►
From YouTube: Lang Team Triage Meeting 2019.10.10
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).
C
C
E
I'm,
not
quite
sure,
I
see
the
point
of
doing
the
group
a
little
bit
just
because
I
mean
I.
Guess
yeah
I'm
wondering
what
that
accomplishes
or
what
it's
supposed
to
do
in
that
regard,
because
we
have
kind
of
the
infrastructure
for
a
group
set
up
and
in
terms
of
having
a
doula
chat
and
repository
I
guess
the
only
value
I
see
is
kind
of
the
announcement
value
that
people
who
follow
new
rfcs
can
see
it.
E
E
A
B
B
A
C
C
B
C
C
A
E
B
An
drawback,
well,
some
I
mean
some
people
argue.
It
gives
too
much
weight
to
people
that
exploit
unspecified
role.
Ub
behavior,
which
I
do
not
think,
is
a
good
document,
but
as.
B
A
A
C
A
C
C
B
Where
they
use
temperature
mute
on
a
global
variable
and
a
static
static
that
has
static
lifetime
and
they
used
it
in
a
way
that
violates
a
blessing
uniqueness
like
they
use
the
road.
They
used
the
meat
of
a
thing
and
then
some
other
cauldron
that
could
access
the
thing
and
then
they
use
the
ordinate
of
the
reference
again
yeah.
C
B
So,
to
be
fair,
I
have
like
we
stack,
Toros
doesn't
I,
don't
know
what
the
stack
pero
story
if
aesthetic
is
its
currently
certain,
not
the
final
version.
But
yes
this
this
would
like
avoid
having
long
live,
mutable
references,
yeah
accidentally
I
mean
you.
If
you
have
a
role
pointer
and
you
the
Empress
mu
star,
you
can
still
get
arbitrary
unbounded
lifetimes,
but
that's
that's
how
world
partners
behave
everywhere?
It's
just
trying
to
reduce
this
particular
foot.
Cam
I,
don't
know
if
it
helps
it
just.
B
Yeah
particular
do
we
have
ampersand
row
and
it
seems
less
invasive
than
deprecating
segment
outright,
which
some
people
refer
getting
for
and
a
lot
of
hosts.
But
some
people
didn't
like
it
that
I
mean
you
would
have
to
add
unsafe
cell
in
places
which
it's
not
bad,
but
well,
it's
not
terribly
necessary
either.
Yeah.
A
B
A
D
C
C
D
Mean
because
there
is
the
existing
code
that
it
makes
impossible
right
is
just
having
static
mutes,
but
with
the
same
no
alias
behavior.
Are
you
saying
you
would
just
go
through
the
whether
that
you
get
from
airbrushing
from
you,
then
dereference
that
to
get
a
beautiful
reference,
you
really
wanted
to
do.
That
I
mean.
B
D
B
Yes,
if
the
middle
of
reference
is
unique
from
the
time
it's
created
at
the
time
is
used
the
last
time
everything
is
okay,
the
same
rules
are
the
same
for
all
of
them
right
and
I.
Don't
I'm
not
entirely
sure
if
the
the
lazy
static
Park
would
have
been
prevented
like
if
they
had
to
use
draw
pointers,
are
mutable
references.
B
It's
not
clear
to
me,
because
if
you
use,
if
you,
if
you
create
a
roll
pointer
to
a
local
and
then
you
use
the
logo
directly
and
then
you
use
the
raw
pointer
again,
that
is
you
be
because,
like
using
the
local
directory,
you
asserted
uniqueness
of
the
local
itself
of
the
place
directly.
So
it's
not
clear
that
the
lazy
static
you
be
would
have
to
invented
by
them
using
world
hundreds
of
mutable
references
is.
B
B
C
D
C
B
C
D
C
B
B
B
C
B
E
A
C
B
The
background
is
that
we
currently
check
the
value
of
constant
and
statics
items
to
make
sense
for
the
given
type.
This
is
something
that
all
he
started
and
I
also
worked
on
around
when
Mary
took
over
city
of
E,
and
it's
is
as
strict
as
we
could
make
it
well
within
reason,
I
mean
you
think
reason,
whatever
other
people
think
about
it.
B
Except
that
embedded
people
like
to
have
like
15
as
their
fathers
and,
of
course
like
for
a
long
time,
is
completely
fine
to
have
these
references.
But
we
are
extra
strict
here
because
those
so
the
the
the
biggest
problem
is
promotion
as
always
implicit
promotion
when
you
have
like
Emperor
and
something
something
something
in
a
function
and
the
company
decides
that
something
something
something
could
be
computed
at
compile
time
to
get
the
M
percent
of
static
lifetime.
B
And
we
need
to
be
really
sure
when
we
do
that
that
we,
when
we
constantly
about
that
code,
there
won't
be
an
error
as
if
there
is
an
error.
We
screwed
up
bad,
because
this
was
code
that
worked
like
we
broke
the
users
code.
The
code
would
have
compiled
fine
and
probably
run
fine
at
runtime,
but
we
decided
to
write
a
compile
time
and
then
program
doesn't
compile
and
users
frustrated.
B
B
I
think
items
might
or
might
not
currently
and
in
the
future
and
so
on
flow
into
this
yeah.
We
are
extremely
conservative
with
what
we
allow
there.
That
said,
we
are
probably
never
gonna
want
to
allow
dereferencing,
a
pointer
and
things
to
promote.
That
seems
like
asking
for
way
too
much
further.
I
already
think
we
are
promoting
way
too
much.
I
I'm
I
have
been
promoting.
He
and
more
explicit
mechanism
that
you
could
read
syntax
like
I,
know,
Const,
curly,
brace
something
something
something
I'd.
F
B
B
B
If
it
all
the
user
in
this
embedded
platform
tasks,
this
GPL
pointer,
which
is
which
is
some
integer
class
to
a
pointer
and
then
they
somewhere,
right
and
percent
and
then
a
bunch
of
code
which
involves
dereferencing
this
pointer,
and
we
would
look
at
that
code
to
be
like.
Oh,
we
can't
promote
this
because
surely
they
want
the
static
lifetime
here
and
then
we
can't
steal
it
and
the
dereference
is
this
integer
and
consumers
like?
No,
you
can't
be
reference,
15
I,
don't
know.
What's
there
and
then
an.
B
A
C
C
B
G
B
D
I
mean
I,
actually
I
have
a
question
about
this
specific
use
case,
and
maybe
this
is
too
in
the
weeds
but
like
when.
Would
it
be
correct
to
have
an
a
static,
ampersand
reference
to
it
if,
like
a
fixed
address,
meaning
that
you're
reading
like?
Is
that
a?
Is
that
a
reasonable
thing
to
do?
Oh
I
mean
this
would
have
to
be
that
doesn't
ever
change
right.
D
B
B
B
B
So
I
mean
yeah
I
mean
you
have
to
make
all
your
weeds
volatile,
obviously
because
data
rates
and
stuff
but
yeah
it
sounds
like
you're
suggesting,
oh
no,
but
that
the
star
is
still
kind
of
weird,
because
you're
doing
a
non-volatile
reads
from
something
that
like
could
change
under
your
feet.
So
it's
kind
of
like
a
data
race.
But
you
could
have
external
synchronization
around
to
see.
D
C
B
D
Worthwhile
pointing
out
that,
then
you
hit
the
issue
that
everybody's
trying
to
mark
every
block
and
every
expression
they
can
con
so
as
not
to
miss
out
on
some
optimization,
which
is
sort
of
yeah
okay.
So
we
have
two
plus
plus
like
patterns
like
this
right,
where
you
just
are
supposed
to
declare
a
million
things
in
order,
I
mean.
F
Yeah
elavil
will
optimize
this
for
sure
anyway.
So.
C
So
I
mean
I
think
this
seems
like
a
fine
proposal
when
I'm
a
little
bit.
How
would
we
like
decide
that,
like
this
kind
of
gets
to
the
question
of
the
cost
of
valuation
story
and
how
its
evolving
and
integrating
into
LinkedIn
put
like,
is
there
somewhere?
Where
there's
like
notes
on
the
current
status,
and
we
can
kind
of
add
this
in
and
be
narrowing
down
the
uncertainty
until
the
rules
are
relatively
fixed,
as
here
I'm
asking
where
we
log
this
decision.
C
B
I
mean
I
would
be
happy
if
there
would
be
designed
about
promotion,
but
they
isn't
okay
promotion.
There
was
an
initial
RFC
which
likes
kicks
out
of
it
and
ever
since
then,
it's
just
been
any
time.
A
new
language
feature
was
added,
people
say
in.
Is
there
a
way
because
possibly
sound
to
promote
this,
and
if
anybody
was
clever
enough
to
come
up
with
one
and
unfortunately
is
really
clever,
then
we
started
promoting
it.
B
So
we
are
promoting
way
too
many
things,
and
this
doesn't
happen
by
this-
didn't
happen
with
design.
This
was
just
like
with
the
current
shaking
infrastructure
and
with
the
current
everything.
Can
we
convince
ourselves
that
promoting
this
okay
and
can
be
implemented
reasonably
and
unfortunately,
that
was
added
him
completely
arcane
behavior,
dictated
by
like
the
way
the
checking
worked,
and
now
a
bunch
of
things
are
promoted,
that
I,
don't
think
should
have,
has
been
promoted
and
then,
whenever
I
see
to
be
promoted,
I,
just
like
oh
yeah.
Let's
just
add
this
to
the
list,
so.
A
B
I
think
I
mean
Ely
would
have
someone
who
was
willing
to
do
some
design
and
like
write
an
RFC
which
hopefully
has
next
visit
promotion
mechanism
and
describes
our
current
existing
interested
in
promotion
routes
and
all
of
that
stuff,
some
of
it
is,
is
well
basically
work.
That
should
have
happened,
but
didn't
so
it's
catching
up
with
like
documenting
status
quo,
and
some
of
it
is
then
like
designing
a
way
forward.
D
C
A
D
G
And
then
there's
also
using
something
as
part
of
an
array
initializer
and
then
passing
something
to
intrinsics,
like
Cindy,
shuffle
where
it
like
really
needs
something
to
be
conservative
as
a
workaround
for
the
absence
of
constant
Eric's
and
then
there's
so
that
is
as
far
as
I
could
tell
not
fully
documented.
So
we
should
probably
do
that
and
here's
the
list
of
operations
which
are
allowed
to
be
promoted.
I
think
it
would
be
amazing
if.
B
B
B
And
it
seems
like
this
might
be
a
place
where
you
have
two
sets
of
promotion
rules.
We
have
one
set
of
emotional
rules
that
we
apply
in
run.
Time
functions,
another
set
that
we
applying
constants
and
not
the
same
rules,
because
it
costs
items
the
user
asked
for
it
being
constables.
We
can
more
aggressively
promote.
A
C
So
it's
time
check
it's
3:40
in
my
local
time
anyway,
it's
40
minutes
past
the
hour
and
what
I
don't
know?
Maybe
some
of
you
want
to
have
our
base
times
anyway?
Sorry,
the
thing
we
have
40
minutes
into
the
meeting
and
we
have
20
minutes
for
me
and
I
feel
like
we
kind
of
got
pretty
I,
feel
like
we're
done
with
this
and
I
wanted
to
check
what
we
should
discuss
next
one.
A
C
G
G
C
Was
reinvent
these
two
wheels,
so
they
were
like
two
super
complex,
amazing
things,
so
I
just
kind
of
like
ties
in
with
what
we
were
saying
earlier
off
about
the
unsafe
code
like
how
should
we
try
to
I
would
basically
like
to
have
some
insight
into
what
you
all
think
are
important
than
what
you're
working
on
and
bring
it
back
here
from
time
to
time
and
I.
Think
things
like
documenting
these
rules
would
be
awesome.
A
A
B
B
B
G
B
G
F
G
G
A
G
D
A
C
B
Have
formal
methods
to
treat
with
I?
Don't
I,
don't
know
anything
about
it?
No,
so
the
the
problem
is,
this
is
the
first
like,
so
there
wasn't
that
wasn't
to
say
that
proxy
has
the
same
ABRSM
fraternity
and
that
passive
ever
a
reasonable
stuff.
But
what
this
also
does
is
add
an
example
for
calling
a
C
function
way
on
the
right
side.
The
signature
we
give
is
a
box
T
and.
D
B
B
This
cover
is
that
you
can
actually
have
very
crazy
kind
of
you,
be
if
you're
doing
bad
things
with
you
F
of
I
imports,
because
whatever
I
imports
are
it's
like
the
same
thing
as
declarations
in
C
and
in
C
its
you
be.
If
you
have
multiple
declarations
that
have
incompatible
types,
incompatible
is
a
bit
unclear,
but
what
this
means
in
LVM
is,
if
you
have
declare
functional
multiple
times,
and
it
is
no
alias
in
one
case
are
not
no
alias.
In
the
other
case,
the
narrators
can
like
move
over.
B
So
basically
the
no
alias
a
trust
adds
to
the
FI.
Import
can
infect
the
C
code.
That
is
important
if
you
do
X
SEO
and
compile
this
all
together.
So
it's
as
if
the
C
code
had
no
alias,
even
though
the
C
code
doesn't,
which
might
not
like
that's,
certainly
very
subtle,
does
mister
can
introduce
you
be
into
a
program,
but
just
adding
a
declaration
that
you
never
a
call,
because
it
adds
an
attribute
that,
like
break
stuff,
which
is
not
even
unsafe
code,
lots
of
us
that's.
B
It
to
the
C
side
as
well,
but
yeah
I,
don't
think
this
is
yeah.
So,
yes,
it's
mostly
orthogonal
and
it
seems
like
something
we
well.
We
could
even
we
could
either
decide.
This
is
a
prop
you're.
Gonna
have
to
sort
of
actually
right
now,
I'm,
not
saying
anything
about
it
and
go
ahead
with
it
or
we
could,
besides,
that,
the
problem
which
we
could
try
to
do
by
mandating
people
to
use
well
upon
us,
but
they
already
don't
so
yeah.
F
C
B
But
we
didn't
decide.
There's
no
example
anywhere
saying
that
you
can
use
em
press
em,
you
don't
FF.
I
imports
at
least
I
haven't
found
one
and
not
enough.
I
export
so
like.
I
don't
even
know
where
to
look
for
this.
I
don't
know
why.
I'm
reviewing
this
PLO
understand
what
does
every
time
you
even
have
to
do
this.
C
B
But
isn't
so
this
is
well.
Maybe
there
is
like
I
would
be
happy
if
there
is,
and
then
I
can
just
our
plus
for
referencing
that
and
then
it's
someone
else's
fault,
but
I
didn't
find
one
and
well
I.
Looked
in
the
reference.
Talk
like
like
the
reference
type,
primitive
type,
STD
documentation,
which
is
what
I
would
have
expected
it
that
there's
nothing
there
and
the
nomicon
FFI
is
chapter
also
I,
think
doesn't
and
then
I
didn't
know.
Where
else
to
look
and
the
FFI
experts
are
important
to
the
Fed.
A
D
C
B
C
B
C
C
B
C
C
B
This
elf
receivers
I
asked
some
time
ago
and
some
PR
I
was
CCD
and
I
asked
about
like
precise
documentation
in
terms
of
what
the
contract
is,
that,
whatever
things
I
involved
have
to
actually
satisfy
to
make
all
that
work
and
what
the
output
was
like
yeah.
We
should
do
that,
and
this
would
only
really
become
relevant
once
we
have
recursive
safe
receivers,
so
it
seems
like
if
they
are
coming,
then
what
is
the
source
of
that
documentation?.
C
B
Of
documentation
yeah,
some
of
those
seems
very
subtle,
from
what's
actually
going
on
at
one
time
perspective
and
like
imposing
contracts,
maybe
on
Dieruff
or
something
maybe
not
I,
don't
know
and
I
mean
so
far,
I'm,
not
sure.
If
there's
any
answer
I've
gotten
to
action
here,
but
one
day
is
I,
have
no
I
mean
probably
will
be
once
there's
usually
find
self
receivers,
because
then
you
have
to
satisfy
a
certain
contract.
I
guess
to
make
that
sound.
C
B
A
B
D
C
B
C
The
only
other
thing
I
want
to
note,
as
we
even
passing,
is
that
for
those
of
you
who
weren't
here,
it's
not
yet
known.
If
we're
going
to
do
an
addition
in
2021
or
what,
but
it
would
be
useful
to
think
about
if
we
were
to
do
it,
what
are
the
kinds
of
things
you
might
try
to
change
at
that
time?
I
think
that
would
affect
the
decision.
C
C
G
B
B
B
So
that
helps,
which
is
also
a
backwards
compatibility
hazard,
so
might
be
an
additional
thing.
So.