►
From YouTube: Backlog bonanza, part 2
Description
No description was provided for this meeting.
If this is YOUR meeting, an easy way to fix this is to add a description to your video, wherever mtngs.io found it (probably YouTube).
B
To
that
dropbox
paper
link,
you
just
posted,
leads
to
a
404.
B
Okay,
so
you
did
yes,
if
I
delete
the
last
character
it
loads.
A
Okay,
so
we
were
going
through
the
rfcs
before
and
we
could
keep
doing
that
and
we
never
actually
took
action,
but
that's
okay.
I
think
I
want.
I
think
I
was
thinking
we
should
get
a
little
farther
before
we
start
posting
comments,
but
I
don't
want
to
wait.
A
A
A
A
B
A
A
A
A
B
A
Not
a
keyword,
although
I
think
that
it
may
be
contextual.
B
I
mean
there's
syntax
bike
shedding.
That
could
be
done,
but
I
think
the
important
question
is:
do
we
still
want
this
and
is
this
vaguely
the
right
approach?
I
think
the
answer
to
both
questions
is
yes.
A
And
of
them
is
pushing
down
to
subfield.
A
A
A
B
A
A
Stop
file
link
team
mcp
for
trying
to
merge
right
right.
B
A
Right,
so
I
think
that
the
design
is
probably
roughly
right
and
it
went
through
already
an
extensive
internals
process.
However,
I
have
some
concerns.
Like
it's
a
large
implementation
effort,
I
think
it's
a
stale
sort
of
a
stale
community
so
to
speak
like
right
with
a
lot
of
action,
but
now
it's
old
and.
A
A
Right,
in
fact,
we
could
revive
that
community,
but
I
guess
what
I'm
saying
is:
I
would
be
hesitant
to
merge
in
until
we
not
because
of
the
design,
but
because,
like
more
of
the
project
management
side
of
this,
like
is
this
a
project
we
want
to
take
on
now,
and
will
we
have
the
energy
to
see
it
through
and
so
forth
right
and
how
does
it?
What
is
it
in?
What
way
does
it
fit
our
priorities
of
closing,
finishing
up
old
work
and
seems
like.
B
B
A
A
A
B
Over
we
could
absolutely
like
poke
the
thread
and
say
we
talked
this
over.
We
think
this
might
be.
You
know,
despite
the
time,
lag
something
that
we're
interested
in
merging
here's,
some
minor
notes
and
are
people
still
around
and
interested
in,
seeing
this
merged
and
willing
to
do
implementation
work.
Let
me
let
me
talk
about
this
a
different
way.
I
feel.
A
A
Size
and
size
of
effort
that
is
in
flight
and
just
having
too
many
big
things
in
flight
like
major
additions
to
the
language
and
so
on,
seems
sub-optimal.
So
I
don't
know,
I'm
I'm
just
a
little.
That's
more
the
perspective
I'm
coming
out
of
like
you
can
poke
the
thread
and
definitely
say
we're
still
interested,
but
I
would
also
want
to
see
how
what
we
see
with
the
other
things.
So
maybe
we
should
move
to
those
other
things
and
come
back,
but.
A
B
B
I
mean
in
general,
I
think
most
rfcs
on
this
list
are
likely
to
be
opening
new
areas
of
work
rather
than
closing
previous
things
unless
they're
things
we're
already
working
on.
So
if
we
don't
feel
like,
we
have
the
bandwidth
to
do
new
things
right
now.
I
think
that
the
answer
to
almost
all
of
these
is
going
to
be
not
right
now
and
I'm
not
sure
that's
where
we
want
to
be.
B
I
guess
this
feels
like
a
relatively
small
addition
anyway,
fair
enough
moving
on
to
2442
for
the
moment,
so
this
one's
mine-
and
I
opened
this
because
I
felt
like
it
would
answer
a
lot
of
different
things
that
were
going
on
at
the
time.
It
also
interacted
with
a
weight
syntax
which
has
now
been
resolved,
so
that's
no
longer
applicable.
B
There
are
still
a
lot
of
use
cases
for
being
able
to
there's
still
a
lot
of
use
cases
for
post-fix
macros,
and
I
still
regularly
see
people
talking
about
things
on
internals,
for
which
this
would
be
a
good
solution.
B
Is
this
so
I'm
going
to
go,
look
it
up
and
apply
it
to
this
expression,
or
do
we
somehow
have
type
level
macro
dispatch
which
is
in
the
territory
of
people
have
done
phds
on
this,
and
I
feel
like,
in
my
opinion,
there's
enough
value
to
this,
to
make
it
worthwhile
to
do
the
lexical
dispatch
and
to
offer
that
as
an
option
and
that
doesn't
stop
us
from
doing
type
based
dispatch
in
the
future.
B
If
we
offered
a
new
kind
of
postfix
macro,
but
I
think
there's
enough
value
in
this
and
in
particular
enough
value
for
the
ability
to
experiment
with
interesting
new
syntax
like
let's
see
what
dot
match
would
look
like
or
what
dot,
if
would
look
like
or
dot,
four
would
look
like
and
be
able
to
experiment
with
that.
I
think
there's
value
in
that,
and
there
was
dispute
on
the
language
team.
B
B
B
B
A
A
B
Structure
right
so
yeah,
when
this
was
pitched,
the
concept
of
mcp
didn't
exist.
I
absolutely
agree.
This
should
not
just
be
fcp
merged
in
its
current
form.
I
think
that
it
would
make
sense
to
file
this
as
an
mcp
and
I
would
be
willing
to
file
this
as
an
mcp.
A
A
A
A
A
Not
just
as
they
come
up,
but
I'm
not
sure
in
any
case,
I
think
it
would
be
would
like
to
have
the
discussion,
but
I'm
not
sure
exactly
the
best
way
for
us
to
have
it,
and
I
don't
know
what
I
think
the
outcome
would
be,
because
I
could
definitely
see
a
work.
A
I
think
I've
previously
been
in
favor
of
type
based
dispatch
also,
but
I
think
I'm
now
in
the
not
in
a
camp
yet,
but
I
see
the
appeal
of
like
okay,
that
does
sound
cool,
but
we
have
to
wait.
We're
gonna
have
to
wait
for
years
to
ever.
You
know
he
may
never
really
do
that,
but
we
could
do
this
today
and
there's
a
lot
of
stuff
people
want
to
do
with
it
and
it
would
mostly
work
just
fine
and
I
think,
that's
probably
true,.
A
A
B
To
make
I'm
just
trying
to
make
sure
that
you
don't
already
feel
the
answer
is
no,
if
you
feel
the
answer
is
let's
talk
about
it,
then
I
am
happy
to
file
an
mcp.
A
A
B
I
think
it's
likely,
I
will
say
this.
I
absolutely
think
it
should
stay
nightly
only
with
a
feature
flag
for
quite
some
time,
not
just
because
of
how
it
interacts
with
other
bits
of
the
language
and
whether
the
feature
itself
was
stable,
but
because
a
critical
part
of
this
is,
let's
see
how
people
in
the
ecosystem
make
use
of
it,
and
if
what
we
see
looks
really
interesting,
that
may
be
compelling
and
if
what
we
see
is
really
horrifying,
we
might
want
to
say
you
know
what
nope.
B
Yeah,
if,
if
macros
2.0
is
still
baking,
then
I
think
it
makes
sense
to
just
say
we
should
have
a
mac,
a
bring
macros
2.0
over
the
finish
line.
Mcp
and
have
this
be
one
of
a
dozen
things
they
look
at,
but
maybe
somebody
could
file
an
mcp
for
let's
get
macros
2.0
over
the
finish
line.
B
A
So
teach
cat
to
join
bite
join
slices.
This
feels
more
lip
length
lips
than
laying,
but
I
guess
this
is
like.
B
There
was
a
bunch
of
discussion
on
this.
I
remember
because
I
was
in
that
discussion
and
I
think
one
of
the
big
points
of
dispute
on
this
one
is
should
can
cat
do
this,
or
should
we
have
a
can
cat
bites?
That
does
this
so
that
it's
type
safe.
I
very
much
argued
we
shouldn't
have.
Can
cat
like
do
different
things
based
on
its
arguments.
We
should
have
two
different
types,
but
otherwise
I
think
it's
perfectly
fine
to
have
a
thing
that
does.
B
A
B
A
A
B
B
B
Yeah
reading
the
section
on
better
naming
of
unnameable
types,
it
sounds
like
this
is
effectively
type
alias
infiltrate
without
having
to
name
all
the
traits
and
instead
just
having
the
compiler
infer
them.
That
seems
like
global
type
inference,
and
my
first
instinct
is
to
say
we
really
don't
want
this.
C
A
A
A
You
know
my
mind
but
such
that
it
gets
inferred
like
because
this
method
returns
a
self
item
you're
allowed
to
you
don't
have
to
like
do
repeat
yourself.
The
main
reason
I
think
I
would
like
that
is
that
it
ties
into
well
just
because
it
seems
like
repeating
yourself,
so
it's
related
to
this
rfc,
but
different,
because
it
could
be
a
much
more
limited
special
purpose
feature
if
we
ever
label
it
anyway.
I
think
this
stupid,
maybe
a
good
idea,
but
we
should
lay
the
groundwork
first.
A
All
right
type,
changing
struct,
update
syntax,
so
I
think
what's
happening
here
is
that
today,
when
you
write
blah
blah
blah
some
fields,
not
that
bar
the
type
of
this
expression
is
the
same
as
bar
today.
A
Then
you
would
rather,
that
the
type
of
this
be
independent
from
the
type
of
bar,
so
that
you
could
change
the
value
of
a
that
makes
sense.
A
D
B
I
think
that's
a
reasonable
question,
because
this
use
case
sounds
completely
justified
to
me
and
the
idea
of
change
the
thing
that
is
dependent
on
the
generic
type
and
keep
all
the
other
fields.
That
aren't
seems
completely
reasonable,
but
changing
from
one
entirely
unrelated
struct
type
to
another,
because
the
fields
happen
to
match.
B
A
A
A
A
I
guess
I
think
I'll
have
to
look
at
what
the
inference
does.
My
expectation
would
be
that
you
make
fresh
inference
variables
for
all
the
generics
you
assign
the
type
of
the
fields
that
were
given,
and
then
you
assign
the
types
of
the
fields
that
came
from
the
construct
that
you
have
the
dot
dot
with
like
the
ones
that
are
not
listed
explicitly
and
that
should
fully
constrain
like
that.
Should
just
be
it.
A
I
don't
know,
it
seems
like
there's
a
lot
of
text
here
and
I'm
not
quite
sure
why
maybe
you're
overlooking
something-
or
this
has
a
more
complex
idea.
A
A
So
I
would
say
probably
this
should
be
merged,
but
it
needs
a
review
discussion
of
the
inference
details.
Only
it's
in
the
discussion,
but.
A
B
B
B
The
latter
the
the
review
of
type
inference
details.
You
said
that
you
would
want
to
review
the
type
inference
details
before
we
considered
this
right.
A
I
think
I
want
to
do
that.
I
don't
think
that'll
take
too
long
like
before
I
before
I
clicked
my
name.
I
would
want
to
do
that.
A
B
C
A
A
A
A
A
B
Yes,
this
there
have
been
several
variations
of
this
one.
I
feel
like
I've
filed
one
of
them,
and
this
is
another
one
and
they
pretty
much
boil
down
to
you
shouldn't
have
to
count
the
number
of
things
in
your
initializer
and
put
that
number
into
the
size
of
your
array.
You
should
be
able
to
write,
underscore
and
have
the
compiler
say
I'll.
Look
at
how
long
your
string
is,
or
I
will
look
at
how
long
your
slice
is.
B
B
The
other
larger
point
of
contention
was,
should
we
have
this
as
a
special
case
for
array
sizes,
or
should
we
allow
more
general
type
inference
more
general
generics?
That
kind
of
thing
in
this
context.
A
I
feel,
like
that's,
pretty
relevant,
like
it's
not
trying
to
solve
the
same
problem,
but
it's
definitely
stabilizing
some
of
the
groundwork
for
solving
this.
The
quote-unquote
principled
way.
A
Yeah
not
today
and.
A
I
could
imagine
trying
to
enable
that,
and
that
is
definitely
const
generics
and
then
why
that
doesn't
it
doesn't
even
parse.
Today,
it's
a
little
surprising
to
me,
but
but
then.
A
C
A
A
Well,
it
doesn't
seem
to
me
it
does
run
into
these
problems
we've
hit
before
of
like
well.
What
about
stuff
like.
B
Right,
how
does
this
interact
with
numeric
inference.
A
Yeah,
so
I
guess
I
could
definitely
see
an
argument
for
we
don't
want
to
support
underscore,
except
in
the
case
of
constant
expressions.
A
Nonetheless,
I
think
that's
how
I
would
expect
to
implement
a
more
general
version
in
the
future.
Yes,
but
I
think
like
it's
not
only
that,
but
it's
like
that's
how
actually,
how
it's
hard
for
me
to
imagine
how
we
would
implement
this
in
the
compiler
without
leveraging
const
generics.
So
I
think
what
we
would
probably
do
is.
We
would
create
a
constant
inference
variable
for
that
underscore
and
we
would
handle
the
type
you
know
and
and
then
read
out
what
value
we
got.
B
So
we
should
talk
to
folks
like
ecstatic,
morse
and
see
about
how
this
interacts
and
see
what
they
think.
A
A
A
A
A
I
really
wanted
this
thing,
but
I'm
kind
of
as
you
can
see
here,
I
remember
deciding
it's
just
not
worth
it.
We
should
close
it
for
now.
There's
too
many
weird
rules
like
this
concern
that
aaron
raised.
A
A
There's
some
analogous
stuff
that
happens
like
when
you
write
self
dot
dot.
You
get
it's
different,
but
when
you
write
self
this
as
a
struct
expression,
you
get
the
same
type
as
the
impul.
But
if
you
write
you
know
what
I
mean
you
write
like.
If
you
have.
Let
me
give
you
an
example.
A
A
This
has
some
inferred
there's
some
subtle
distinctions
like
that,
and
this
is
a
similar
one
where
self
colon
colon
item.
A
A
A
A
C
A
A
One
thing
I
was
thinking
side
note:
we
have
this.
I
would
really
like
it
if
we
take
the
output
from
these
meetings.
A
And
maybe
it's
just
going
to
be
comments
in
the
rfc,
but
I
think
it
would
be
nicer
if
we
had
like
maybe
on
the
lane
team
website.
I've
always
dreamed
that
for
these
ideas
we
can
kind
of
write
up,
write
up
these
concerns
and
thoughts
and
then
link
to
the
rfc
like
in
shorthand
form,
but
so
that
you
can
easily
remember
wait.
What
was
the
major
concern
of
that?
Oh
right.
B
I
mean
we
could
do
that
in
the
rfc
threat,
at
least
for
the
time
being,
all
right,
we
should
start.
A
A
A
A
D
D
So
I
don't
know
what
happened
with
it.
Maybe
it
was
an
mcp
or
something,
but
it
seemed
sort
of
an
updated
version.
B
A
A
A
A
A
A
It's
interesting
I'm
reading
simone's
comments
here.
The
2019
road
map
includes
finnish
long-standing
dangers.
I
argue
resolving
a
1.0
tracking
era
issue
fits
into
that
discussion.
That
actually
sounds
like
a
reasonably
good
case,
but
basically
he
was
kind
of
saying
look.
I
did
the
work
of
trying
to
make
the
intersection.
A
You
know
the
forward
compatible
intersection
of
so
we
don't
have
to
solve
the
hard
problems,
but
we
can
solve
the
easy
ones
right
and
then
we
should
do
that,
and
I
agree
with
that
concept,
but
I
think
the
right
thing
I
have
to
reread
it
now.
I
remember
thinking
when
I
read
it
last
time
that
it
didn't
really.
B
So
this
does
not
pin
down
some
portion
of
the
abi
in
a
way
that
will
constrain
us
in
the
future.
A
Maybe
the
right
answer
is,
I
don't
know
the
answer
to
your
question.
I
don't
think
so,
but
I
have
to
reread
it.
I
don't
have
it
fresh
enough
in
my
mind,
it
might
be
that
it
committed
us
to
something
that
I
didn't
think
was
a
big
deal.
A
A
List
it's
two
o'clock.
I
could
go
a
little
more
we're
making
pretty
good
progress.
B
A
A
These
are
like
tuples
of
name
fields:
okay,
I'm
of
the
opinion
that
we
should
not
do
this.
I'm
gonna
put
it
right
out
there,
not
because
I
think
it's
a
bad
idea,
but
just
because
it's.
A
B
You
know
I
wouldn't
support
fcp
merge
either.
I
do
long
term
feel
like
this
is
a
thing
that
we
should
have,
but
I
also
think
that
it,
this
does
not
like
fill
in
a
corner
or
make
a
language
make
the
language
more
orthogonal.
A
Unlike
union,
I'm
not
sure
if
I
would
say
it
touches
everything.
I
mean
what
like
the
major
difference
between
the
union
rfc
and
this
one
is
that
union
added
a
new
capability
right.
A
Yeah
it
basically
either
you
can
either
imagine
destroying
it
to
a
fresh,
struct
sort
of
to
a
fresh
struck
declaration
or
just
to
tuples
without
names.
Just
reordered
right,
there's
some
slight
difference
in
semantics.
I
imagine,
but
not
much
so.
A
A
D
A
They
were,
we
have
a
challenge
to
figure
out
what
is
the
canonical
representation,
but
maybe
they
get
sorted,
but
that's
you
know
so
anyway.
My
point
is,
it
doesn't
add
a
fundamental
capability,
so
union
is
requiring,
like
unsafe
guidelines,
interaction,
all
kinds
of
interaction
which
is
fine
because
we
needed
that.
But
this
doesn't
this
one
isn't
as
bad
as
that
in
terms
of
touching
everything,
but
also-
and
by
the
same
token,
it's
not
as
important
as
that.
A
B
And
okay,
so
this
is
a
different
subset
of
the
problem.
This
one
is
more
interesting.
This
one
actually
seems
like
a
great
idea
to
me.
It
kind
of
got
dropped
on
the
floor
at
some
point,
but
I
personally
am
hugely
in
favor
of
having
this
available.
A
A
A
There's
a
claim
that
it's
going
to
work
out
just
fine,
I'm
not
sure.
If
I
believe
that
claim,
I
guess
I
think
I
would
recommend
I
would
say,
recommend
for
project
proposal.
A
This
seems
like
a
great
starting
point.
It
may
be
that
we
want
to
adopt
it
wholesale,
but
I
would.
C
A
I
was
just
gonna
say:
I'm
not
sure
whether
one
thing
I
would
want
to
consider
in
in
that
is
conditions,
and
whether
like
this
is
something
we
might
face
in
across
an
addition
boundary.
If
there
are
some
ways
that
we
can
make
inference,
behave
better
that
wouldn't
be
strictly
backwards
compatible,
I'm
not
sure.
B
I
think
we
should
ask
for
a
project
proposal,
because
we
want
to
ask
somebody
to
commit
to
shepherding
it
all
the
way
through
the
process
like
be
there
as
the
champion
of
the
rfc
to
talk
to
the
implementer
or
potentially
be
the
implementer.
Considering
that
varcour
was
the
one
to
propose
it
and
they
certainly
know
the
compiler.
B
So
I
think
we
should
ask
for
a
project
proposal,
but
not
for
changes.
Just
for,
please
be
the
project
group.
We
should
charter
one.
So
that's
really
interesting.
A
And
I
agree
with
you,
and
I
think
you
gave
words
to
an
intuition
that
I
had
earlier
around
delegation
and
some
of
these
other
projects
and
some
of
these
other
things
too,
like
underscoring
type
aliases,
maybe
or
even
postfix
macros,
which
is
kind
of
like
for
a
change
that
is
sufficiently
visible
to
language
users.
A
I
think
that
I
wanted
to
have.
I
wanted
to
go
through
the
project
process
and
I
want
to
have
us
like
talk
about
it,
freshly
and
get.
You
know
what
I
mean.
I
don't
want
to
just
like.
Take
some
stale
rfc
and
approve
it
and
then
make
a
tracking
issue
and
call
it
done,
and
I
would
rather
that
we
try
to
do
it
in
a
more
healthy
way
and
have
somebody
who's
going
to
see
it
through
and
so
forth,
and
so
it
might
be
that
that
also
involves
changes
to
the
design.
B
But
quick
side
question
here,
so
we're
going
through
a
large
number
of
rfcs
I've
taken
a
little
time
along
the
way
on
two
or
three
of
them
to
go
ahead
and
write
a
comment
or
to
do
rfc
bot.
But
a
lot
of
these
were
just
making
notes
here.
B
A
B
I
just
wanted
to
mention
before
we
moved
on
there's
a
quick
note
about
the
whole
like
enums
are
regularly
talked
about
the
other
facet
of
this
that
keeps
getting
talked
about,
and
I
I
feel
like
there
is
definitely
a
we
should
talk
about
this
as
a
problem
before
seeing
solutions.
Kind
of
thing
going
on,
there's
been
like
three
variants
of
a
discussion
on
internals
about
anonymous
enums,
which
basically
amounts
to.
B
If
we
want
to
do
this
or
if
there's
another
solution
for
the
same
problem,
I
just
bring
it
up
because
there's
a
lot
of
internals
discussion
and
I
feel
like
there
hasn't
been
any
serious
lane
team
discussion
of
it.
It's
true
and
I
don't
want
to
end
up
in
another.
One
of
those
situations
of
somebody
does
a
ton
of
work
before
there
is
any
blank
team
feedback.
D
It
does
seem
like
that
sort
of
area
where
there's
sort
of
a
really
wide
solution
space
and
we're
not
even
sure
we
want
to
explore
the
we
might
want
to
explore
the
solution
space.
But
we
don't
know
if
we
want
to
identify
solutions
and
implement
them,
is
sort
of
the
exact
intent
behind
project
groups.
Exactly.
D
We
would
like
someone
to
open
a
project
proposal
with
the
intent
of
following
through
by
forming
a
project
group
and
sort
of
getting
resolution
space
explored
and
such.
B
Minor
phrasing
knit
I
think
we
actually
are
interested
in
enum
variant
types
with
the
specific
proposed
solution
here,
but
the
anonymous
enums
are
the
ones
where
it's
like.
Yes,
we'd
like
to
talk
about
this
at
a
problem
in
the
problem
space
rather
than
the
solution
space.
A
A
So,
a
long
time
back,
we
were
like
one
of
the
first
rc's
was
to
out
of
something
we
called
virtual
struct
at
the
time
which
was
not
popular,
and
then
it
was
kind
of
an
attempt
to
get
make
it
make
you
able
to
model
like
the
dom
essentially
more
efficiently
in
rust,
which
is
still
a
problem,
still
an
unsolved
problem,
and
so
there
was
because
it
was
so
popular.
We
pulled
it
back,
but
then
we
sort
of
solicited
rfcs
and
there
were
four
or
five
competing
rfcs,
aiming
at
the
same
design
space.
A
A
A
B
I
think
we
need
one
coherent
project
proposal
here
and
we
can
let
the
project
group
evaluate
things
like
you
know.
What
are
the
trade-offs
between
having
a
new
opt-in,
built-in
trade
versus
having
a
separate
trait
versus
all
sorts
of
other
ways
this
could
be
handled.
I
know
we
had
a
half
dozen
different
discussions
on
this,
so
I
would
say.
A
A
A
Maybe
we
want
to
have
a
list
of
like
open
problems
like
that's
more,
this
lengthy
page
I
keep
talking
about.
I
think
part
of
the
reason
I
would
like
it
is
that,
when
we're
thinking
about
what
to
do
next,
we
should
consult
it
and
be
like
here
were
some
ideas
and
some
of
the
summaries
which
of
these
do,
we
think
are
any
of
these
like
something
that
seems
like.
The
time
is
right.
A
B
I
remember
this
one
and,
as
I
recall
this
nico,
could
I
ask
you
to
turn
off
the
recording
for
a
minute.
Please
sure,
maybe.
A
B
B
Like
I
see
people
on
chat,
for
example,
writing
this
as
a
quick
note
and
then
realizing.
Oh
right,
that
doesn't
actually
work
and
I
think
because
people
are
used
to
things
like
config
things
where
it
does
tend
to
work
or
derive.
Actually
is
a
really
good
case
where
you
can
write,
derive
trait
1,
trait
2,
trait
3,
trait
4,
and
this
is
the
kind
of
thing
where
the
same
kind
of
thing
at
the
top
level
feels
like
it
ought
to
work.
But
it
doesn't.
A
It
doesn't
feel
like
we
need
a
project
proposal
around
this.
If
we
were
going
to
do
it,
it
would
either
be
merged
or.
B
It
doesn't
so
it
doesn't
exactly
need
design
work.
The
one
thing
that
has
to
happen
before
this
could
be
merged
is
somebody
needs
to
very
carefully
review
the
token
stream
specification
for
how
does
this
interact
with
like
attribute
proc
macro
other
attributes?
What
does
it
expand
to
and
what
tokens
does
each
one
get.
A
Yeah,
honestly,
I
think
what
that
alone
makes
me
inclined
to
say:
I
think
we
should
defer
it.
There
are
some
issues
and
it
just
doesn't
seem
worth
it.
You
have
enough
problems
and
revisit
it
later,
like
it's
some,
it's
basically
a
bandwidth.
B
I
would
be,
I
wouldn't
object
strongly
to
the
notion
of
we
like
the
concept
in
principle.
The
fact
that
it
needs
further
specification
makes
us
inclined
to
differ.
A
Yeah,
I
think
that's
like
I
feel
like
if
we
open
the
project
proposal,
that's
sort
of
what
I
would
argue
for
is
I
don't
expect
procedural
macros
in
particular
are
an
area
where
I
feel
like
we're
under
we
have.
We
are
right
under
strong
strength
and
it
we
need
to
move
carefully,
and
this
doesn't
seem
worth
it.
B
A
Okay,
2600
so
now
we're
up
to
2602
attributes.
B
B
All
over
the
place,
I
remember
this
one.
B
I
think
this
part
of
the
proposal
is
theoretically
syntactic
in
that,
like
we're
not
specifying.
Yet
what
the
semantics
of
what
these
things
are,
we're
saying
that
you
can
parse
them
and
then
that
would
allow
a
proc
macro
to
deal
with
them
and
then
separately
we
could
spec.
Some
of
those
out
like
this
opens
the
door
to
being
able
to
say
that
a
function
pointer
has
a
wrapper,
or
some
you
know
attribute
in
a
some
parameter
in
a
closure,
has
a
wrapper
or
something
along
those
lines.
B
But
for
the
moment
I
believe
this
is
purely
for
the
benefit
of
proc
macros
allow
parsing
attributes
in
more
places.
A
A
A
B
B
I
tend
to
generally
be
in
favor
of
let's
parse
a
thing
and
allow
proc
macros
to
have
fun,
but
I
also
completely
understand
that
it
will
take
a
lot
of
work.
B
The
thing
that
makes
me
mildly
inclined
is:
there's
all
sorts
of
binding
generators
and
ffi
and
c
plus
and
similar,
where
it
would
be
really
helpful
to
have
attributes
nested
in
the
middle
of
various
kinds
of
types
to
say
here
is
something
important
that
you
need
to
know
as
the
binding
generator
that
isn't
part
of
rust,
syntax.
A
So
that's
true,
I
guess
I
would
say
so.
Binding
generators,
maybe
like
I
could
imagine
sort
of
tooling
for
extended
proof,
tooling
or
so
on.
Right.
A
B
B
I
think
my
inclination
would
be
similar
to
the
previous
one,
which
is
concept,
seems
okay.
There
are
some
practical
considerations
in
this
case:
ambiguity,
issues
in
the
grammar
and
because
of
that,
we
would
want
to
see
this
come
through
as
a
project
proposal
with
somebody
who
is
prepared
to
put
a
great
deal
of
grammar
work
in,
but
not
opposed
kind
of
mildly
in
favor.
But
we
want
to
postpone
anyway,
because
complications.
B
A
Postponed
because
complications-
and
enough
is
because
interaction
with
grammar
and
procedural
macros,
two
areas
were
not
especially
well
specified.
A
B
Got
through
almost
another
whole
page,
yeah
page
two,
if
I
might
suggest
one
other
that
I
think
we
could
disposition
very
quickly,
because
I
think
that
there's
an
established
project
group
that
should
be
called
into
play.
B
2633
ff5
returns
twice
for
specking
functions
like
long
jump
and
sorry
set
jump.
We
now
have
an
ffi
unwind
working
group
that
is
currently
working
on
the
concept
of
plain
old
frames,
and
when
can
you
use
set
jump?
I
think
it's
really
clear
that
the
correct
disposition
for
ffi
returns
twice
rfc
is:
please
work
this
through
the
ffi
unwind
working
group,
which
is
actively
discussing,
set,
jump,
long,
jump
and
unwinding
right
now,
yeah.