►
From YouTube: Lang Team Triage Meeting 2019-11-14
Description
Triage Meeting 2019-11-14
B
B
A
B
A
So
I
didn't
yet
make
a
lot
of
progress.
Okay,
we
ran
the
prayer-meeting
crater.
There
was
some
impact.
I
was
able
to
identify
a
few
changes
that
would
be
easy
to
fix
and
I
need
to
do.
A
rerun
I
found
out
that
there's
a
way
to
use
crater
where
it
only
runs
the
like,
where
you
give
it
a
list
of
Creed's
versus
checking
all
the
grades.
It's
done.
C
A
A
A
So
yeah,
so
that's
a
little
more
data.
That
would
be
useful,
but
it
seems
like
the
impact
is
a
small
number
of
widely
used
crates
more
than
a
large
number
of
crates
period,
which
is
always
good,
but
not
as
good
this
small
number
of
grades.
So
I,
don't
know
if
that
makes
a
difference
to
us.
I,
don't
really
know
what
the
alt
term,
okay,
Yoda
I,
don't
presently
have
an
idea.
That
is
not
what
the
one
that
I
presented
so
far.
There
might
be
some
variations
that
make
it
a
little
more.
E
C
A
B
B
A
A
E
B
B
Okay,
exports,
meaning
sorry
imports,
meaning
those
in
next
earned
lakhs
right,
and
so
so.
This
caused
some
problems
in
serve
and
in
other
places
and
people
contained
a
bit
and
the
new
winds
and
reverted
that
PR,
because
as
Lindbergh's
right
now,
if
you
have,
if
you
have
a
pointer
to
to
type
which
is
science,
then
that's
one
fine,
but
the
link
complains
about
anyways
and
as
all
and
has
always
done
so,
but
now
it
triggers
for
more
now,
no
more
cases.
It
unwise.
E
A
A
Times
how
I
understood
our
course
comment
was
the
fog
'''l
sort
of
division
between
the
lint
currently
has
there's
like
out
now
you
be,
and
then
there's
squid
guns
and
that's
true
in
both
cases
and
I
guess.
My
point
is
that
the
set
of
things
that
are
the
like
line
of
how
likely
something
is
to
be
a
foot
gun
is
different,
also
separately,
like
it's
more
likely
to
be
a
problem
on
a
see,
function,
you're
calling,
although
it's
not
it's
still,
mine,
guarantee
I.
F
A
D
A
D
A
B
F
A
A
Do
you
think
you
can
summarize,
but
the
problem
is
that
they're
trying
to
solve
what
the
mechanism
is.
Both
the
problem
is
they're
trying
to
solve
serializing
and
deserializing
trade
objects
and
I
kind
of
forget
what
this
mechanism
ever
imposing
is,
but
I
think
they're,
basically
allowing
yeah,
okay,
I
think
what
they're
trying
to
do
is
there's
some
linker
mechanisms
that
let
you
have
different
crates
each
contribute
pieces
of
data
that
get
collected
at
the
end,
and
then
you
can
sort
of
iterate
over
them.
He
totally
has
a
crate
that
lets
you
do
this.
A
D
A
The
idea
is
they're
going
to
try
to
use
this
mechanism
to
collect
information
trade
objects
that
get
used
by
different
crates,
which
they
would
then
potentially
expose
to
enable
deserialization,
I
think,
obviously,
the
trickiest
part.
The
interesting
question
here
is
how
they
can
expose
that
weight.
B
A
G
B
F
Boarding
on
their
mind,
so
this
is
a
little
stale
at
the
moment.
This
was
originally
proposed
in
the
midst
of
the
arguments
about
amplify
unwind,
and
this
was
effectively
hey.
Let's
go
back
to
aborting,
but
only
on
state
functions,
so
this
was
kind
of
a
proposal
in
the
midst
of
the
controversy.
I
think
the
right
answer
now
is
to
just
close
it
as
postponed
and
say
a
waiting
decision
from
the
epified
working
group.
F
The
point
to
merge
a
thing
now
I
would
sooner
merge,
Ralph's
thing
that
drops
the
know
unwind
attribute
from
LLVM
in
the
short-term.
We
don't
seem
able
to
do
that
either,
but
this
one
the
main
controversy
is.
It
means
that
unsafe
would
suddenly
start
having
a
semantic
effect
on
the
program
and
I'm
really
uncomfortable
with
that
idea.
Well,.
A
A
A
F
B
B
E
Feel,
like
my
attitude,
is
sort
of
the
same
as
I
think
if
the
cement
at
this
point,
the
semantics
that's
being
suggested
here-
is
the
semantics.
We
would
obviously
want
the
only
argument
that
I've
seen
that
I
can
recall
for
not
merging.
This
is
just
the
sort
of
weird,
oh
we're
supporting
this
case,
not
the
other,
but
it's
not
like
I,
don't
think
we're
gonna
discover
at
different
semantics
for
the
out
of
line
module
case.
I'm.
E
Sorry,
I'm,
not
gonna,
discover
a
new
semantics
for
the
in
my
module
case
through
the
exploration
of
the
out
of
line
module
case.
So
the
only
question
is
just
this.
You
know
level
surprise
people
of
oh,
this
is
only
supported
for
inline
modules
and
maybe
they
are
gonna.
You
know.
Is
that
surprising
and
then
I
don't
know
I
just
don't
see
I'm,
not
a
certain
petrichenko
camp
of
like
I.
Don't
understand
why
we're
doing
stabilizing
this,
except
that
it's
just
a
little
weird,
only
stabilize
this
case.
F
Time
we
talked
about
this
I
supplied
a
sentence
that
might
be
usable
as
documentation
to
explain
why
it
is.
There's
only
applies
to
my
modules:
module-
that's
sitting
here
at
the
last
line
on
the
bullet
points.
So
if
that's
useful
in
writing
up
a
summary
comments,
then
please
feel
free
to
use
it
I
think
at
the
moment
it's
not
relevant
because
we
actually
will
just
error.
A
A
E
Only
other
thing
I'll
mention
is
that
I
can
imagine.
Maybe
this
is
sleeping
stuff
now,
I
think
I
think
it's
the
same
thing
last
week
that
I
can't
escape
it
or
chunk
up
saying.
Are
we
gonna
bait
it
back
for
this
or
not
like
stabilizing
this,
which
I
assume
just
no
we're
just
not
only
okay?
Now
we
are.
B
A
B
B
B
They're
using
a
fuel
pump
from
I
guess,
the
question
is
whether
we
should
stop
doing
that
for
fun.
Specific
yeah.
F
F
G
A
Okay,
you
can
go
below
I.
Think
what
you're
saying
Scott
is
that,
like
you,
are
correct,
you're,
not
forced
to
use
the
same
name,
so
it
depends
how
much
we
care
about
local
variables.
Having
sneaked
case
names
your
month
or
whatever
and
I
feel
like
I
care.
Very
little
I
know
that
there
are
definitely
times
when
I
choose
to
deviate
from
the
rest.
F
F
Think
it's
not
zero,
actually
I
commonly
see
crates
that
disabled
the
snake
case
requirements
at
the
top
of
a
file
that
was
generated
by
bhaijan
or
otherwise
written
to
match
some
external
API,
but
everywhere
else
in
the
crate.
They
want
this
enforced.
So
I
think
it's
perfectly
reasonable
to
expect
that
it's
turned
on
at
the
code
and
turned
off
at
the
structure.
Definition
I.
A
E
A
A
C
B
A
F
F
F
B
I
think
this
one's
gone
lawyer
and
I
think
this
will
continue
to
go
nowhere.
It's
like
too
many
links
to
go
to
and
I
think
we
should
go
through
them,
one
by
one.
That
means
votes
made
the
Commons
about
guidelines,
there's
a
lot
yeah,
okay,
I
think
just
about
guidelines
make
sense,
but
I
would
suggest
that
we've
sort
of
develop
the
dresses
like
tracks
and
some
guidelines
by
doing
by
going
through
them
like
developing
guidelines,
it
seems
like
an
ideal.
A
F
There's
one
thing:
I
want
to
mention
that
I
think
is
related.
That
came
up
in
the
cargo
meeting
yesterday,
which
I
think
is
worth
discussing
very
briefly
here,
because
we
do
commonly,
as
a
resolution
two
things
say:
hey
that
ought
to
be
a
clippie
lint,
not
a
regular
lint,
and
so
we
care
a
little
about
the
status
of
Clippy,
because
we
kick
things
over
to
cookie,
sometimes
yeah.
F
There
was
a
general
feeling
and
consensus
in
the
cargo
team
team
that,
based
on
when
their
full
request
that
was
trying
to
fix
Clippy
issues
and
add
them
to
CI.
There
was
a
feeling
that
the
defaults
on
Clippy
are
much
less
curated
than
the
different
mints
on
rusty
and
much
less
focused
on
avoiding
positives.
There
was
a
feeling
of
if
somebody
would
sit
down
and
curate
a
list
to
remove
some
of
the
things
that
people
very
commonly
disabled
and
the
remainder
would
be
a
lot
more
useful.
A
F
From
this
issue,
I'm
just
bringing
this
up
because
it
came
up
and
Clippy
is
relevant
to
the
Lange
team,
because
we
kicked
things
over
too
quickly.
Sometimes
I
thought
it
was
worth
bringing
up
that
we
might
want
to
join
in.
With
that
conversation
with
the
dev
tools,
team
working
on
Clippy
and
worst
case,
if
it
doesn't
get
solved
by
then
I
think
we
ought
to
have
this
happen
at
the
audience.
I
want
to
float
on.
I
want
to
argue
the.
A
Opposite
of
what
I
just
argued
a
little
bit,
actually
speaking
of
that
I
kind
of
I
think
there
could
be
a
deafening
a
case
for
let's
let
the
Clippy
team
on
Lintz.
Let's
work
with
them
to
develop
like
we
agree
on
the
sense
of
what
belongs
in
which
place
but
look
like.
Do
we
really
need
to
micromanage
which
lint
you
know
we
can
give
our
feedback
if
something
goes
wrong
and.
A
E
A
F
Said
there
was
a
pretty
detailed
review
of
here's.
What
should
be
uplifted?
Here's
what
shouldn't
somebody
needs
to
actually
curate
it,
but
that
curation
aside,
yet
there
was
a
pretty
extensive
review
on
5:30
2
to
4,
and
there
was
a
list
of
here's.
What
wouldn't
have
false
positives
and
is
finding
real
bugs
right.
They
go
over
the
list.
A
A
There's
there's
some
pretty
good
categories
here,
and
maybe
we
want
I
forget
already
like,
for
example,
like
looking
over
this
list.
I
forget
the
argument
for
why
clone
on
an
ampersand
empty
as
a
correctness
but
I
think
it
comes
down
to
you.
Don't
expect
the
type
that
you
get
out
from
there,
but
and
I
think
there
were
some
examples
where
these
two
bugs,
but
you
could
argue
it
but
I'm,
just
kind
of
saying
I
guess
you're
gonna
message,
but
I
think
it
would
be
good
for
us
to
sort
of
let
them
go
and
there's
something.
A
E
A
G
D
A
A
B
I
general
points
about
mins,
I,
think
they're,
quite
different
from
Diagnostics
in
that
think
they
sort
of.
If,
if
you
had
a
bunch
of
mints,
they
they
they
have
the
power
to
change
a
lot
about
valves
languages
is
written,
so
I
think
it
made
sense
for
the
members
team
to
like
keep
dealing
with
things.
I
do.
A
Want
to
yeah
and
that
same
language
team
has
nothing
to
do
with
this
I'm.
Just
saying
I
don't
want
to
have
like
a
whole
lot
of
process
when
there's,
oh
I,
don't
know
if
we
need
a
whole
lot
of
process
when
we
have
with
experience
there's
buggy
patterns.
Most
of
these
things
are
not
related
to
like,
so
so
what
I
would
suggest
is.
B
G
On
this
is
I
read
the
RFC
I
like
there's
a
lot
of
stuff
in
here
and
I'm,
not
sure
I,
know
Sara
Lee
agree
with
some
of
the
positions
in
there,
but
I
do
agree
with
a
lot
of
the
a
lot
of
the
actual
in
like
to
the
previous
phrasing
of
like.
Maybe
we
don't
actually
need
an
RFC
for
this
or
whatever
or
one
super-mega
issues
the
wrong
way
to
go
about
it
or
something,
and
if
people
just
make
be
ours
for
things
and.
A
A
G
F
G
The
other
thing,
I
think,
would
be
really
interesting
and
potentially,
if
we
wanted
there's
a
bunch
of
those
that
are
like
unit
makes
no
sense
here
or
if
you
pass
zero,
as
this
particular
parameter
will
always
throw
so.
We
wrote
a
lint
about
that.
I
think
some
sort
of
extensible
diagnostic,
something
or
others.
G
B
F
A
G
B
E
A
Not
really
enough,
but
we
get
started
so
I
sent
out
this
mail
about
whether
to
have
a
second
meeting,
especially
to
discuss
their
roadmap,
but
also
other
things,
but
I
wanted
to
give
a
little
context,
especially
because
nobody
answered
the
doodle
and
I'm
not
sure.
Maybe
it's
a
bad
idea.
I
want
to
get
some
feedback.
People
have
people
think
so.
I
I
wrote
down
a
few
of
the
concerns
that
are,
on
my
mind
that
I'd
like
to
sort
of
address
in
cases.
A
A
Like
it
to
be
very
clear
at
any
given
time,
what
are
the
things
both
to
us
and
to
the
community,
what
are
the
things
that
we're
like
actively
working
on
and
to
try
to
have
it
so
that
RFC's
and
discussion
is
really
concentrated
on
those
things
also,
so
that
we
know
what
to
pay
attention
to
I?
Think
that's
not
the
case
right
now,
right.
We
have
a
lot
of
open
pr's
in
the
arts,
or
I
see
we
put.
Most
of
them
are
like
not
a
high
priority
from.
A
It's
hard
to
figure
out
which
ones
we
should
care
about
so
in
a
related
way,
I
think
we
don't
have
a
good
channel
for
lower
priority
things
to
kind
of
make
progress
you
wind
up.
Instead
of
using
RFC's,
we
wind
up
people
of
nominating
issues
and
PRS
sort
of
sidestep
the
process.
It
works
okay,
but
it
doesn't
feel
ideal.
It
feels
right
now
like
if
it's
small
enough,
you
can
squeak
through
that
way.
A
I
would
like
to
have
not
everybody
can
make
this
meeting
on
a
regular
basis,
notably
Ralph,
but
I'm
sure
this
must
apply
to
lots
of
people
and
I
would
like
sometimes
to
have
longer
conversations
period
and
also
time
for
them
to
make
it
that
requires
in
scheduling.
So
what
I
was
thinking
about
was
taking
a
bit
of
a
card
from
the
way
the
copilot
team's
been
operating
and
having
two
meetings
per
week.
It's
worth
noting
that
we
currently
spend
at
least
some
of
us
90
minutes
doing
the
pre
triage
in
the
tree
guys.
A
The
idea
would
be
to
sort
of
not
do
that
reconfigure
the
time
so
that
we
a
detailed
discussion,
meaning
that
we
stood
on
one
track
and
then
like
a
shorter
triage
meeting,
where
we're
not
spending
as
much
time
on
things.
I
was
even
thinking
of
having
a
timer
literally
keep
us
on
track
and,
if
something's
taking
too
long,
we
shut
it
off
to
the
other
meeting
or
to
a
site.
A
However,
and
the
idea
would
be
that
we
discuss,
we
mostly
discuss
things
from
our
shepherded
list
or
issues
that
have
arisen
that
are
complex,
but
we
have
a
sort
of
while
we're
scheduling
like
every
ends.
Meeting
we
try
to
keep
as
a
wild
card
to
handle
some
lower
priority
thing
that
maybe
we
do
want
to
make
some
progress
on
so
that
there's
like
a
side
channel
for
new
stuff
to
come
in
and
I
wanted
to
hear
what
people
think.
A
E
A
A
E
Yeah
I
said
this
in
the
Zula
channel:
I'm
not
prepared
to
commit
to
a
time
I
feel
like
I
like
to
mix
my
time
now.
It's
just
going
to
be
in
a
month,
I'm
gonna
say
oh
that
time,
maybe
that
tells
to
work,
but
why
not
cuz
I'm
under
six
hours,
a
six
hour
shift
in
time
zone?
It's
a
good
plan
wait
a
month
soon.
Maybe
we
should.
F
A
B
A
All
right:
well,
let
me
discuss
the
timing.
I
would
be
okay,
delaying
the
start
for
a
little
bit
anyway,
because
it'll
probably
take
us
some
time
to
get
organized
I
think
we
could
use
the
time
to
sort
of
write
up
and
discuss
a
little
more,
especially
if
we're
gonna
couple.
This
I
wanted
to
make
an
RFC
about
the
migrating
of
RFC's
and
liaison
policy,
because
I
think
that's
a
big
enough
thing
that
I
would
like
people
to
see
about
it
and
hear
about
it
and.