
just
writing
an
issue
for
the
powershell
sdk
saying:
okay
you're,
depending
on
the
the
latest
ga
of
the
net
sdk.
That
makes
sense,
but
it
would
be
really
good
if
everyone
could
be
ready
for
when
we're
ready
to
do
the
2.0
which
will
have
breaking
changes
for
net.
C
G
Yeah
definitely,
I
think
right
now
that
the
workflow.net
sdk
does
not
depend
on
the
the
cloud
events
one,
but
that
would
be
really
nice
actually
to
kind
of
combine
them
together
on
both
the
the
java
and
the
go
and
net
and
and
and
I
will
definitely
try
to
join
the
sdk
meetings
for
those
discussions.
If
that's
okay
with
you
guys.
B
Cool
all
right,
any
other
questions
for
team
are
on
the
workflow
stuff.
B
B
H
It's
easier
to
get
there:
okay,
yeah,
so
one
pr
that
we
added
in
the
master
branch
is
to
add
a
github
action
that
gives
a
green
check
mark
if
there's
conventional
commits
right.
H
The
goal
of
this
is
to
especially
for
folks
that
don't
actively
look
at
the
spec
every
week,
and
so
you
want
an
overview
it's
to
enforce,
or
at
least
encourage
folks
to
use
conventional
commits
when
they're
writing
their
commits,
and
so,
if
we
go
maybe
to
like
fix
some
typos,
maybe
the
first
pr
there
so
and
we
go
press
the
red
x
under
the
fix
or
or
we
look
at
that
yeah
under
the
details.
H
You
can
see
that
the
linter
has
ran
and
the
commit
message
is
like
not
a
conventional
commit.
It
says
the
subject
may
not
be
empty
and
the
type
may
not
be
empty,
and
so
the
fix
here
would
be
to
prefix
with
the
type
like
a
docs
colon
and
fix
a
few
typos.
H
And
so
then,
when
we
get
a
lot
of
these
commits
in
our
spec
folks
can
just
look
at
well.
They
serve
can
ignore
the
the
like
doc
changes
that
don't
really
affect
anything
and
then,
if
there's
a
feature,
it'll
have
the
feet
prefix
and
then
then
you
can
look
over
commits
more
easily.
That
way.
B
Okay,
slinky
has
his
hand
up.
Just
like
you
have
a
question.
F
Yeah,
to
be
honest,
I
don't
I
I
don't
understand
why
we
need
this,
for
the
very
simple
reason
that
our
git
flow
is:
we
do
prs,
doug
merged
them,
so
at
the
end
of
the
day,
is
doug
that
squashes
the
pr
and
types
the
actual
commit
message
comment
title.
So
I
I
think
what
we
need
is
just
to
have
doug
adjust
any
the
automatic
message
generated
by
github
when
squashing
prs
and
merging
them.
H
Yeah,
so
that
sounds
fine
so,
and
that
I
mean
the
overall
goal
is
when
a
developer
looks
at
the
spec.
Maybe
I
mean
a
lot
of
developers,
don't
look
at
it
every
week
or
even
every
month,
but
they'd
still
like
to
understand
what
are
the
top
changes
in
the
repo?
H
I
What
I
found
myself
having
to
do
is
you
know,
as
I'm
iterating
on
comments
on
the
pr's
I'm
having
to
do
squashes
every
single
time,
and
so
I
wasn't
sure,
if
is
it
just
the
first
commit
in
the
chain
that
some
that
this
rule
applies
to
or
every
single
commit
that
goes
through
and
you
know,
is
there
an
expectation
that
either
I
then,
when
we're
ready
squash
the
whole
thing
and
give
it
to
doug
as
a
single
item
or
the
maybe
a
slinky
said
that
doug.
I
Does
it
at
the
time
that
he
does
the
merge,
because
I
mean
I,
I
think
it's
it's
it's
sort
of
affecting
the
way
that
I'm
I
tend
to
work
iteratively.
So
if
people
put
some
comments
in
I'll,
do
a
you
know,
I'll
do
another
tweak
and
a
commit
and
a
push
to
address
those
comments.
So
you
sort
of
get
this
backlog
of
changes
which
eventually
you
want
to
squash,
whereas
I
think
this
seems
to
think
that
every
commit
I
do
could
be.
The
final.
B
H
Let's
get
point
I
think:
maybe
we
want
to
change
the
commitment
just
to
run
when
on
the
master
commit
it
shouldn't,
affect
anybody
just
developing
and
trying
to
create
a
pr.
H
Or
or
well,
I
guess
I
mean
at
least
the
goal
is
to
the
the
only
real
change
is
the
the
squash
commit
at
the
end,
and
that's
really
where
we
care
about.
If
this
passes
or
not
cool.
B
Okay,
so
so
let
me
ask
a
question:
if,
because
I
actually
had
a
similar
thought
process
to
what
I
think
lance
was
saying
in
the
chat,
which
is,
I
thought
we
were
doing
this
to
auto-
generate,
in
other
words
like
release,
notes,
kind
of
stuff
right,
your
change
log
right
and
we
don't
have
that
process
defined
in
here
or
something
I
was
wondering,
one
weather
grant.
B
B
This
change
log
stuff,
I'm
sorry
this
this
this
conventional,
commit
stuff
to
apply
to
every
single
pr
or
just
prs
that
we
think
are
significant
enough
to
warrant
mention
in
the
release.
Notes.
H
H
If
we
start
using
conventional
commits,
then
we
can
just
auto
generate
the
change
log
and
yeah.
A
further
pr
would
introduce
the
auto
generated
change.
Log.
B
H
B
Okay,
because
I'd
like
to
get
a
sense
from
the
group
in
terms
of
the
process
everybody
wants
to
follow,
because
if
we
want,
if
we
want
to
be
consistent,
then
it
seems
like
every
pr
should
follow
the
same
process
and
while
I
don't
necessarily
mind
doing
extra
work.
When
I
hit
that
when
I
hit
the
merge
button,
I
can
do
that
people
want.
It
also
makes
me
a
little
nervous
that
I
would
be
the
one
editing
or
typing
the
title.
B
That
appears
in
the
release
notes
and
I
kind
of
feel
like
it
should
be
the
author,
who
at
least
does
it
themselves,
and
maybe
I
can,
I
could
fix
a
wording.
You
know
a
typo
kind
of
thing,
but
I
I'm
not
sure
I
should
be
the
one
that
comes
up
with
it
as
opposed
to
the
author
themselves,
because
there's
a
good
possibility
to
get
it
wrong.
So
it
makes
me
a
little
nervous.
J
Yeah,
a
couple
of
things
so,
first
of
all
there
are
there,
are
tags
for
commits
that
are
are
meant
to
be
like
not.
This
is
this
is
not
significant
for
the
changelog.
I
think,
but
I
think
my
convention
chore
is
is
one
of
those
that
generally
is
just
kind
of
ignored
by
the
changelog
generation
tools.
J
So
if,
if
we
want
to
be
consistent
about
having
every
commit
be
conventional,
we
can
just
you
know,
have
a
type
of
chore
or
something
that's
that
is
used
for
you
know
things
that
are
not
significant
to
the
other
question
about
whether
we
want
so
kind
of
having
the
commit
author
being
the
or
the
the
pr
author
driving
the
these.
J
I
guess
the
text
of
the
of
the
line
in
in
the
in
the
change
log.
I
think
that's
the
that
was
kind
of
the
point
of
having
a
lyncher
like
this
is
to
kind
of
cause
or
force
the
the
pr
authors
to
participate
in
that
process
and
to
get
in
the
habit
of
writing
a
conventional
commit
such
that
their
understanding
that
this
is
you
know
this
is
going
to
go
into
the
change
log,
and
so
I
should
think
about.
J
You
know
how
this
should
appear
and
write
the
the
commit
message
accordingly,
so
is
it
I.
I
think
that
was
kind
of
the
point
of
it
is
a
tool
to
kind
of
force
us
to
think.
In
those
terms.
J
So
that
said,
if
there
are
issues
with
like
well,
I
want
to
add
more
commits
to
my
pr
and
they're.
Not
you
know,
and,
and
now
it's
forcing
me
to
to
change
my
workflow-
I
don't
know
if
there
might
be
ways
to
to
convince
to
configure
the
the
linter
in
certain
ways.
I
I
think
by
default.
It
might
assume
that
every
every
commit
needs
to
be
needs
to
be
conventional.
J
But
if
we're
gonna
squash,
maybe
there's
a
way
to
to
to
configure
it
to
only
check
the
first
one,
because
I
think
if
you
have
more
than
one
commit
then
by
default,
git
uses
the
the
the
title
of
the
pr
rather
than
the
commit
messages
itself
as
the
default
or
as
a
starting
point
for
your
your
final
squash
commit
name.
So
there
might
be
ways
to
to
configure
it
to
to
know
that
hey
we're
gonna
squash,
let's,
let's
optimize
for
that,
workflow.
B
Okay,
thank
you.
Slinky
you're
up
next.
A
Hey
so
I
you
know,
based
on
every
other
project
we
work
on,
but
we
use
a
convention
where
we
we
use
a
special
tag
block
and
then
kubernetes
has
this
release
notes
tool
that
goes
and
looks
at
every
commit
that
landed
in
that
that
time
frame
and
then
extracts
that
that
little
block,
plus
potentially
labels
that
were
applied
to
the
pr
and
so
it's
it
has
nothing
to
do
with
the
commits
and
the
commit
messages
in
the
squashing.
It
has
everything
to
do
with
github
and
the
description.
A
Maybe
we
would
like
to
do
that
because
it's
a
little
easier
for
the
both
the
the
ci
process
that
we
use
in
github
and
poor
doug
to
not
have
to
rewrite
commit
messages
on
smashing,
because
this
is
so
here's
the
text
of
what
the
commit
conventional
commits
uses.
So,
basically,
it
says
it's
doug's
job.
B
A
Yeah,
that's
right.
I
could
maybe
next
week
I
could
show
a
little
demo
and
we
have
some
some
wrapper
github
action
tools
that
help
you
kind
of
like
extract
a
markdown
file
based
on
two
hashes.
F
And
and-
and
I
want
to
add
that
it's
even
simpler
for
you
for
the
users,
because
if
we
have
a
template
for
the
issues,
we
will
encourage
people
to
write
the
release
notes.
So
it's
even
better
because
we
kind
of
force
the
user
to
do
that.
A
B
Okay,
anyway,
so
much
time
in
there.
Otherwise
I'm
going
to
mention
a
proposal,
then,
okay,
so
scott,
you,
you
asked,
or
you
volunteered
to
to
share
that
process
with
us
on
next
week's
call.
So
why
don't
we
go
ahead
and
do
that
and-
and
I
agree
with
you-
daniella
yes,
we'll
modify
the
contributing
doc
if
we
even
have
one
or
create
one
if
necessary,
but
yeah
whatever
process
we
come
up
with
we'll
make
sure
we
document
it
someplace.
B
If
we
don't
think
they're
worthy
of
enough
to
mention
that,
in
the
release
notes,
I
can
just
go
to
emerge
those
and
not
worry
about
the
bot
complaining.
However,
if
there
is
one
that
needs
that
we
think
should
be
in
the
release
notes,
what
I
want
to
do
is
make
sure
people
understand
what
things
should
look
like.
So
grant
is
I'm
trying
to
figure
out.
Is
this
a
good
example
of
what
the
commission
message
should?
H
Yeah,
exactly
so,
you
add
a
prefix
of
the
type
and
optionally
a
scope,
and
so,
as
daniel
mentioned,
chore
ones
will
not
be
added
to
the
change
log.
And
so
we
can
sort
of
ignore
those.
But
significant
ones
like
new
features
or
fixes
can
have
those
prefixes.
B
B
Okay,
so
before
we
jump
into
the
prs-
and
I
will
merge
this
one
now-
slinky
any
other
topics-
you
want
to
bring
up
before
we
jump
into
pr's-
I
got
a
nice
long
list
today,
all
right,
let's
go
back
to
slinky,
because
it
is
the
slinky
show
today,
oops.
B
All
right
back
to
this
one,
I
believe
a
couple
people
asked
for
this
one
to
be
held
off
a
little
bit.
I
think
I
might
have
been
one
of
them.
I
personally
I
I
did
do
some
thinking
about
this,
but
I
haven't
changed
my
nervousness
about
it,
but
I
also
don't
want
to
say
block
it,
since
we
can
always
change
it
again
later,
so
I'm
okay
with
this
going
in
for
now,
but
what
other
people
think
about
this
one?
K
Yeah,
just
commenting
that
last
time
we
we
had
this
discussion
about
how
other
how
database
processed
this-
and
I
did
a
few
experiments
on
sql,
fido
and
basically,
oracle
sequel,
server
and
all
the
major
ones
they
throw
an
exception.
But
I
feel,
like
I
think,
was
my
sequels
simple
and
a
few
others.
They
basically
returned.
No,
so-
and
I
know
we
didn't
have
this
discussion
of
whether
returning
no
was
a
valid
option.
F
To
be
honest,
I
strongly
disagree
with
that
thing
now.
It
adds
a
whole,
more
complicated
set
of
problems
that
I'm
trying
to
avoid
as
much
as
possible.
So
I'll
be
honest
with
you
and
if
we
want
to
add
null
it's
not
just
about
adding
nil
for
this
specific
program,
it's
about
adding
null
in
the
whole
spec
and
language
design,
which
is
the
same
thing
I
I
I
explained
for
the
error
for
handling
errors.
This.
F
This
is
a
consequence
of
how
and
lingeros
is
defining
this
language
and
I'm
very
open
to
change
it
or
modify
it.
But
then
that's,
I
think,
if
we
want
to
do
that,
we
need
to
bring
up
the
discussion
at
a
more
higher
level.
So
how
do
we
end
the
errors
in
the
language.
B
B
That's
that's
thing
that
keeps
running
through
my
mind,
but
I
don't
want
it
to
be
just
one
vote
either
way
kind
of
thing.
Okay,
daniella
says
I
prefer
not
to
change
it
until
we
have
a
clear
benefit
of
returning
something
different
than
zero.
B
B
F
B
Okay,
let's
move
on
to
the
next
one
then
clarify
in
semantics.
I
think
this
was
there
last
time,
yeah.
F
So
so
this
one
we
had
the
discussion,
whether
other
other
databases
act
on
the
typecasting
when
using
the
in-
and
I
I
in
the
conversation
I
wrote
my
findings,
I
tried
with
my
sequel
and
oracle
sequel
I
think
and
yeah.
It
should
be
fine
this
way.
So
this
way
is
more.
Like
other
sequels,
let's
say,
but
can
you
open
the
conversation?
F
Because
if
I
call
correctly
on
this
one
oracle
sql
behaves
differently
from
mysql
this
one
yeah
yeah
yeah
yeah,
because
oracle
sql
doesn't
have
the
boolean
literal
stormfalls
and
in
my
sequel
the
implicit
type
casting
is
only
valid
for
binaries,
but
not
for
numbers.
Here
I
may
be
doing
some
mistakes
here.
F
To
be
honest,
so
at
the
end
of
the
day
they
are
not
very
the
same
between
rockport
sequel
in
my
sequel,
so
I
just
chose
to
to
go
with
whatever
seemed
reasonable
to
me
and
and
for
the
user
of
my
sequel,
which
is
just,
of
course,
always
the
types
to
the
left
argument
type.
So
that's
it
and
that's
that's
what
I
wrote
in
this
pack
so.
B
B
Done
easy,
okay,
cool
john.
C
Yes,
so
this
is
replacing
a
fairly
brief
couple
of
paragraphs
with
significantly
more
detail,
attempting
to
squash
out
any
ambiguity
and
also,
hopefully
make
life
a
little
simpler
and
more
reliable.
C
So
it's
encoding
and
decoding
http
headers
that
are
meant
to
represent
cloud
event
attributes
and
it's
basically
saying
perform
percent
encoding
on
anything.
That's
not
ascii
or
is
a
space
or
is
percent
for
obvious
reasons
or
is
a
double
quote,
and
the
reason
for
encoding
spaces
and
double
quotes
is
so
that
we
don't
then
need
to
perform,
quoting
as
per
rfc
7230,
but
sdks
still
must
perform
decoding
via
rfc
7230
for
compatibility
with
previous
versions
of
this
spec.
C
So
my
hope
is
that
it
is
compatible
and
it
looks
like
the
comments
so
far
have
been
encouraging
on
that
front.
No
one
said
it
fails
in
this.
This
respect.
I've
tweaked,
a
couple
of
bits
of
wording
have
folks
had
a
chance
to
look
at
it
as
much
as
they're
likely
to
all
feedback.
Welcome
any
comments.
B
C
And
I'm
definitely
while
I
would
obviously
like
to
get
this
in
partly
so
that
I
can
get
the
c-sharp
implementation
in,
although
if
if
the
noises
are
encouraging,
I'm
happy
enough
to
merge
the
c-sharp
implementation
in
a
sort
of
this
looks
like
it'll
go
in,
but
I
definitely
don't
want
to
rush
this.
Let's
get
it
right,
rather
than
fast
right.
B
Okay,
so
tell
you
what?
Let's
do
that,
because
this
is
this
is
a
significant
change
so
hold
on
we'll
vote
next
week,
whoops.
C
B
B
And
normally
I
do
this
stuff
no
later
than
tuesday,
and
just
this
week
just
got
away
from
me,
so
I
hopefully
after
next
week,
my
schedule
will
get
a
little
bit
more
normal
and
I'll
be
able
to
get
back
to
getting
these
things
done
timely.
So
I
apologize.
Actually,
I
suspect
most
of
these
are
not
too
new.
So
forget
this
I'll.
Just
remove
that.
Okay,
let's
go
back
to
slinky
then
because
we
like
hearing
from
slinky.
B
F
It
yeah
it's
so
this
pr
is
basically
putting
some
constraints
on
how
how
bad
you
can
mess
up
your
engine,
your
expression,
language
engine,
so
I'm
basically
explaining
that
the
overloading
is
allowed
but
yeah,
but
there
is
limit
on
how
you
can
overload.
F
There
is
a
sentence
that
is
not
completed
because
of
inclusive
casting
null
functions
with
the
same.
Oh
okay.
I
need
to
fix
that
anyway,
and
so
he
put
some
constraints
on
overloading
and
on
bariatric
functions.
So
I'm
just
being
more
explicit
on
on
how
how
you
can
define
custom
functions,
that's
basically
or
on
how
the
spec
can
define
functions
and
how
engine
can
allow
to
define
custom
functions
so.
B
B
B
Yeah
okay,
hold
on
hold
for
one
typo
cool
okay
can
do.
B
C
How
do
we
go
about
versioning
cloud
events,
and
this
is
a
topic
I'd
been
working
on
internally
within
google
and
obviously
had
strong
opinions
because
I
tend
to,
but
it
felt
like
something
that
deserved
to
be
in
the
spec
repo,
so
we
created
an
issue
here
discussed
it
about
a
month
or
so
ago,
and
I
came
up
with
this
proposal,
which
I
hope
reflects
the
conversation
from
then
now,
given
that
it's
a
month
ago,
I
suspect
that
almost
everyone
has
forgotten
what
we
actually
said,
but
basically
this
this
suggests
that
the
the
type
should
be
used
as
a
sort
of
normative
way
of
saying
do
not
expect
these
two
to
be
backwardly
compatible
because
they
are
different
versions
and
you
know
you
can
define
a
producer
can
perform
their
own
use
of
semantic
versioning
or
not.
C
However,
they
want
to
do
that.
But
one
thing
I
would
like
to
be
checked
within
this
is
the
part
around
the
data
schema
attribute.
So
I
have
asserted
without
any
particular
evidence,
other
than
gut
feeling
and
previous
conversations
exactly
this
paragraph
that
you're
looking
at
now
around
data
scheme
or
attribute
is
expected
to
be
informational.
C
So
I
can
see
it
being
really
useful
for
tools
that
let
you
inspect
cloud
events
dynamically,
but
my
guess
is
that
most
code
that
is
consuming
a
cloud
event
won't
use
the
data
schema
at
all.
It
will
expect
things
to
be
as
they
were
defined
when
when
the
code
was
written
or
compatible
with
that,
so
please
shout
if
I
have
completely
misunderstood
the
point
of
data
schema.
I
don't
think
I've
had
any
comments
on
this
one
at
all.
Yet
I
could
be
wrong
again.
C
I'm
not
in
this
is
even
less
of
a
hurry
to
to
get
merged.
If
folks
could
take
some
time
to
have
a
look
and
see
whether
I've
written
complete
nonsense
or
not
that'd
be
great
right.
Cool
yeah.
B
B
Just
let
me
be
even
more
forceful
with
that
question,
because
I
that
is
something
I've
often
wondered
myself,
because
I
tend
to
agree
with
you
john.
I
tend
to
think
that
people
tend
to
talk
about
a
phrase.
Sort
of
statically
create
the
code
that
they're
expecting
to
come
in
and
the
idea
of
dynamically
changing
what
you're
going
to
get
and
analyzing
the
schema
at
runtime
it
just
I
did.
I
don't
know
how
often
you'll
actually
do
that.
B
I
always
had
a
gut
feel
like
you
that
they
probably
don't
it's
a
little
more
static
than
that,
but
I'd
love
to
know.
Does
anybody
on
the
call
have
the
opposite
opinion
that
there
are
lots
of
cases
where
no
people
actually
do
analyze
the
schema
at
runtime
beyond
a
superficial,
schema
checker
thing:
do
people
actually
use
it
for
anything
real.
C
I
I
can
imagine
some
cloud
events
with
particularly
dynamic
data
where
the
the
cloud
event
provider
producer
doesn't
really
specify
the
schema
themselves.
They
just
pass
on
data
from
something
else,
so
you
could
have
10
different
events
all
with
their
own
different
data
scheme
or
attribute,
but
that
would
be
for
a
very
particular
kind
of
cloud
event.
I.
L
Okay,
okay,
you
can
speak
up
for
a
moment.
I
that
I
was
going
to
speak
up
and
then
you
made
that
little
quibble
about
using
the
schema
for
validation
purposes
and
I've
been
writing
node
for
a
while
and
because
the
untightness,
I
there's
a
whole
lot
of
code
that
I
don't
have
to
write.
If
I
use
the
schema
validator
to
basically
say
this,
this
message
that
I'm
receiving
meets
exactly
the
requirements
that
I
have
for
that
data,
and
so
I
don't
think
I
don't
see
the
scheme
as
informational.
B
So
john,
that
sounds
like.
Maybe
this
paragraph
here
might
be
then
a
little
too
loose
or
something
because
it
doesn't
take
into
account
what
eric
was
just
mentioning
there,
which
is
well.
They
may
not
necessarily
do
anything
like
dynamically.
Create
objects
on
the
fly
because
of
it
like
that,
but
they
people
may
use
it
for
schema
validation
at
one
time.
So
it
isn't
strictly
a
development
type
thing.
C
Yeah
and
I'm
fine
with
it
being
used
at
execution
time,
although
I
would
have
expected,
if
you're,
trying
to
validate
that
the
data
you've
received,
is
reasonable,
validating
that
against
what
the
event
itself
says
is
reasonable
feels
like
it
doesn't
buy
you
very
much
it's
sort
of.
Do
you
say
that
you're?
Okay,
you
do.
C
Oh,
that's
all
right,
then
I'm
being
overly
flippant,
but
I
would
expect
it
to
be
more
likely
to
validate
against
a
schema
that
you
have
previously
that
you
knew
about
at
build
time,
but
I
could
be
really
wrong.
I
would
be.
I
would
love
to
hear
more
in
written
comments
that
I
can
then
try
to
inwardly
digest,
if
possible.
L
B
F
B
B
B
With
separator,
thank
you
lou,
that
makes
sense
doi
we
missed
the
obvious
there.
All
right,
just
double
check
anybody
just
see
anybody
have
a
anybody,
have
an
issue
with
approving
it.
B
Thank
you
slinky
this
one.
F
B
F
Well,
well,
there
is
one
semantic
exchange
which
is
now
you
is
bull
returns.
True,
if
you
pass
a
boolean
as
a
value
and
then
enter
when
you
pass
an
an
integer
returns,
the
identity
bull
when
you
pass
a
boolean
returns
the
identity
string
when
you
pass
a
string,
returns
the
identity.
B
B
F
B
M
I
I
you
know
partly
did
this
as
a
mental
exercise,
but
there
was
an
outstanding
issue,
so
it
does.
You
know
it
does
address
that,
take
it
or
leave
it.
I
think
that's
what
it
is.
I
mean
this
is
it.
It
shows
why
it
shows
a
mechanism.
You
know
that
you
can
express
cloud
events,
as
you
know,
in
xml
format.
If,
if
you
want
to
do
so,
yeah
nobody's
mandating
that
you
have
to
do
it.
C
Every
language
ever
exactly
yeah,
I
suspect
it's
actually
far
more
useful
than
if
we
ended
up
with
a
yaml
format
or
something
like
that
cool
kids,
stuff
yeah.
I
was
pleased
to
see
this
at
which
point
I'll
say
I
haven't
reviewed
it
in
depth,
but
I
was
pleased
to
see
that
it
existed.
B
B
Fine:
okay,
any
questions
off
the
bat.
I
I
mean
there's
a
little
example.
I
think
at
the
end
of
this,
which
sort
of
sort
of
will
give
you
a
glance
into
what
the
actual
event
structure
would
look
like
yeah,
so
yeah.
So
that's
what
that's,
basically,
how
you
can
think
about
it.
I
I
you
know,
I
will
admit
I've
not
written
any
xml
stuff
for
about
10
years,
but
I
do
remember
that
it
was
much
more
efficient
to
be
attribute
heavy
and
you
know,
support
a
more
streaming
style.
I
I
With
that
again,
I
think
this
is
a
constant
battle.
I
have
I
don't.
You
know
this
needs
to
be
humanly
legible,
but
it's
machine
interpreted
yeah,
so
anywhere
that
I
tend
to
try
and
be
as
efficient
as
a
as
I
can.
There's
no
need
for
verbosity
there
that
so
I'm
all
about
bites
on
the
wire,
essentially,
especially
when
we're
talking
about
xml,
which
is
naturally
a
little
bit
verbose
anyway.
So
I
can
show
you
you
know.
B
Yeah,
just
the
the
the
anal
retentive
side
of
me
likes
consistency.
That's
the
only
reason
I
kind
of
was
thinking
about
that.
So
let
me
pick
on
some
folks.
C
I
So
when
you're
street,
if
you're
stream,
processing
yeah,
you
need
to
get
you
know
in
my
mind,
you
need
to
get
the
version
as
early
as
you
can,
so
you
can
then
make
assertions
about
the
stuff
that
follows
it,
that
that
was
my
only
way
of
thinking
I
mean
originally.
I
This
was
also
a
way
of
enforcing
the
version
to
be
present.
I
didn't
I
didn't
go
down
the
road
that
I
did
with
the
protobuf
stuff
of
having
all
the
required
elements
present,
because
it
just
looked
a
little
bit
messy
but
having
the
the
version
stuff,
there
forces
it
yeah
you
just
oh.
C
I
And,
depending
what
sort
of
xml
generation
processor
you're
using
you
couldn't
always
control
the
order
which
those
elements
are
going
to
be
emitted?
Could.
B
I
C
I
D
B
F
B
Yeah,
okay,
so
minor
changes,
oops
good,
golly,
okay,
looking
to
approve
next
week.
B
I
I
We
doug,
if
we
have
a
chance,
could
we
look
at
the
next
one,
because
it's
really
really
minor
and
I
think
john
was
the
guy
who
prompted
me
into
it.
Sure.
I
There
so
just
for
reference
for
people.
This
is
for
some
reason.
I
I
had
a
mental
brain
fart
and,
I
didn't
add,
batch
support
when
we
originally
defined
the
protobuf
format.
So
this
adds
that
it's
a
really
simple
construct,
just
a
a
repeating
set
of
events.
C
John,
does
everything
look
okay
to
you?
It
does
indeed
this
one.
I
have
actually
reviewed
and
I
think,
even
if
I'm
able
to
give
approval,
I
think
I
possibly
probably
did
yeah
basically
gemini
had
the
same
thoughts
at
roughly
the
same
time
and
he
was
good
enough
to
implement
it
instead
of
me
doing
so.
Okay,.
I
B
I
If
we,
if
people
are
willing
to
do
so,
or
you
know,
obviously
just
give
a
thumbs
up
on
the
pr,
if
anyone
has
well.
B
Any
objection
to
approving,
even
if
you
you
know
well,
I
already
asked
if
one
more
time:
okay,
I'm
not
hearing
any
objection,
especially
since
we
have
had
somebody
who
was
who
knows
the
stuff.
Like
john
he's,
okay
with
it,
I
see
no
reason
to
wait.
It's
been
seven
days,
so
one
last
chance
any
objection,
oh
gosh,
okay,
cool!
In
that
case,
we
are
done.
B
All
right
cool
in
that
case,
if
you're,
not
if
you're
a
non-sdk
person
you're
free
to
leave
anybody,
have
any
topics
for
the
sdk
call.