►
From YouTube: Lang Team Triage Meeting 2019.10.03
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).
A
Let's
talk
of
a
Nico's
Facebook
account,
so
let's
see
welcome
to
our
current
meeting
we're
doing
a
new
thing
by
the
way,
because
papers
so
grabby
with
email
addresses
and
annoying
I'm
moving
the
minutes
to
get
afterwards.
So
we'll
just
keep
the
same
duck
every
week.
Instead
of
making
new
duck
since
I
used
to
do.
C
A
B
A
Or
did
they
get
that
wrong
like
from
okay?
So
anyway,
okay
I
did
talk
some
to
Ralph,
but
he's
currently
on
vacation
I
was
talking
to
him
about
the
unsafe
code
guidelines
and
CTF
II
in
general,
and
I'm
gonna
defer
that
util
discussion
for
later,
but
we
basically
agreed
that
what's
needed
is
to
create.
A
He
basically
said
a
lot
of
good
conversation
happening,
but
what's
not
always
happening
is
that
that
conversation
doesn't
always
get
taken
from
an
issue
into
like
some
kind
of
summary
and
brought
before
a
wider
audience,
and
so
we
were
talking
about
a
big
improvements
there
about
any
case
to
bring
him
in
for
a
meeting
he's
currently
touring
around.
So
it'll
have
to
be
you
next
week
or
so
so,
I'm
gonna
try
to
follow
up
on
that
I've
been
making
this
list
of
shepherded
items,
projects
that
we're
trying
to
carry
through
to
conclusion.
A
B
B
First,
it
was
not
doing
it
inside
because
that's
sort
of
like
how
one
expects
not
exactly
to
to
pay
respect
to
inside
and
outside,
and
so
there's
a
stabilization
here
and
that
David
made
and
who
and
and
and
then
and
then
there
is
like
follow-up
work
on
maybe
doing
the
the
addition
that
David
told
me
wanted
to
use
in
its
in
which
you
can
read
more
about
in
the
truck
dealership.
The
limp,
sorry
on
bit,
tired
snakes
or
not.
That's
good.
Basically,
we're.
A
Done,
which
is
awesome,
David
I,
think
David
said
he
was
implementing
the
limp
too.
So
that's
cool!
Oh
yes,
oh
yeah,
you
mentioned
something
about
it.
On
somewhere,
okay,
going
up
from
there,
I
haven't
done
anything
on
the
rebalance
coherence,
but
all
that's
really
waiting
for
is
one
more
pass
over
the
test.
Since
we
found
a
hole,
I
wanted
to
like
try
to
put
them
into
some
something
so
I
could
see
if
there's
any
more
obvious
gaps
central
said
they
might
also
do
it.
A
That
would
be
fine,
but
and
in
terms
of
this
coherence
I'm
in
like
broaden
this
a
little
bit
actually
I
think.
E
A
Thing
that's
going
on
here
is
like
didn't
rate
stuff,
I
need
a
better
name,
but
in
particular
there's
the
coherence,
maybe
bypassed
by
indirect
bull
stuff
that
we
talked
about
and
the
objects
say
for
dispatch
or
I've
seen
at
those
for
a
while
back
and
they're
kind
of
intertwined,
which
is
why
I
approved
it
we're
making
progress.
We
were
hitting
some
cycles.
I
figured
out
what
the
problem
is
now
we're
looking
to
work
around
it
and
I'll
post
it
more
I,
don't
have
a
know
exactly
what
we're
going
to
do
yet,
but
yourself--all
does.
A
C
A
A
Certain
we're
asking
something
was
object,
safe
and
while
we're
assembling
the
list
of
candidates-
and
then
the
problem
would
be
that,
like
thing
impulse,
we
might
use
which
is
a
reasonable
place
to
do
it,
but
then
sometimes
some
of
the
chucks
to
see.
If
something
is
object,
safe
have
to
do
normalizations
and
they
were
hitting.
A
And
yes,
we
can
I
have
to
figure
out
the
best
way
to
sidestep
this
I
think
one
way
to
do
it
might
be
to
assume
there
is
an
imple
there,
but
have
that
if
you
wind
up
using
it
you'll
get
an
error
down
the
line.
I
think
that
would
work,
for
example,
but
but
I
have
to
look
the
other
way
to
do.
It
might
be
just
to
split
so
that
we
have
a
more
narrow
check
that
determines
whether
there's
a
nimble
or
not,
and
then
a
that's
actually
used
in
this
code
anyway.
A
Whether
the
trade
implements
itself
from
both
of
these
cases
and
that's
the
rough
casting
is
more
about
my
converting
between
types.
So
this
is
like
within
one
trade
type,
and
that
is
across.
A
A
C
The
the
goal
is
to
explain
kind
of
in
a
public
place
which
has
already
sort
of
been
done
in
Nico's
last
comments
in
that
RFC.
What
the
plan
is
to
go
forward
with
making
progress
on
this
future
and
to
make
sure
that
there's
a
clear
intent
to
go
forward
that
the
interested
parties
can
be
aware
of,
and
so
we
broke
this
into
bullet
points
about
that
I
guess
I'll
just
go
through.
So
the
first
thing
is
to
actually
create
the
project
group
I'm
sure.
C
A
C
A
C
Also
a
little
nebulous,
but
yes,
it
is
to
make
sure
that
interested
parties
have
a
place
to
collaborate
and
a
means
to
collaborate.
That's
not
just
comments
in
that
in
one
long
RFC
and
towards
that
end,
the
first
sort
of
effort
is
defining
the
scope
which
this
summary
does
with
some
hazy
edges
in
some
places.
C
So
the
first
sort
of
simple
at
there
about
sample
use
cases
is
really
the
core
purpose.
I
guess
the
the
core
thing
that
we
know
we
need
to
do
as
a
working
group
and
that's
to
you
know
clearly
to
limit
here's
the
things
that
rust
provides
that
we
claim
should
work
that
we
support
and
that
can
have
some
cross
platform
issues,
but
in
general
we
think
that
the
working
group
is
definitely
going
to
be
able
to
tackle
that
limited
subset
of
the
overall
problem.
C
Then
the
next
bullet
point
the
next
major
bullet
point,
this
RFC,
that's
just
where
we
want
to
go
with
27.53,
specifically
or
with
Turing
into
the
discussion,
we're
talking
about
just
closing
that
and
replacing
it
with
a
new
one.
But
the
idea
is
get
a
short
I
guess.
This
was
already
the
goal
of
275
three,
but
have
a
a
small
self-contained
RFC.
That
is
not
intended
to
be
anything
more
than
the
initial
effort
from
the
working
group.
A
A
A
A
A
A
Defines
what
the
rust
panic
mechanism
how
its
implemented,
but
mainly
what
that
means
in
practice,
is
if
we
diverged
from
the
native
mechanism,
we
would
have
to
do
some
form
of
adaptation
and
that's
like
other
shaming,
for
example.
Yes,
that's
something
we
could
design
at
the
time.
We
should
be
able
to
do
it,
though,.
B
A
Join
him
all
right
hold
on
any
more
thoughts
on
the
unwind
FFI.
Everybody
comfortable.
Is
this
one
thing
I'm
sorry
then
go,
but
one
thing
I
want
to
say
is
I.
Think
our
plan
is
to
close
the
old
RFC's
and
replace
it
with
this
by
the
way
third
time's
a
charm,
but
expect
to
move
quickly
to
an
SUV
third.
C
D
D
Just
wanted
to
jump
in
trying
to
read
over
that
the
text
here
in
the
that
the
project
in
fine
wine
summary
that
Niko
has
here
I'm
a
I
personally
am
a
little
confused
by
the
part
that
seems
like
it's
that
there's
a
bullet
that
says
native
panic,
slash
lawn
jump,
which
is
soon
a
type
of
for
long
jump,
traversing
and
no
destructor
Russ
frames.
To
me.
That
sounds
like
it's
saying
that
long
jump
is
a
native
panic
mechanism
which
I
guess
is
sort
of
true,
but
it's
yeah.
A
D
Like
there
isn't,
it
uses
okay,
right,
right,
Hugh
searches
and
handling.
So
therefore,
that's
worked
he's
long
jump
there.
Okay,
fine
I,
guess
my
needy
reaction
was
to
say
long
jump
would
not
be
a
super
way
to
implement
this.
The
unwind
extern
thing
but
I
guess
when
your
platform
does
use
structured
exception,
handling,
has
long
jump
than
it
works.
A
Alright
and
then
but
the
other
point,
this
may
or
may
not
be
like
fully
in
scope,
I
think
we
were
debating
about
it.
One
of
the
things
is
that
there
is
there
it's
useful
to
this.
We
were
distinguishing
this
no
destructor
cases,
because
they're
kind
of
relatively
less
complex
than
the
case,
where
you
have
to
decide
like
which
destructors
might
run
and
so
on
and.
C
A
You
have
a
notion
of
a
rust
frame
that
doesn't
have
destructors,
that's
also
quite
useful
for
a
long
jump
which,
and
that
comes
up-
didn't
see
api.
So
it
seemed
like
it's
not
exactly
what
we
were
setting
out
to
do,
but
it's
like
so
close
that
it
might
be
worth
saying.
If
we
do
have
this
notion,
then
we
can
say
that
yes,
it's
okay
to
long
jump
over
frames
as
long
as
they
meet
this
definition,
and
maybe
there
would
even
be
like
a
lint
that
could
help
you.
A
B
C
B
B
F
A
A
A
It
allows
you
normally
you
specify
on
the
rusty
command-line,
extern
foo
equals
path,
and
that
makes
the
crate
foo
available
from
this
given
path.
This
allows
you
to
just
say
foo,
which
means
that
the
crate
is
in
the
prelude
exactly
as
any
other
create
so
for
the
universal
paths
and
so
on,
but
the
path
is
determined
by
searching
the
whatever
is
called
the
search
path
or
something
I
forgot.
A
What
we
called
it
by
searching
around
different
directories
same
way
that
it
would
be
if
you
have
extra
crate
foo
this
group,
and
the
main
advantage
of
this
is
that
you
no
longer
have
to
write
extern
crate,
proc
macro
or
rather,
cargo
can
write.
Dice
extra
talk,
nut
grille
and
you
can
use
this
park
macro
great,
like
any
other
quick.
A
B
G
So
this,
like
Paco
and
issue
like
they're,
really
excellent,
like
people
have
been
like
picking
out
in
different
ways
and
I
feel
like
I.
Don't
know
that
it
would
necessarily
need
to
block
this
cuz
I
think
was
like
maybe
just
to
looking
ahead,
regardless
of
how
we
solve
that
problem,
but
like
weirdly,
somebody's.
C
G
A
A
G
B
B
B
G
Like
I,
don't
I
don't
know
if
it
like
should
block
doing
this
just
like
allows
like
maybe
like,
for
example,
maybe
a
stable
curve,
a
feature
that
you
can
experiment
with.
That's
a
solution,
and
that
would
be
great
but
like
yeah
I,
just
I
feel
like
the
larger
picture
here
is
like
you
need
someone
to
share
so.
A
A
A
B
A
G
B
E
A
It
also
seems
to
me
that
these
don't
have
a
lot
like
the
proc
macro
pace
in
the
Alek
case,
despite
being
superficially,
similar
are
actually
kind
of
different
and
the
work
that
needs
to
be
done
around
no
stood
and
Alec
wouldn't
really
affect
the
yeah,
so
I
do
think
we
could
go
forward
with
this
mechanism.
Let's
say
you
know,
but
we
don't
think
like
this
may
be
a
useful
feature
to
solve
the
Alec
problem,
but
that
is
a
bigger
problem
that
would
sort
of
want
someone
to
yeah.
G
A
G
A
G
A
A
B
A
The
future,
as
I
said
it
only
it's,
basically
a
question
of
who
provides
the
path,
but,
from
the
language
point
of
view,
it's
not
very
different,
like
like
from
the
language
going
to
do,
there's
some
set
of
crates
that
have
a
path
that
were
given
on
the
command
line.
G
A
A
This
I'm
gonna
skip
over
these
and
talk
about
include
stir.
So
in
the
past,
we've
talked
about
this
change
to
normalize
new
lines
include,
stir
clearly
a
backwards.
Incompatible
change
are
arguably
a
bug-fix
because
we
do
it,
but
we
we
said:
okay,
fine,
you
can
go
ahead
and
do
this
if
nobody
complains
basically
and
the
bar
side,
including
at
least
chris
smoove-
did
some
scraping.
A
A
B
A
E
Crlf
sand
HTTP
messages
that
people
were
building
and
stuff
like
that
were
plausible
examples.
Maybe
my
thought
on
this
one
now
I
think
with
the
pushback
is.
This
was
fine
if
we
could
sort
of
slip
it
in
and
nobody
really
noticed
or
cared,
but
now
that
people
care
enough
I
are
starting
to
show
up
and
say
they
care
I.
Think
it's
hard
to
argue
that
this
is
a
bug
fix
now
is.
A
E
Of
a
bug
fix
but
not
like
trivially,
obviously,
a
bug
fix
no
I
would
be
good,
I.
Think
I
would
propose.
We
say:
okay,
FCP,
you
worked
as
designed.
We
got
additional
comments.
That
said,
we
don't
want
to
do
this
anymore
and
we
put
this
on
a
list
or
maybe
we
consider
it
on
an
addition,
boundary
or
a
new
macro,
or
something
like
that.
B
A
B
A
E
A
A
C
A
B
E
A
B
C
D
Yeah
stabilized
brought
the
stabilized
brach
macros
chittering
macro
rules,
items
I,
don't
know
if
it's
fair
to
say
that
we're
all
relying
on
it,
but
it's
okay.
Well,
you
better
check
about
a
good
yeah.
A
Caracas
he's
pretty
young
widely
used.
Okay,
so
you'll
start
out
to
be
on
that.
Yes,
okay,
while
its
FC
peeing
I'm
gonna,
move
it
over
here,
so
that
we
well
guess
what
Felix
you're
the
shepherd.
To
the
extent
that
one
is
you
shouldn't
be
every
since
it's
not
it's
almost
in
the
last
phase
shouldn't
be
very
yeah.
The
report
is
already
done
right.
So,
okay,
cool.
A
B
B
So
it's
just
the
biz
thing:
yep,
you
might
yeah
summers
so
so
right
now
the
parser
works
it
up
a
steerable.
We
have
three
different
forms
of
like
items.
We
have
items
at
the
module
level.
Three
items
I
will
call
them.
We
have
trained
by
the
items
in
Chase
definitions
and
then
we
have
items
in
input,
definitions
both
for
inheritance
and
trade
rates,
and
and
so
it
happens
that
when,
when
you
are
because
then
so
the
plans
are
for
traits
definition,
specific
functions
or
trained
technicians.
B
Items
in
in
traits
do
not
support
invisibility
because
well
we
have
no
weight.
We
don't
support
that
semantically,
you
can't
say
like
crates
function,
food
with
insider
trades
bar,
but
syntactic
lead
up
and
that's
a
different
matter,
but
right
now
the
restriction
is
syntactic,
which
certainly
creates
a
problem
for
for
macros.
So
in
this
specific
issue,
the
problem
is
that
one
person
is
trying
to
use
the
same
macro
to
define
some
methods
in
in
both
the
in
both
the
trades
on
these
animals,
but
the
since
the
traits
traits
might
some
parser
doesn't
support
disabilities.
B
B
B
My
larger
point
is
that
we
might
want
to
unify
the
the
grammars
for
for
the
different
1940s
in
in
trade,
in
imports
and
in
module
level
and
then
moved
to
semantic
errors
instead
and
a
subset
of
that
might
be.
Let
me
just
do
it
for
functions.
For
example,
the
difference
between
the
right
now
is
in
methods.
We
support
cell
syntactically
in
we
don't
in
at
one
level
at
the
modular.
That
would
be
supports
lot
of
thoughts
at
the
end.
B
E
B
B
E
So
on
the
second
half
I
feel
like
the
language
definition
says
that
visibility
isn't
allowed
there
and
whether
that's
a
the
parser
accept
it
accepts
it
and
it
gets
rejected
later
or
it's
rejected.
If
the
parser
level
is
a
implementation
choice,
not
something
that
I
would
personally
write
into
a
spec
specifically.
B
D
B
We
allow
you
to
in
in
2015
in
traits
you
can
write
an
anonymous
promise
room,
but
just
in
type
not
not
the
button,
but
in
2018
you
have
to
supply
it.
So
the
pattern
on
the
type
so
yeah
we
can
sort
of,
like
plus
in
flags
and
and
stop
for
for
traits
and
non-paid
context,
and
that's
more
or
less
just
like
tossing
in
a
bull
and
propagating
it's
done.
Instructors
and
things
shouldn't
be
too
difficult
and
sort
of
doing
it
for
parameters
to
clean
things
up
and
unify
things.
But
that's
mostly
in
communication,
media.
D
B
E
A
Having
visibility
in
traces
when
I
migrated
the
quinces
but
great,
but
I
would
say
that
I
just
think
like.
Overall,
we
probably
want
our
grammar
to
be
concise,
as
we
can
make
it
than
having
like
too
close.
But
if
you
were
to
read
the
published
grammar,
it
would
be
like
too
similar,
but
different
definitions
of
functions.
It
just
seems
like
a
recipe
for
annoyance,
hey.
We
write.
B
E
The
on
the
minimal
part
I
think
because
macro
what
backward
based
on
the
tokens
that
they
omit,
not
the
actual
matcher
that
originally
matched
them,
because
you
can
use
an
ident
to
make
a
structure
a
union.
It
seems
like
just
a
straight-up
bug
that
if
he
is
literally
empty,
you
can't
have
it
there.
Yeah.
G
B
B
B
A
A
E
A
A
It's
there
are
some
cases
around
identifiers
specifically
but
like,
for
example,
if
you
parse
something
as
a
type,
that
is
just
one
word,
and
then
you
try
to
match
it
as
an
expression.
It
should
work
because
it's
just
identify
or
just
both
the
type
in
an
expression,
but
it
doesn't
yeah,
so
it
does
actually
matter
Wednesday
movies,
we're
in
the
implementation.
It's
actually
carrying
an
a
tag
DST
around
in
that,
unfortunately,
that
leaks
out
I
wish
they
weren't
the
case.
B
A
A
B
A
One
thing
I
will
say
is
never
mind.
Let's
just
call
it
I.