►
From YouTube: 2020-07-06 Triage Meeting
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
Should
this
better
so
I'm
about
so
first
of
all,
I
filed
this
meeting
proposal
just
two
seconds
ago,
which
I
didn't
have
time
to
write.
Much
and
I'll
write
a
little
more
some
more
concrete
thoughts
here,
but
this
came
out
of
a
conversation
Josh
and
I
were
having,
but
I
would
like,
ideally
this
week.
If
we
had
time
to
talk
about
more
kind
of,
follow
up
on
the
discussions
around
the
new
procedures
and
the
Pat
will
talk
about
laying
team
path
to
membership.
A
In
some
ways
we
could
manage
that,
and
but
one
of
the
things
I
want
to
talk
about
is
okay,
we're
starting
to
get
some
some
major
change
proposals.
We
need
to
think
about
how
we
process
them
and
how
we're
gonna
answer
them,
and
one
of
the
goals
of
the
whole
idea
of
MCP
was
to
sort
of
force
us
to
ask
the
questions,
and
sometimes
the
answer.
A
No
or
you
know
if
we
don't
have
bandwidth
or
if
we
do
and
basically
to
be
a
bit
more
honest
and
clear
about
what
we
are
doing,
or
maybe
honest
is
the
wrong
term.
We're
always
honest,
just
clear,
so
I
bring
this
up
to
point
out
that
you
have
these
three
groups.
So
far
we
haven't
gotten
back
to
anybody.
Who's
filed
an
NCP
which
seems
suboptimal
to
me.
A
We
had
talked
about
two
ways
to
answer,
or
three
may
be
one
is
to
say:
no,
we're
not
gonna
do
that
at
this
time,
because
nobody's
excited
whatever
reason.
Another
was
to
say
that
we
do
have
interest
in
doing
it,
but
there's
the
person
who
is
interested
in
it.
It's
kind
of
got
their
hands
full
right
now.
It
doesn't
really
have
time
to
follow
up
and
does
expect
to
have
time,
I.
A
A
So
if,
if
anybody
here
wants
to
signal
that
they
do
have
time
or
they
don't
have
time
that
they
are
interested
and
they
think
they
would
have
time,
maybe
after
the
summer
or
something
like
that
or
like
they
can
sort
of
identify
when
they
might
have
time,
then
I
think
there
should
talk
on
how
to
do
that.
I
think
you
would
just
self-assign
and
leave
a
comment
and
I
would
basically
move
it
over
here
to
this
short
list
of
stuff
we
might
do
or
stuff
we're
waiting
on.
A
We
were
in
queue
to
do
next
and
we
could
discuss
exactly
what
you
want
to
do
in
the
meantime
like,
while
that's
in
the
shortlist
I
don't
know
if
it
should
have
its
own
zoo
live
stream,
so
that
the
people
who,
like
maybe
the
lang
team
member,
is
not
able
to
interact
very
much,
but
other
people
want
to
keep
pushing
on.
In
the
meantime,
that
seems
okay,
although
at
that
point
it's
almost
a
project
group.
So
maybe
that
also
doesn't
make
sense.
A
So
yeah
I
guess
that's
all
I
had
to
say
just
to
point
out
that
we
have.
These
I
would
like
to
kind
of
start
to
answer
people.
One
of
the
things
in
this
list
is
this
one,
which
was
me
saying
something
I
would
like
to
do,
but
I
want
volunteers.
I
haven't
really
thought
about
how
we
should
advertise.
This
I
have
been
thinking.
I
want
to
do
more
I
kind
of
given
up
on
19
blog
posting
and
I
plan
to
change
that
by
writing
up
the
review.
A
Safe
transmute
meeting
just
write
a
short
blog
post,
saying:
hey
this
happens,
it
was
cool.
Tipic
is
cool.
Here's
a
little!
Here's
a
movie
you
can
watch
I
could
imagine
us
using
that
opportunity
like
if
we
had
regular
blog
post,
we
might
advertise.
You
know,
what's
happening
with
the
major
change
proposals.
Keep
people
abreast.
These
things
were
filed.
There
are
some
Help
Wanted
requests
if
you're
interested
in
getting
involved.
This
might
be
something
to
pick
other
words.
I
should
actually
ask
myself
if
I
really
had
time
to
work
with
this.
B
Wanted
to
expand
on
the
notion
of
assigning
yourself
in
the
interests
of
like
expressing
interest.
I.
Think
part
of
the
goal
here
would
be:
let's
try
to
provide
more
visibility
into
our
respective
backlogs.
So
if
you
already
have
something
going
on-
and
this
is
more
I'm
tagging
this
because
I'm
interested-
then
we
can
tag
it
like
that
and
then
leave
it
in
nobody's
working
on
this
right
now.
B
But
we
are
generally
in
favor
in
as
a
column
here
and
separate
from
that,
if
you,
even
if
you
don't
have
something
you're
currently
serving
as
liaison
for
it's
still
possible,
to
tag
something
as
being
interested,
but
not
actually
have
it
be
at
they've.
Yet
so
I
think
it's
important
that
we
maintain
this
distinction
of
who
would
be
willing
to
be
a
liaison
given
appropriate
bandwidth
and
actually
having
bandwidth.
C
B
That
was
the
intention
is
there
might
be
for
people
who
are
self
assigned,
because
they're
all
willing
to
liaison
then
when
one
of
them
gets
bandwidth
and
wants
to
slot
something
new
in
their
queue,
if
they
take
something
they've
already
self-assigned
remove
the
other
people
assigned
to
it
and
put
it
in
the
active
list.
I
mean.
A
B
C
A
A
The
idea
was,
if
you
assign
yourself
to
a
proposal,
it's
saying
I
want
to
see
this
go
forward,
but
you
know
you
can
leave
a
comment
clarifying
the
timeline
you
have
in
mind
and
I
think
the
shortlist
is
totally
fine
and
wildon
like
not
if
not
immediately
as
a
valid
timeline.
C
A
C
A
Thing
that
sounds
right,
I
think
this.
This
project
board
is
incomplete
and
doesn't
contain
all
the
old
like
either
old
or
current
activity
and
I
sort
of
think
it
should.
So.
You
know
whether
that
means
creating
an
MCP
or
we
can
add
arbitrary
things
on
here.
So
I
could
add,
like,
like
I,
have
here
actually
an
existing
RFC
inactive
projects
column.
A
We're
going
to
be
a
little
more
selective
I
would
like
at
some
point
to
sit
down
and
I.
Had
it
the
way
I
imagined
it
is
that
we
would
do
some
through
the
new
system
until
we're
kind
of
happy
with
how
it
works
before
we
start
doing
this,
and
then
we
would
go
over
those
and
like
pick
the
ones
that
seemed
most
promising
and
encourage
them
to
open
a
new
MCP,
but
or
do
it
ourselves,
but
like
and
move
it
through
that
system
or
just
make
a
group,
we
don't
necessarily
make
an
MCP.
A
A
A
Tracking
tainted
expressions
and
generally
stabilizing
the
exclamation
point,
I've
been
talking
to
blitz
there
a
little
bit
I'm,
not
sure,
we'll
see
where
that
goes,
I'm
still
kind
of
looking
for
someone
to
implement
that
on
these
topics.
Some
of
these
that
are
kind
of
stalled,
I'm
thinking
of
making
another
column
for
like
stalled
projects
is
to
recognize
that
distinction.
A
I,
think
this
coherence,
bypassed
by
indirect
imple,
might
be
a
good
example
of
a
stalled
project
which
I
do
you
want
to
see
us
pursue
and
they
haven't
had
the
time
to
get
back
to
it,
but
Stallman's
we
did
to
call
the
knob,
because
that
might
be
a
place
for
somebody
else
who
hasn't
time
to
pick
it
up
all
right,
no
update
here.
Unless
anyone
has
any
comments
on
coherence
can
be
bypassed,
but
projects
safe,
transmute,
so
now,
I'm
meeting
is
there
much
to
say
beyond
that
I
don't
know
any
updates
for.
B
This
I
think
that
debate
on
the
basis
of
that
meeting,
people
are
proceeding
forward.
Earth.
Okay,
what's
the
next
step
and
I
think
the
the
question
is
going
to
be:
what
can
we
do
to
unblock
ongoing
work
in
that
area?
We
expressed
some
general
interest
of
this
sounds
about
right.
You
know
you're
in
going
in
the
right
direction.
The
question
is
now:
what
is
the
how
much
further
should
they
go
before
checking
back
and
saying?
Here's
where
we're
at.
A
A
D
The
other
thing
I
remember
there
was
having
that
intrinsic
or
auto
implemented,
trait
whatever
it
is.
That
would
enable
the
library
team
to
go
and
figure
out
what
this
should
actually
look
like
to
users,
but
from
a
language
that
it
was
mostly
just
exposing
some
internal
things
so
that
that
was
even
possible.
A
B
B
C
E
B
That
we
generally
gave
the
guidance
or
had
talked
about
the
idea
that,
if
a
feature
isn't
even
Specht
by
the
end
of
2020
start
of
2021,
it
probably
shouldn't
be
going
into
the
2021
edition.
But
something
that
has
an
accepted
RFC
and
implementation
is
being
experimented
with
in
2020.
Then
stabilizing
that
in
the
2021,
Edition
seems
pretty
reasonable
hand
wave
and
wave.
So.
A
A
C
A
A
A
A
B
One
quick
follow-up
on
the
FFI
on
one
thing,
I
would
say,
based
on
some
of
the
discussion
that
we
had.
If
there
were
an
ABI,
we're
unwinding
is
just
in
an
eighth
part
of
the
the
ABI.
We
might
want
to
consider
whether
that
should
be
the
default,
but
for
the
most
parts-
and
you
know,
system
and
win
API
and
all
of
the
other
api's
were
aware
of.
Don't
have
that
as
a
consideration
that
there
isn't
an
innate
everybody
uses,
unwinding
and.
A
B
Said
what
I
would
like
jesting
is
like
if
hypothetically
some
ABI
like
standard
call
was
specifically
used
primarily
by
functions
whose
error
handling
was
always
some
sort
of
structured
exception,
handling
based
on
unwinding,
then
it
wouldn't
be
very
useful
to
have
a
non
unwind
variant
of
that,
and
we
would
talk
about
hey.
Should
the
default
be
this,
so
we
don't
break
a
ton
of
people.
We
don't
have
such
an
API
that
we're
aware
of
so
I
think
it
makes
sense
that
all
of
their
ABI
is
followed.
The
same
pattern
of
default,
no
unwind.
A
Right
I
guess
like,
for
example,
in
Windows,
it
is
quite
quite
normal
to
unwind,
but
only
in
the
event
of,
but
only
the
forest
unwind
case,
which
we
already
said
can
still
occur
family.
However,
what
I
would
say,
probably
in
the
case
like
you're
describing,
is
that
we
should
still
they
may
not
have.
We've
always
still
want
to
name
it.
Fou
unwind,
for
consistent
by.
B
A
C
B
A
B
B
B
A
B
A
See
so
is
this:
this
is
removing
so
that
we
now
support
impulse
for
any
arbitrary
limits.
B
That's
strange
zoom
keeps
kicking
me
out
of
the
meeting
and
making
me
rejoin
and
I.
Don't
know
why
so
yeah
we
took
that
conservative
approach
so
that
you
could
always
drop
in
manual
impulse
or
macro
generated
impulse
up
to
a
certain
fixed
size,
and
if
we
accept
this
PR,
we
are
committing
stabili
to
the
notion
that
we
will
always
have
enough
of
constant
Eric's
available,
at
least
in
nightly,
to
the
compiler
and
library
itself,
standard
library
itself
to
be
able
to
implement
things
for
arbitrary
sized
slices.
B
That
doesn't
mean
we're
committing
to
stable,
constant
Eric's
or
to
any
particular
cement.
It
just
means
we
are
stabili
providing
a
set
of
impulse
for
any
arbitrary
slice
size,
which
means
we
always
have
to
have
a
means
of
doing
so
in
the
future.
Whatever
that
means
might
be
I
think
that's
reasonable,
I
think
that
that's
not
a
full
commit,
or
we
will
always
have
constant
generics
and
we
will
have
them
available
stable
yet,
but
I
think.
The
idea
that
we
have
some
feature
available
for
stood
to
use
is
reasonable,
agreed.
B
A
B
You
were
given
and
use
it
exactly,
and
that
would
be
a
subset
that
may
be
easier
to
stabilize
where
you
can't
stay
in
+1
and
you
can't
necessarily
say
I'm
using
the
size
of
this
other
thing
as
computed
in
the
type
system
as
my
slice
length,
but
you
can
take
an
in
you
were
given
and
put
it
in
the
end
of
a
slice
and
that's
enough
to
implement
this,
for
example.
So
it
would
be
kind
of
a
conservative,
Constitution
Eric
so
that
we
might
be
able
to
stabilize
sooner
min.
B
A
A
A
Okay,
I
would
like
to
point
out
this
target
feature.
This
is
the
question
of
whether
Cindy
the
next
order
of
business,
I-
guess
seven,
three,
six
three
one,
whether
or
not
Cindy,
whether
or
not
closures
inside
a
function
should
be
able
to
assume
the
target
features
fold.
In
other
words,
once
you
enter
a
target
feature
function,
do
we
know
for
all
time,
then
you
can't
take
that
target
feature
back,
Josh
I
think
you
were
in
in
favor
of
this.
Oh
I
see
yes
yeah.
So
I
just
wanted
to
point
out
it's
here
right.
B
I
want
to
state,
for
the
sake
of
completeness,
the
assumption
that
this
makes
is
that
once
you've
run
code,
that
is
part
of
a
target
feature
function,
then
you
will
always
be
able
to
use
that
target
feature.
So
you
should
be
able
to
make
a
closure
that
uses
that
feature
and
pass
it
around.
The
caveat
to
that
is
in
theory.
You
could
have
some
complicated
architecture.
Where
say
you
have
a
symmetric
multiprocessing
with
big
cores
and
little
cores.
B
You
know
either
arm
or
Intel
or
other
implementations
of
that
concept,
where
some
of
the
cores
have
a
particular
processor
features
that
feature
and
some
of
the
cores
do
not
so,
theoretically,
you
could
have
to
deal
with
that.
That
would
break
a
huge
amount
of
code
already
in
the
wild,
not
just
rust
code,
but
C
code
and
other
people's
code.
That
just
assumes
you
can
check
cpuid
once
or
the
equivalent
once
and
then
run
code,
and
so
any
introduction
of
such
a
mechanism
there's
been
discussion
in,
say.
B
The
Linux
kernel
community,
for
example
of
you,
could
ask
for
an
explicit
lease
to
use
scoped
features
and
you
have
to
say
I'm
willing
to
do
that
or
you
don't
get
those
features
at
all
and
you
get
instead,
the
subset
that
runs
on
every
processor,
so
the
notion
would
be
yeah.
You
know
this
processor
has
av
X,
but
this
one
doesn't
or
this
processor
has
I'm.
Sorry
I
can't
remember
the
name
of
the
arm,
vector
extension,
but
the
same
kind
of
concept.
B
Then
you
would
have
to
ask
for
it
and
then
you
will
only
run
on
the
big
cores
until
you
stop
asking
for
it,
but
and
that
would
be
the
kind
of
thing
that
would
be
necessary
to
avoid
breaking
code.
That
makes
assumptions
like
this.
If
that
were
ever
done,
then,
yes,
we
need
some
additional
mechanism.
If
we
wanted
to
track.
Oh,
you
shouldn't
keep
around
closures
that
assume
that
you
can
use
this
target
feature
after
you
have
dropped
the
lease
that
says
you
can
use
this
target
feature.
B
There
are
ways
we
could
make
that
work
more
safely,
without
keeping
it
for
the
duration
of
a
process,
but
at
the
same
time,
I.
Don't
think
that
we
should
block
our
current
simplistic
architecture
like
something
that
works
for
the
majority
of
cases
for
the
majority
of
people
on
the
particular
implementation,
details
of
some
future
CPU
architecture
that
doesn't
necessarily
even
have
this
property.
Yet.
D
B
A
A
A
When
you
or
when
you
enter
into
a
target
future
specific
thing,
the
unsafe
obligation
on
your
part
is
to
prove
that
that
feature
is
available
and
what
we
are
kind
of
changing.
Well,
we
I
guess
what
you
would
have
had
before
is
that
you
must
prove
that
it's
available
for
the
duration
of
the
function.
A
Another
thing
is,
you
have
to
prove
that
it's
available
for
the
duration
of
the
function
or
anything
that
escapes
from
that
function,
which,
in
the
case
where
which
is
easy
to
do
in
the
case
where
you
don't
are
not
in
this
a
symmetric,
multiprocessing
setup,
because
you
don't
need
to
know
what
escapes
from
the
function,
because
it
never
changes
in
the
case
where
you
are
in
that
setup,
then
it
becomes
more
difficult.
You
would
actually
have
to
know
the
function,
definition
or
like
know
something
about
what
might
escape
from
it
right.
B
Matter
you
could
have
a
lifetime
based
approach.
Like
hypothetically,
you
could
have
a
scoped
I'm
requesting
this
target
feature
which
has
a
lifetime,
and
if
you
run
a
function
inside
that
that
uses
that
target
feature,
it
can't
outlive
the
lifetime
of
that
lease.
For
example,
yeah.
There
are
ways
we
could
do
this,
but
that
would
require
designing
in
the
future
that
takes
into
account
stuff
that
may
not
exist
yet
so
I
would
suggest
we
not
worry
about
it
right
now
that.
A
Have
for
a
long
time
had
the
observation
that
maybe
function?
Pointers
should
have
had
a
lifetime
bound
on
them
because,
for
example,
with
dynamic
loading,
it
would
be
nice
to
be
able
to
say
there's
a
function
pointer,
that's
valid,
but
not
forever,
because
I'm
gonna
unload
this
library
later
on,
and
we
can't
currently
do
that.
I
think
it
seems
analogous,
and
there
may
be
a
mechanism
that
could
be
introduced
it
somewhere
in
the
future
to
address
both
problems.
A
A
A
Hybrid
architecture,
where
CPU
features
can
change
you
would
either
you
would
have
to
know
whether
function,
pointers,
function
and
closure
pointers
escape
and
be
sure
that,
and
for
how
long?
This
may
imply
new
annotations
that
assure
you
that
they
do
not
escape
or
perhaps
that
in
a
lifetime
annotation
lifetime
annotations
could
also
be
useful
for
dynamic
loading
unloading
scenarios
for
our
functions.
A
A
We
talked
about
this
before
this
is
pretty
old,
RFC
I,
don't
know,
I'm
somewhat
inclined
to
just
cancel
that
day,
we're
still
in
favor,
potentially
basically
move
to
postpone
the
RFC
on
the
basis
of
this
sounds
like
a
good
idea.
Maybe
somebody
could
pick
this
up
and
file
an
NCP
if
they're
still
interested
I
mean
might
or
might
not
have
been
way
to
do
it,
but
this
is
still
enough
that
I
think
it
should
be
closed
and
this
current
champion
is
not
around.
So
does
anyone
disagree.
A
Okay,
so
no
plans
to
cancel
and
move
to
postpone,
in
fact,
maybe
I'll,
just
close
I,
suppose
one
but
encouraging
moving
to
postpone
and
moving
to
close,
always
feels
a
little
good
makes
sense,
but
also
in
kind
of
everybody's,
going
to
go
check
their
boxes.
Just
to
do
nothing
encouraging
folks
to
consider
opening
and
MCP
for
parts.
A
A
A
A
B
A
B
A
quick
update
on
move
ref
pattern,
the
one
where
we
said
it
would
be
nice
to
see
usage
in
the
ecosystem.
I
just
checked
a
complete
search
of
github
for
anybody
who
hasn't
found
it
yet
grep
dot
out
is
a
gray
github
search,
I
just
checked
and
there
is
no
reference
to
the
move.
Ref
pattern
feature
gate
in
we're
outside
of
the
rest:
compiler.
The
rust
blog
announcement
this
week
in
rust
and
a
list
of
rust
feature
gates.
So
nobody
on
github
has
used
it
at
all.
B
A
B
A
Leave
a
comment
to
that
effect:
I'm
I
personally,
would
still
go
forward
with
it.
Just
cuz,
it's
like
weird
code
that
you
know
logging
things
for
no
particular
reason.
Alright
things,
everyone
good.