►
From YouTube: Open RFC Deep Dive Meeting - Wednesday, April 22nd 2020
Description
In our ongoing efforts to better listen to and collaborate with the community, we're piloting an Open RFC call that helps to move conversations and initiatives forward. The focus should be on existing issues/PRs in this repository but can also touch on community/ecosystem-wide subjects.
A
Just
wanna.
Note
quickly
note
that
these
calls,
of
course,
are
all
under
the
are
orchestrated
and
operated
under
the
code
of
conduct
that
we
have
listed
in
the
and
referenced
in
the
MPM.
Rfc
is
repo,
and
so
any
conversation
that
we
have
here
and
also
in
those
repos,
we
we
hope,
is
constructive
and
everybody's
thoughtful
and
mindful
of
each
other,
and
in
these
calls
we
we
just
ask
that
people,
you
know,
raise
their
hands
as
usual
when
they
like
to
speak,
and
you
know
these
calls.
A
The
intention
here
is
to
have
a
forum
to
interact
with
community
and
hopefully
move
for
these
conversations
and
they've
been
fruitful.
So
far,
so
we
hope
that
we
keep
getting
more
folks
coming
to
them
before
we
jump
in
and
jump
into
the
actual
agenda.
I'd
love
to
know
if
they
there's
any
announcements
from
folks,
any
goings-on
that
we
should
be
aware
of.
A
If
not
we'll
just
get
into
it,
I
see
German
just
jumped
on
as
well
again,
if
you
want
to
add
yourself
or
follow
along
in
the
hack
MD
document,
I've
pasted
it
there
in
chat
and
also
put
it
into
the
issue
itself
after
the
call.
So
the
topic
today
again
is
resolution.
Slash
overrides
Isaac
I
know:
you've
put
together
the
RFC
and
there's
being
a
little
bit
of
conversation
about
that.
I'm,
not
sure
if
you
want
to
kick
things
off
and
sort
of
go
over
the
nuts
and
bolts
of
what
your
proposed
sure.
B
B
We've
been
exploring
a
lot
of
the
subtler
edge
cases
and
and
making
sure
that
we
can
hit
most
of
the
use
cases
that
we
that
we
care
the
most
about
the
big
one,
that
the
big
kind
of
thing
that's
changing
is
the
statement
in
the
so
to
to
big
things
in
the
in
the
RFC
as
it
currently
stands,
I
just
pushed
just
a
work-in-progress
commit
I,
haven't
finished
up
all
the
implementation
updates
to
it,
but
essentially
we
had
previously.
You
know
in
our
initial
discussion
of
this,
had
talked
about.
B
B
The
way
the
the
the
sort
of
specific
mechanics
of
how
we
merge
together
and
deduplicate
packages
based
on
the
the
nested
override
rules
needed
a
little
bit
more
subtlety
to
it
as
well.
We
identified
a
handful
of
cases
that
would
be
sort
of
reasonably
D
duplicatable,
but
we're
not
going
to
be
deduplicated
in
in
the
the
proposal
as
it
initially
stood.
B
So
on
the
the
one
issue.
There
is
a
case
where
you
have,
let's
say:
I
want
to
set
all
versions
of
react
to
16
and
also
I
want
to
override
one
of
the
dependencies
of
react
16.
So,
in
order
for
that
to
work,
you
have
to
be
able
to
apply
both
a
string
override
to
say
version.
16,
you
know
react
should
be
this
version
and
also
an
object.
Override
that
says.
React
at
sixteen
should
have
these
updates
to
its
dependencies.
B
It's
it's
a
little
bit
kludgy,
but
the
you
know
we
don't
what
we
don't
need
to
do
and
what
I
push
kind
of
pushback
on
was
having
some
way
of
having
a
value
in
the
overrides
object,
which
is
bolt.
You
know,
which
is
like
an
array
of
like
a
string
and
an
object
to
do
like
nested
as
well
as
a
direct
override.
B
B
C
B
B
A
C
It's
worthy
so
I
was
just
gonna
say
though
I
feel
like,
so
this
is
probably
I
would
say
edge
Casey,
so
I
think
we
can
accept
a
little
bit
of
less
than
optimal
things
in
what
we
go
with.
That
said,
it
looks
really
weird
like
as
it
as
like
a
you
know.
If
I
was
seeing
this
really
Beca,
Jason
I'd
be
like
what's
going
on
here,.
B
B
D
B
So
if
you
look
at
multiple
string
value
overrides
there,
there
is
kind
of
an
example
of
that
not
using
star
star
empty
string
and
latest,
which
will
all
be.
You
know
the
same
thing,
but
essentially
the
first.
So
the
first
string
value
that
that
hits
it
does
not
evaluate
any
more
string
overrides
for
that
particular
node.
B
D
B
B
D
D
In
the
case
where,
in
my
specific
momentary
dependency
graph,
it's
irrelevant
because
you're
right,
like
that
catch-all
rule,
might
be
reasonable,
and
maybe
tomorrow
a
new
version
gets
published
that
invokes
it.
What
I'm
saying
is
in
the
case
where
I
have
first
specified
at
star
or
no
version
at
all.
That
means
that
fertilis
they
react.
That
means
that
any
further
mention
of
react
is
a
stream
form
can't
do
anything,
no
matter
what,
and
that
seems
like
an
error
condition
to
me
like
like
no
one
would
ever
be
doing
that
on
purpose.
C
And
unqualified
is
the
same
as
saying
react
between
one
and
two
and
and
then
following
it
with
react
greater
than
one
point
one
and
two:
it's
like
well
yeah.
We
have
overlap,
but,
like
I
mean
there's
tons
of
cases
where
we'll
end
up
having
that
as
a
possibility
do
we
do,
we
need
to
then
go
and
check.
All
of
them
is
that
I.
B
C
D
B
What
this
feature
is
it's
it's
only
in
the
route
package.json
that
it
matters,
and
so,
if
you're,
adding
it
like
you're
you're,
adding
an
override
right
there,
probably
because
you
want
something
to
resolve
a
particular
way
and
you're
probably
going
to
look
at
what
gets
resolved
as
you're,
adding
the
override
right
like
you're
doing
it,
because
I
know
why
am
I
getting
react?
15
here,
oh
I,
see,
let
me
add
an
override
to
solve
that
problem.
I'm,
not
just
gonna
like
add
the
override
and
leave
and
I
am
not
so
sure
that
in.
A
B
I
mean
I
I
hear
you,
but
you
know
this
is
also
like
again,
even
if
even
if
you're
kind
of
like
you
know
in
a
very
you
know,
brute
force
and
somewhat
clumsy
way
or
saying
we
we
have.
We
have
hereby
declared
that
there
shall
be
no
react.
15
at
this
company
and
everything
gets
an
override
added
to
package.json.
That
explicitly
sets
react
to
version
16
everywhere.
B
Well,
that's
either
gonna
break
stuff
or
it's
not
I,
don't
know
and
I
mean
the
the
the
hazard
of
there
being
the
hazard
of
there
being
an
override
role
which
doesn't
get
hit
is
just
it
doesn't
get
hit
and
like
if
that's
a
problem,
you
go
and
look
at
it
and
I.
Don't
feel
like
it's
that
hard
to
unpack
right,
like
oh
yeah,
I,
think
it's.
D
A
defensible
position
to
say
that,
like
if
they're
not
aware
that
there
is
a
problem,
then
it's
probably
fine,
but
my
sensibilities
tell
me
that
people
are
often
have
problems
they're,
unaware
of
and
that
when
tooling
can
surface
that
in
very
loud
faily
ways.
It's
helpful.
That
doesn't
mean
we
have
to
add
this
feature
here,
I'm
just
kind
of
talking
about
that
loud
sure.
B
D
B
The
other
thing
that
that
is
kind
of
weird
or
interesting
is
that
the
the
piece
on
the
left
side
is
not
necessarily
a
name
at
cember
range.
It's
it's
a
named
package
specifier.
So
that
could
actually
be
you
know
a
git
repo
link
or
or
even
a
get
repo
a
link
with
a
particular
commit,
or
a
particular
branch
name.
B
B
Establishing
that
that
one
thing
is
always
going
to
be
overridden
by
another,
it
becomes
even
more
sort
of
complicated
and
I'm.
Just
I
know
that
yeah
we
could.
We
could
say:
okay,
there's
a
special
case
for
star
and
an
empty
string,
because
those
are
literally
identical,
but
then
we
also
have
to
have
like
you
know,
star
empty
string
and
greater
than
equal
to
zero
zero
zero.
B
And
then
you
know
it
just
becomes
kind
of
a
ballooning
set
of
special
cases
if
it
was
straightforward
and
I
think
it
is
somewhat
straightforward
to
say
you
know
if
December,
if
it,
if
they
are
both
number
ranges
and
one
is
entirely
a
subset
of
the
other
and
it
comes
after
the
other
one.
We
could
do
that
we're
just
we're
getting
into
like
Oh
n
cubed
territory
on,
albeit
for
a
relatively
small
end
right,
you're
you're
unlikely
to
have
hundreds
of
overrides
in
a
package.json,
I'm
sure
that'll,
never.
D
B
Or
even
like
a
user
land
tool
to
do
it,
we
have
some
bird
subsets
now
in
the
latest
version
of
node
semper,
so
it
wouldn't
be
impossible
to
just
build
sort
of
a
standalone
thing
that
would
analyze
your
overrides
and
see
if
there
are
subsets
of
numbers
that
come
after
in
the
in
object,
it's
more
it's
more
that
like
doing
it
as
a
special
case
is,
is
inelegant
and
doing
it
elegantly.
I
worry
about
the
performance
and
so
kind
of
says
to
me:
let
the
pundit
for
now.
So
we
can
move
on
past
that
and.
B
Right,
so
that
was
that
was
an
interesting
idea.
We
we
touched
on
it
briefly,
I
think
in
another
select
conversation
and
just
didn't
didn't
end
up
pursuing
it,
but
the
idea
there
for
posterity
is
I
want
to
say
you
know
all
versions
of
foo
at
one
X
should
map
to
my
fork
of
foo.
You
know
NPM
:,
my
fork
/foo
at
the
same
version
number
right
so
maybe.
D
D
B
One
way
that
we
could
tackle
that
is
to
when
we,
when
we
sort
of
resolve
what
the
the
right-hand
side
of
those
string
override
expressions
would
be.
We
can,
we
can,
you
know,
have
some
some
string,
interpolation
or
capturing
that
goes
on.
So,
if
you
do,
you
know
like
dollar,
curly,
brace
version
close
curly
brace,
then
that
will
be
like
whatever
version
number
was
matched
I
see
Wes
kind
of
like
grimacing
at
the
subject.
This.
C
Is
like
the
this
is
the
concern
I
brought
up
earlier,
which
is
like
in
the
initial
conversation.
If
we
start
going
down
this
route,
we're
gonna
end
up
with
something
just
so
complicated
that
most
users
are
gonna,
not
get
it
and
they're
gonna.
You
know,
have
a
lot
of
problems
getting
it
wrong
and,
and
we
ended
up
having
a
lot
of
support,
burden
and
stuff
where
we
you
know,
is
maybe
unnecessary.
B
So
that
doesn't
work
right,
because
the
the
thing
that
we're
replacing
it
with
is
not
a
version
number.
The
thing
that
our
remember
so
the
left-hand
side
is
a
name
package.
Specifier
the
right
hand.
Side
is
an
unnamed
package
specifier,
which
uses
that
has
to
use
the
same
name
right,
because
it's
replacing
in
the
same
spot.
So
if
you
want
to
change
the
package
name,
you
have
to
use
the
alias
specifier
NPM
:
package,
name
at
version.
C
B
B
B
You're
sort
of
suggesting
that
if
we
have,
if
the
right-hand
side
is
an
alias
specifier,
then
we
and
it
doesn't
have
a
version.
Then
we
tack
a
version
number
onto
it.
The
challenge
there
is
it
makes
it
so
that
you
can't
just
override
to
the
latest
and
greatest
version
of
my
thing
right
so
I
might
want
to
say,
like
let's
say,
I'm,
not
publishing,
all
versions,
let's
say
I'm,
just
I
keep
up
to
date
with
the
one
that
we
should
be
using.
C
C
B
Yeah
I
can
certainly
appreciate,
like
the
the
the
user
facing
complexity
of
having
like
capture
groups,
or
you
know,
special,
like
environment
variables,
that
I
get
pasted
in
there
or
whatever
I
mean
we
got
requests
for
for
putting
environment
variables
in
NPM,
RC
and
I.
Think
that's
really
useful
in
a
lot
of
CI
scenarios
where
people
do
like,
you
know
token
equals
dollar
sign
NPM
token,
but
yeah.
B
It
also
leads
to
some
really
kind
of
confusing
cases
when
users
do
try
to
push
it,
so
it
would
not
be
like
every
NPM
config
that
we
just
sort
of
slap
up
in
there.
We
would
just
define
like
look.
You
can
use
these
three
strings
like
virgin
or
resolved,
or
you
know
whatever,
like
that's
all
you're
gonna
get.
A
D
It's
it's
saying
that
I
want
to
that
that
I
am
maintaining
a
an
alternative
to
an
existing
package
that
might
be
used
in
various
versions
throughout
my
def
graph
and
I
want
to
transparently
swap
out
any
any
instance
of
that
package,
with
my
better
replacement
at
the
same
version.
So
you
could
like
the
the
example.
The
contrived
example
I
came
up
with
to
talk
about
in
slack,
but
Isaac
was
a
dependency,
is,
has
raw
source
and
also
it
trans
piles
to
CJ
s
and
I
want
to
ship
native
ESM.
So.
D
D
B
A
D
D
A
D
D
A
So
in
the
sorry
I
guess,
I
was
going
off
of
the
object
overrides
example
or
sort
of
like
carrying
on
that
work-in-progress
that
you
shared
Isaac
and
then
he
sort
of
nested
object
version.
Then
I
was
saying
like
essentially
then
all
depths.
That's
why
I'm
trying
to
utilize
the
left
hand
like
wild
card
for
is
right.
A
So,
on
the
left
hand
again
we're
utilizing.
The
star
is
like
special
wealth
carved
for
the
package
names
like
all
depths
and
then
the
right
hand.
Side
means
the
Sun
fur.
So.
A
A
A
D
Other
thing
I
mentioned
in
slack,
was
in
the
same
way
you
can
do
like
get
colon,
slash,
slash
or
SSH
colon,
slash,
slash
or
get
+
ssh
colon,
slash,
slash,
there's
the
NPM
colon
thing,
I
wonder
if
we
could
do
something
like
alias
+,
NPM,
:
and
a
package
name
and
then
I'm
saying
all
foods
become
map
to
this
alias
+,
whatever
it
is
and
which
I
can
specify
a
version.
But
if
I
don't
specify
one,
it
just
inherits
the
version
across
I.
Don't
know
just
throwing
that
out.
D
B
So
I
mean
I
feel
like
as
as
much
as
it
is
sort
of
inventing
a
new
programming
language,
the
just
having
a
string,
a
string
capture
on
certain
fields,
from
the
left-hand
side
or
from
the
node
being
overridden
from
an
implementation
side.
It
is
much
much
simpler
right.
It's
just
like,
let's
just
replace
this
particular
string
with
this
other
string.
B
B
If
we
start
adding
other
kinds
of
other
kinds
of
syntax
to
either
one
of
those
like
like
adding
new,
you
know
new
dependency,
specifier
types
right
like
somebody's
gonna,
want
to
be
able
to
do
NPM,
install
X
at
alias,
plus,
NPM,
:
or
like
put
that
in
their
dependencies
list
or
whatever.
So
this
way
we
can
say
everything.
That's
you
know.
The
right-hand
side
is
always
something
that
can
kind
of
be
like
a
key
value
in
your
dependencies
list.
A
B
B
Yeah
yeah
well
just
in
terms
of
like
the
the
the
algorithm
that
I
imagined
originally
for
doing
this
is
it's
not
actually
gonna
work,
it's
a
little
too
naive
because
what
it
doesn't
allow
you
to
do
is
sort
of
inherit
overrides
from
a
parent
object.
Once
you
go
on
to
a
new
object
override
it
like
forgot
about
all
of
its
parent
roles,
so
you
couldn't
have
something
like
you
know.
B
B
Because
it's
a
declarative
matching,
yes,
no,
it's
it's
actually
kind
of
similar
in
some
ways,
order,
dependent,
etc,
but
where
I
think
we
landed
was
and
what's
described
in
the
text,
but
not
in
the
implementation
section
is
that
rule
sets
are
inherited
effectively
and
only
overridden
by
the
the
child
rule
set.
So
if
you
have,
if
you
have
that
scenario,
then
then
Y
would
get
one
two
three
versions
of
both
X
and
Z
and
its
dependency
graph.
B
The
implication
there
is
that
the
duping
logic
has
to
be
a
little
bit
more
involved,
because,
if
I
have
I
can
I
can
end
up
in
a
case
in
the
dependency
graph,
where
I
have
two
different.
Two
different
modules
that
were
both
sort
of
subject
to
a
combination
of
inherited
rule
sets
and
end
up
with
the
same
effective
rule
set
of
overrides,
and
in
that
case,
all
of
their
dependencies
should
be
able
to
be
deduplicated
because
they're,
you
know
they
were
kind
of
subject
to
the
same
rules.
B
So
we
can
sort
of
read
deduplicate
those
up
to
the
top
level
as
well,
or
you
know,
to
the
highest
level
that
they
could
normally
go
if
there
were
no
overrides
and
so
just
tracking
like
which,
which
override
rules
were
relevant
in
which
portion
of
the
graph
kind
of
becomes
an
interesting
little
crossword
puzzle
interesting
little
Sudoku
to
do,
which
is
arborists
reason
for
existence.
So
that's
sort
of
the
natural
place
to
stick
a
bunch
of
this
logic
there
and
that
build
ideal
tree
step.
A
Is
there
any
like
so
in
terms
of
like
feedback
that
you're
looking
for
especially
I,
know,
there's
a
couple
West
even
brought
up
a
comment
that
you
want
to
add
in
terms
of
you
know
how
folks
might
want
to
be
using
overrides
are
sharing
overrides.
Is
there
any
specific
feedback
you're
looking
for
right
now
in
terms
of
this,
this
implementation
so
I
think.
B
Yeah
I
think
we
have
pretty
good
alignment,
unlike
what's
what's
gonna,
be
in
v1
and
in
in
these
conversations
where
we've
been
getting
to
I,
feel
like
we're
not
really
we're
getting
to
the
point
of
either
proposing
whole
new
stuff,
which
would
be
cool
and
gray,
but
really
kind
of
needs
its
own
RFC
in
its
own
examination
and
is
not
precluded
by
anything.
That's
in
this
or
where.
C
So
are
we
sure
that
the
RFC
doesn't
preclude
any
of
these
ideas?
So
if
we,
if
we
were
to
go
not
with
the
capture
group
but
with
the
just
pull
the
version
over,
that
would
be
a
breaking
change
right.
So
that
would
be
because
there
today
would
be
the
possibility
of
aliasing
without
a
version
which
would
be
like
latest
yeah.
B
C
A
C
So
so
here's
the
so
we
have
this
at
Netflix
in
our
Java
ecosystem.
So
this
is
why
I'm
bringing
something
so
obviously
there
are
solutions
which
do
work
for
this
it
in
other
ecosystems.
So
generally,
the
idea
is
a
central
team
manages
a
product
that
product
may
be
used
in.
You
know
a
bunch
of
different
use
cases,
but
when
they
look
at
the
dependency
graph,
they
want
to
apply
some
restrictions
across
the
entire
company.
C
Deduplication
x'
across
the
entire
organization,
on
what
versions
of
specific
transitive
dependencies
we
want
them
to
use
right.
We
do
this
via
this
thing.
In
our
nebula
visited
the
Java
side,
they
have
a
wrapper
around
Gradle,
which
tells
Gradle
hey
here's
some
constraints
that
we
want
you
to
resolve
for
in
the
JavaScript
ecosystem.
We
definitely
don't
have
anything
like
this.
We're
discussing
a
few
different
options
with
something
like
a
shared
dependency
that
everybody
installs,
which
uses
peer
dependencies
as
sort
of
a
hack
right.
C
So
you
would
install
you
know:
Netflix
recommended
you
know
peer
depp's,
and
then
it
would
have
some
some
strong
constraints
on.
We
say
we
only
want
react
above
the
you
know,
14
or
something
right,
and
then
anybody
who
had
that
there
install
would
say.
Oh
you're,
not
meeting
the
peer
dependency
of
react,
greater
than
14
you've
got
react,
12
and
that's
a
warning
not
an
override
right.
So
the
ability.
A
B
A
B
B
C
A
Would
this
be
similar?
I
know
it's
like
the
only
other
example
I
can
see
is
like
no
dirt
as
a
flag
for
referencing.
Let's
say
like
a
file.
We
would
have
something
like
override
path,
override
file,
yeah,
override
file,
override
paths.
So
then
you
could
be
distributing
like
specifically
overrides
JSON
file,
yeah.
C
It
would
be
really
nice
if
it
could
be
remote,
but
I
understand
why
that
adds
a
whole
of
other
complexity,
so
yeah,
if
we
just
had
one
file
at
that
point,
though,
it
still
seems
like
publishing
a
package
that
contains
overrides
and
letting
a
user
configure
their
project
to
say
my
override
should
come
from
this
package
would
be
a
better
way
than
just
having
it
be
some
arbitrary
file
in
the
file
system
right
because
it
might
be
different
package.
Ecosystems
have
different
bundles
of
overrides.
A
C
B
Yeah
and
the
other,
the
other
downside
that
I
was
going
to
say
like
so
we
we
can
put
objects
and
you
know
multiple
string,
keys
and
everything
else
in
the
ini
file
format.
It's
a
little
janky
and
less
humane
for
expressing
objects.
Then
then
JSON
is
obviously
but
like
can
be
done.
The
bigger
downside
there,
whether
we
use
NPM
RC
as
it
currently
is,
or
we
define
some
other
like
JSON
file
that
defines
these
objects.
B
The
bigger
downside
for
me
is
now
you're,
taking
the
config
away
from
the
thing
that
it's
configuring
and
we
get
into
kind
of
spooky
action
at
a
distance
where
it's
like.
It's
not
clear.
Why
I'm
getting
like?
Why
am
I
not
getting
the
latest
version
of
react?
I
keep
doing,
NPM,
install
reactant
latest
and
getting
version
14.
What
the
hell
like
that's
gonna,
be
really
much
more
confusing
I
think
so.
C
So
this
the
Java
cou
system
has
this
problem
and
the
way
that
they
solve
it
is
by
having
basically
an
explain
that
tells
the
user
what
things
are
being
applied
when
they're
installed.
So
if
they
have
an
override
it'll
say
like
hey,
you
know
it
we're
applying
this
override.
That's
that's
why
this
package
is
not
getting
exactly
what
you're
asking
for
yeah.
B
Yeah
I'm,
my
experience
is
that
people
tend
to
not
read
those
things.
C
B
Defining
it
based
on
a
package
seems
interesting
right.
We
could
kind
of
get
get
most
of
the
way
there
in
user
land
with
a
like
post
install
script
or
something
like
that
which
just
says.
Okay,
let
me
let
me
look
at
you,
know,
walk
up
the
tree,
find
the
package.json
and
just
go
ahead
and
mutate.
It
kind
of
verges
into
the
territory
of
bad
behavior,
for
a
dependency
to
do
that,
but
I
mean.
D
B
C
I
think
to
Jordan's
point.
We
could
also
have
that
install
package
instead
of
actually
applying
the
override
say
on
pre
install
so
like
the
installation
doesn't
do
the
change
it
says.
If
you
want
to
have
this
run
regularly,
add
this
pre
install
script
and
then
the
users
opt
it
into
a
you
know
more
and
then
pre
installed
would
just
be
like
npx.
B
Would
that
ever
we're
getting
into
like
implementation,
details
of
user
and
stuff?
So
you
know
probably
the
right
answer
is:
let's
try
it
all,
but
some
things
to
consider
there
is,
if
it's
a
pre-installed,
just
a
message
that
that
exits
with
status
code,
zero
that
won't
actually
be
shown
right
because
we're
we're
we're
hiding
those
for
all
kinds
of
good
reasons.
If
you
want
to
beg
for
money,
you
gotta
use
NPM
fund
and
provide
us
the
actual
URL
to
go.
B
B
C
Yeah
so
they're
the
the
really
problem
here
is
it's
a
chicken-and-egg
right.
You
need
to
apply.
You
need
to
have
the
overrides
before
you
can
apply
them
and
npm
installs
the
thing
that
a
that
you
know
would
get
them
if
they
were
in
a
package.
So
if
there
wasn't
some
sort
of
top-level
consideration
for
like
do
this
thing,
like
pre-install
timing,
right
that
that
can
go
and
fetch
the
oh
I.
C
That's
kind
of
why
they're
something
where
it
was
like
a
if
you
had
a
configuration
that
said
fetch
this
remote
URL
as
your
you
know
and
and
apply
that
override
as
if
it
was
appended
to
the
overrides
list
right.
That
would
be
one
way
to
saw
the
problem,
but
it
again
require
like
an
explicit
feature
in
the
CLI,
for
this
use
case
right.
B
This
I
mean
this
really
feels
like
a
B.
You
know
kind
of
kind
of
add-on
feature
once
we
have
all
of
the
like
implementation
details
of
just
just.
What
do
we
even
do
with
overrides
like
assume
we
get
them
from
somewhere
now,
I
have
them,
then?
What
right,
like?
That's
theirs,
there's
plenty
of
territory
there
to
still
kind
of
iron
out
being
able
to
fetch
that
that
data
set
from
somewhere
else
is
not
shouldn't,
be
terribly
complicated.
Right,
like.
C
I'm
happy
to
also
proof
of
concept
some
of
this
out,
because
I
mean
we
that's
effectively
what
we
did
in
our
in
our
Java
ecosystem
is
we
build
some
stuff,
then
discussed
with
the
Gradle
folks?
How
that
could
look?
You
know,
supported
by
the
tooling
directly
and
I'd,
be
happy
to
just
repeat
that
sort
of
process
yeah.
B
What
I,
what
I
wouldn't
want
to
do
is
have
somebody
you
know,
install
something
that
like
depends
on
something
that
tries
to
hack
with
overrides
and
now
you
know
their
whole
tree
gets
gets
mucked
up
because
they
installed
the
install
the
adapt
like
the
the
spooky
action
at
a
distance
is
just
it's.
We.
A
So,
just
quickly
time
check,
we
got
about
four
minutes
left
before
the
hour.
I
want
to
see.
If
there
was
you
know,
follow-ups
here
that
we
are
gonna
have
to
have,
or
if
we
can
pare
this
down
and
then
punt
some
or
add,
as
you
said,
Isaac
these
other
use
cases
the
like
these
other,
like
user
land,
features
that,
like
potentially,
we
can
discover
and
tooling
beyond.
Just
like
what
what
we're
planning
on
offering
with
this
first,
like
phase
of
overrides
they're
like
tangible,
is
this
just
like?
B
B
A
If
not
I'll
give
folks
two
minutes
back
and
appreciate
everybody
jumping
on
for
for
the
call
as
usual.
Well,
you
have
our
open,
RC
general
open
RFC
calls
next
week,
so
those
are
biweekly
alternating
with
these
deep
dives.
So
we
can
queue
up
another
topic
for
the
next
bi-weekly
deep
dive
in
that
call,
and
also
hopefully
it
let's,
let's
bring
up
any
any
topics
or
any
discussion
about
this
specific
RFC
in
the
PR
itself.
If
we
missed
anything,
yeah
I
appreciate
everybody
jumping
on.