►
From YouTube: Lang Team Triage Meeting 2020.06.01
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
A
A
Looking
at
our
list
of
things,
I
see
talking
about
type
aliases
and
bounds
which
might
be
relevant
to
upcoming
editions.
They
decide
to
do
anything.
There
was
a
little
bit
of
prep
required,
but
not
that
much
one
save
code
guidelines,
I,
don't
really
think
we
have.
It
would
be
useful
to
talk
there.
I,
don't
know,
but
I'm,
not
sure
if
we
want
to
or
going
over
things
like
copying
out
of
references
and
safe
transient
working
group.
That's
where
we
kind
of
have
now
so
far.
A
A
C
B
B
B
If
that
is
implementing
everything
that
we
need,
how
that
might
transition
in
the
future
to
traits
generated
by
the
compiler
or
similar,
and
somebody
had
an
experiment
in
the
last
week
or
so
that
managed
to
use
constant
Eric's
to
almost
entirely
figure
out
things
like
whether
a
structure
has
alignment
requirements
entirely
in
the
type
system,
so
that
it
wouldn't
need
special
compiler
support.
So
there's
some
feeling
like.
If
we
fix
things
like
pasta
generics,
then
we
could
do
less
in
the
compiler
to
enable
this.
A
Okay,
that's
intriguing:
how
did
it
do
that
I?
Guess
you
take
advantage
of
the
we?
Don't
do
pentimenti
tail
thorns
and
links
on
that
topic,
constant
valuation,
anything
burning
to
add
your!
Oh!
Nothing.
D
D
E
Was
trying
to
remember
what
we
were
talking
about
around
when
we
want
to
ask
the
key
things,
for
that
seems
in
a
pretty
reasonable
spot
that
maybe
it
would
be
appropriate
to
take
it
forward,
but
I
was
trying
to
remember
if
we
said
we
wanted
to
get
basic
consensus
in
one
of
these
meetings
before
we
did.
That.
A
A
A
E
A
Ffi
unwind
I
think
Kyle
was
doing
some
work
towards
fine
is
like
a
few
minor
edits
just
want
of
time
hoping
to
have
an
artsy
soon.
One
minor
terminology
thing
we
settled
after
numerous
back
and
forth.
I
think
we
settled
on
a
term
plain
old
frames,
taking
off
of
plain
old
data
from
C++
for
frames
that
don't
have
a
pending
destructor
or
catch
unwind
sort
of
things
that
are
inert.
A
What's
the
phrase
that
was
using
before
those
are
significant,
because
there
are
the
kind
of
things
you
can
long
jump
over,
one
minor
point
that's
being
discussed
is
whether,
if
you
have
a,
if
you
have
a
destruct
like,
if
you
have
a
variable,
that
would
need
a
destructor,
but
it's
been
conditionally
moved
and
you
happen
to
know
that
that
move
occurred
somehow
by
program
logic.
A
Can
you
still
consider
it
a
plain
old
frame
or
not
like
if
the
compiler
doesn't
know,
but
you
do
and
I
I
don't
know
the
answer
to
that
question
actually,
because
it
kind
of
depends
a
little
bit
on
LOD
M
since
we're
gonna
flag
it
as
potentially
needing
to
run
so
I
think
we're
gonna.
Leave
that
I
don't
know
what
the
consensus
was.
I
should
look
back,
whether
to
I
think
I'm
I'm.
You
wanted
to
wanted
to
say
that
it
still
counts
as
a
plain
old
frame.
A
A
A
A
So
this
PR
is
applying
the
unsafe
up
in
the
unsafe
function,
lint,
which
we
recently
added
to
live.
Alec
I'm,
not
gonna,
add
it
to
the
paper
unless
there's
a
reason,
I
think
it
was
as
far
as
I
can
tell
it
was
highlighted,
so
we
can
take
a
look
and
kind
of
see
what
we
have
brought
and
decide
if
we
like
it,
I'm,
not
sure
if
it's
really
blocked
on
us
or
not,
it
seems
more
like
an
implementation
decision,
whether
to
do
it,
but
for
those
of
you
who
are
interested,
you
could
take
a
look.
A
It's
kind
of
what
you
expect,
there's
more
unsafe
bugs
in
the
bodies
of
things.
I
did
find
it
nice,
I,
guess
one
thing
that
was
sort
of
interesting
was:
you
know
cases
like
this,
where
the
actual
unsafe
operations
are
more
highlighted,
I
guess
than
they
would
have
been
before,
where
it's
a
bigger
function.
A
It
is
also
worth
noting
that
any
automated
automated
rewrite
well
at
least
what
we
have
I,
don't
know
you
can
sort
of
see
also
places
where
it
was
just
added
cover
the
entire
function,
and
it's
not
that
illuminating.
In
that
case,
as
you
might
expect
all
right,
there's
anyone
want
to
make
any
comments
on
that
I
did.
B
Look
through
a
great
deal
of
the
code
and
I
thought
that
the
overall
balance
of
having
many
of
them
become
more
fine-grained.
It
suggests
that
we're
going
in
the
right
direction.
It
also
suggests
to
me
that
there
may
be
further
improvements
we
could
make
in
terms
of
making
unsafe
I,
don't
want
to
say
easier
to
use,
but
rather
harder
to
misuse
or
easier
to
do
at
the
granularity
that
you
need
to.
But
nonetheless
this
seems
like
it's
showing
that
this
helps
at
least
to
me.
B
A
A
A
B
B
Stabilizes
anything
beyond
that,
so
I
think
that
we
have
some
flexibility
to
improve
this
in
the
future.
One
item
that
came
up
in
particular
not
in
this
discussion,
but
in
another
discussion
elsewhere.
Somebody
observed
that
instrumentation
attributes
don't
always
capture
the
full
details
of
what
function
was
Paul
and
it's
worth
investigating
whether
that's
true
of
trash
hauler
like
whether
it
passes
the
details.
A
A
A
A
A
A
We
never
changed
that
a
long
time
ago
and
I
think
this
is
trying
to
make
it
into
an
error,
which
I
think
is
what
we
thought
would
be
best
so
that
you
basically
changing
the
definition
of
C
like
to
be
cannot
implement
drop.
It
looks
like
we
did
create
a
run.
There
was
no
impact,
it's
up
on
up
to
us
to
decide.
If
we
want
to
approve
her
and
do
we
need
an
MCP,
I
guess
I.
A
A
E
B
B
C
E
A
E
A
E
C
Go
on
Felix
I
want
to
ask
because
I
think
it
was
Scott
said.
Well,
you
know
it's
copy
for
not
a
couple
of
things
copy
for
copy
things
and
move
for
non
copy
things
bo
is
trying
to
think.
Is
there
any
other
case
where,
as
works
on
non
copy
things,
I,
don't
I
can't
recall
any
other
precedent
appointed
for
what
to
do
here?
Ochs.
A
B
A
We're
so
that's
sort
of
a
compiler
team.
Technically
sorry,
the
rules
the
compiler
team
has
committed
you
to
would
permit
like
a
clean
grader
run,
allows
us
to
get
away
with,
will
run
down
nightly
and
see
what
happens
like
a
good,
a
good
practices
startup
cycle,
for
example,
to
give
as
much
time
for
people
to
raise
reports.
B
A
We
would
say
I
mean
it
depends
who
that
person
is,
but
if
it's
someone
who's
has
a
large
impact
on
the
ecosystem
and
they're,
probably
testing
lately
yeah
I,
don't
think
we
would
do
a
stable
backboard
in
thankies.
A
B
A
A
C
A
I
think
it's
good
for
us
to
delineate
our
own
just
our
own
decision-making
power
here,
but
I
I
think
we
can
leave
a
comment
that
says
so
consider
a
bug
fix
might
prefer
a
warning
period
or
one
release
and
I
probably
wouldn't
hard
to
put
that
either.
You
should
update
the
reference
yeah.
It's
actually
no
reason
not
to
do
a
warning
period
is
very
easy
to
do.
The.
D
A
Kind
of
a
WG
prioritisation
that
was
one
of
the
mission
statements
that
we
haven't
really
followed
up
on
I
think
was
that
they
were
supposed
to
help
us
keep
track
of
that
calendar
and
advance
it
all
right.
Let's
move
on,
because
I
want
to
get
there's
a
lot
of
stuff
here.
Raw
ass
macros
nominated
just
to
call
attention
to
yes,
probably
Felix
Scott
be
aware
that
we're
trying
to
add
these
raw
protein
macros
just
a
few
tons.
A
B
A
A
No
new
comments:
okay,
let's
leave
it
denominated
issues.
C
Before
before
we
move
off
with
this
stuff,
I
did
want
to
double
check
about
the
raw
rough
math
worst
thing
is
that
also
need
I
saw
ecstatic
Morris
mentioned
two
weeks
ago,
that
I
need
sign
up
in
T
lives
for
the
per
se
syntax
immunization.
Does
that
wait
until
stabilization?
Do
we
need
to
broke
them
in
now?
Our
next?
C
A
A
good
point
I
suppose
it
does
I
didn't
think
of
it
because
and
Rob
is
kind
of
purely
laying,
but
I
don't
know
like
why?
Don't
we
ask
them?
A
B
A
D
E
A
So
nominated
issues,
async
execution
can
get
lost.
This
is
all
about
option
of
must
use
nominated
again
because
I
don't
think,
there's
no
that
we
need
to
keep
talking
about
this
until
some
development
occurs.
A
A
A
A
A
B
Reading
the
I
definitely
want
to
understand
why
it
is
that
these
are
duplicative,
because
it
seems
like
one
is
saying
it:
let's
stabilize
the
ability
to
specify
one
and
the
other
is
saying,
let's
specify
the
behavior
when
you
don't
specify
at
those
seem
like
they're
compatible
as
if
one
is
saying.
Let's
panic,
if
you
don't
have
one
one
is
saying:
here's
how
you
can
stay,
please
specify
one.
C
B
Think
that's
true,
insofar
as
people
want
the
ability
to
handle
this
in
some
reasonable
way,
and
one
way
was
to
default
to
panic
and
then
it's
possible
for
people
to
override
a
panic,
Handler
and
another
way
is
to
let
people
actually
specify
an
allocator
error
handler.
So
it's
true
that
you
only
need
one
to
solve
the
problem
that
people
had,
which
was
handling
no
standard
with
allocation,
but
that
doesn't
mean
we
only
want
one.
That
just
means.
We
only
require
one.
A
Yeah
so
I
guess
it's
like,
then
we
all
right
now.
That's
enough
context
for
every
member.
What's
going
on
I
guess
but
I
think
the
objection
is
then.
Can
we
come
up
with
some
motivation
that
cause
like?
We
should
list
more
motivation
than
to
enable
people
to
write
things
on
standard
I.
Guess
I,
don't
know
it
seems
like.
The
motivation
is
not
that
hard
to
imagine
that
one
wishes
a
different
recovery,
I.
B
A
A
C
B
B
Great
so
config
version
and
configure
accessible.
There
is
a
desire
to
have
those
so
that
you
can
reference
things
that
may
or
may
not
exist
and
use
that
conditionally
to
decide
whether
you
should
implement
something.
If
we'd
had
it
back
in
the
day,
people
would
have
used
it
for
things
like
do.
I
have
you
128,
so
I
can
implement
types
for
implement
traits
for
it.
B
There
was
some
concern
expressed
in
previous
line
team
meetings
about
stabilizing
version
before
we
have
accessible,
because
that
may
then
lead
to
lots
of
version
check
throughout
the
ecosystem.
There
was
an
additional
concern
that
the
tone
they
raised,
which
was
that
the
way
that
we
decided
to
resolve
version
handling
for
nightly
such
that
150
nightly
will
report
itself
as
config
version
149,
but
not
config
version
150.
We
decided
on
that
because
150
is
not
exactly
a
complete
150
nightly
is
not
a
complete
150.
B
It
would
not
necessarily
have
all
the
features
of
what
stable
150
would
have,
whereas
beta
theoretically
does.
They
might
potentially
get
turned
off
if
there's
some
critical
issue,
but
there
won't
be
new
ones
added.
So
that
was
the
motivation
we
had
in
that
previous
meeting.
The
statement
from
D
tone'
was
that
that
would
also
prevent
people
from
testing
on
nightly
and
testing
the
same
code
they'd
like
to
use
on
stable
because
they
can't
just
add
a
config
version
for
150.
B
Config
version,
150
or
nightly
or
they'd
have
to
otherwise
somehow
handle
this
so
I
think
that's
worth
at
least
discussing
that
use
case
and
covering
is
this
the
desired
behavior?
Do
we
want
something
else?
I
have
thoughts
on
that,
but
I
just
finished
spending
a
bunch
of
time
talking
on
a
summary,
so
somebody
else
would
go
first,.
A
So
if
you
Tony's
workflow,
if
I
understand
is
the
idea
is
feature
gets
stabilized
at
a
certain
date
that,
after
sometime
after
that,
but
before
the
next
release,
various
libraries
insert
config
versions
that
will
start
enabling
that
feature
like
presumably
before
this
they
have
some
other
form
of
versioning,
but
now
they're
going
to
config
versions,
because
they're
testing
only
stable
things,
then
making
keep
testing
nightly,
builds
and
now
we'll
find
out.
If
indeed
they're
like
stabilized.
A
Some
problem
starts
to
occur
with
the
soon
to
be
stable,
behavior
and
if
that
makes
sense
to
me,
I
guess,
I'm
very
wary
ii.
Don't
know
like
I.
Think
library,
consumers,
like
you,
told
Nate,
have
a
lot
of
sway
here.
So
if
he
feels
really
strongly
one
way,
you
better
have
a
strong
argument.
Why
it's
wrong
make
you
like
our
prior
arguments
where
that's
wrong.
B
Insofar
as
if
a
particular
feature
has
been
stabilized
in
a
given
nightly
release,
then
once
that
feature
is
stabilized
it'd
be
nice
say:
do
I.
Have
that
feature
rather
than
a
Midas
version
and
I
would
expect
a
nightly
compiler
that
has
stabilized
that
version
or
that's
in
nightly
and
therefore
has
that
to
support
a
config
feature.
This
name.
A
Working
la
that's!
What
that's!
What
this
comment
says,
which
I
think
the
belief
is
that
we
one
would
primarily
be
using
config
accessible
for
sort
of
new
api's
that
have
been
stabilized,
but.
A
Right
there
might
be
some
features
that
are
not
accessible,
being
accessible.
A
Like
if
you're
on
the
nightly
ecosystem,
I
guess,
what
goes
wrong
is
that
you
you've
updated
for
a
newer
nightly,
but
your
consumer
is
using
an
older
nightly
right
right
and
in
principle
that
should
be
addressed.
The
gut
is
a
problem
period
and
in
principle,
well,
I,
don't
know
that
we
have
any
proposals
that
have
ever
talked
about
that
level
of
resolution
in
terms
of
being
able
to
specify
you're
compatible
with
this
version
of
rusty.
Normally
we
would
say
we
do
that
to
tie
it
to
rusty
versions,
I
guess
I.
B
I
would
express
a
mild
degree
of
skepticism
regarding
there
was
a
pushback
and
part
of
the
pushback
said.
You
know,
people
don't
run
the
old
Knightley's
I'm,
not
necessarily
saying
I
mean
they
told
me.
They
specifically
said,
for
instance,
providing
any
more
than
an
accidental
support
for
old
night
leads
as
a
lost
cause.
B
I
would
push
back
slightly
on
that
in
that
people
who
are
using
nightly
do
have
a
tendency
to
pick
a
version
in
that
version
and
then
build
with
that
version
until
they
pick
test
and
in
a
newer
version,
and
if
the
version
you
happen
to
pick
is
a
week
into
the
1:50
nightly,
and
that
version
declares
itself
as
150
and
you
leave
that
version
pinned
for
three
months
or
six
months.
You
may
experience
breakage
if
some
crates
says
I
want
150
and
it
expects
a
feature
that
came
later
in
the
same
nightly
cycle.
A
A
A
B
I
believe
it
originated
with
central,
both
config
version
and
config
accessible
and
I
heavily
pushed
for
the
availability
of
config
accessible
and
was
also
an
advocate
for
config
feature
of
some
kind.
I
was
never
an
advocate
for
config
version,
I,
don't
know
who
on
the
team
currently
is
that
doesn't
mean
there
isn't
someone?
We
also
don't
have
that
much
forum
today,
yeah.
A
A
B
A
Now
I
see
the
point
of
whoa
okay,
no,
it
seems
to
me
like
if
I
were
gonna
resolve.
This
I
would
want
to
sort
of
collect
the
use
cases
and
write
up
the
constraints
and
be
able
to
just
tell
like
it
sort
of
doesn't
matter
that
much
as
long
as
we
can
tell
it
does
matter,
but
if
we
can
tell
people
what
they
ought
to
be
doing
and
then
we
know
how
people
want
to
do.
We
pick
the
one
that
seems
overall
this
onerous
or
you
know,
and
for
whom.
A
A
A
A
D
A
A
C
A
C
A
Justin
report
I
would
check
my
box,
barring
any
surprises
with
him
like.
B
C
B
By
the
way
and
the
time
we
have
left,
we
have
three
of
the
six
members
of
the
team
currently
I'm.
Sorry,
four
of
the
six
members
of
the
team
currently
present
do
we
know
that
this
time
is
actually
working
for
all
six
members
and
there's
just
a
conflict
today
or
is
there
a
standing
issue
or
conflict
that
we
didn't
anticipate
when
we
sent
out
the
doodle
that
may.
A
A
A
C
Yeah,
it's
it's
tough
part
of
me
is
like
these
examples.
I'm
sit
here
and
go
well.
I
wouldn't
want
to
code
like
this,
but
it's
not
mine
and
then
make
a
clip.
You
lint
or
something
to
you
know,
try
to
rule
it
out,
and
I
too
would
like
something
to
highlight
that
it's
a
mutations
happening
but
I've
already
been
sort
of
instructed,
but
it's
not
really
consistent
with
where
things
happen
elsewhere,
I
mean.