►
From YouTube: 2020-10-22 RFC 2229
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
B
Yeah
that
would
kind
of
solve
all
the
places
in
one
place.
Maybe
it
might
lead
to
other
problems
down
the
line,
which
is
the
kind
of
annoying
thing
another.
A
Another
thing:
well,
I
guess,
there's
two
questions
right.
One
is
that
we
had
a
delay
span
bug.
Why
did
that
not
trigger
anything?
Did
we
crash
before
that
and
if
we
crash
before,
then
that
would
answer
my
question.
Second
would
be
that,
yes,
I
guess
the
possible
three
values
for
it
from
the
looks
of
it
are
going
to
be
error,
infer
and
double
itself,
but
the
method
subs
where
we
do
this
probably
cannot,
I
guess,
does
not
return,
does
not
have
an
inference
contact,
so
we
would
have
to
go
and
basic.
B
Sorry
why?
But
we
have
to
add
an
inference
context
so.
A
Already
I'm
saying
like
just
to
handle
a
similar
case
in
other
places
like
today.
It
happened
here
is
possible
since
we
made
the
scene
in
multiple
other
places
we
might
see
like,
like
the
the
error
happened
here
like
this
is
where
we
had
a
ty
kind
error,
but
it's
possible
that
at
some
point
t
by
kind
as
t
y
kind
error
in
other
parts
of
the
code.
We
just
don't
have
an
example
for
it
and,
and
that
might
and
those
might
be
places
we
might
need
to
resolve.
A
Code
yeah
in
the
original
code.
We
don't
need
to
in
this
one,
but
we
we
do
similar
thing
in
multiple
other
places
like
where
we
actually
have
to
resolve.
So
in
that
case,
we
would
probably
look
for
a
tuple
and
instead
of
an
inference
variable
as
kind
of
what
I'm
going
with
so
let's
let
him
pull
it
up.
I
guess
that
might
be
here
hold.
B
A
B
A
Oh
I'm
not
saying
that
I'm
not
against
the
other
type.
I
am
more
saying
that
you
probably
should
do
something
similar
around
when
we
use
resolve
as
well.
B
B
The
only
thing
I
don't
know
is
whether
that
behavior
should
be
just
return,
an
empty
list,
which
probably
is
the
right
thing
to
do
most
of
the
time
or
or
to
return
a
result
or
something
which
would
be
more
annoying
but
would
allow
us
to
like
you
can
kind
of
call
or
default
if
you
want
the
emptiness
behavior.
But
if
you
you
know,
maybe
there's
some
logic
where
it
expects
to
find
something
in
that
list,
and
it
would
ice
if
it
doesn't
find
it.
A
I
guess
you
would
have
to
try
it
out
my
what
I
was
going
for
was:
how
does
how
do
we
handle
the
error
case
in
rest
of
the
parts
of
the
compiler
like
what,
if
we
get
a
ty
kind
error.
B
B
Actually,
another
option
would
be
to
return
a
single
entry
list
with
just
error
in
it.
But
anyway
I
don't
know
if
that's
actually
better
the
like
the
main
thing
that
you
do
want
to
try
to
avoid.
If
you
can
is
you
know,
either
isis,
of
course,
or
like
further
errors
being
recorded
as
a
result
of
getting
error
so
like.
If
there's
a
type
mismatch,
you
don't
want
to
necessarily
report
that,
because
that
seems
weird
that
usually
just
means
downstream.
B
Stuff
that
the
author,
the
user,
has
to
ignore
because
it
doesn't
apply.
I
guess
what
I
would
probably
say
is.
B
B
My
guess
is,
most
of
the
time
it's
to
be
an
empty
list,
because
you
just
don't
want
to
do
the
work,
that's
being
done,
but
it
might
sometimes
not
be
or
make
it
return
an
empty
list
and
still
do
a
just.
Do
a
grep
either
way
it's
kind
of
like
we
gotta,
go,
inspect
the
callers
and
see
what
yeah.
A
Right,
that's
fair.
One
thing
wait.
I
had
give
me
one
second
around
this
was
so
this
was
caused
because
of
a
delays.
Okay,
actually,
I'm
I
I
lost
my
friend,
I
thought
it's!
It's
fine
yeah,
okay,
yeah!
That
makes
sense.
We
can
one
question
that
we
were
trying
when
we
were
running
this
was
what
is
the
ui
dependencies
set
of
tests
we
supposed
to
do
or
if
you
are
full,
that's
that's
as
a
sibling
directory
to
ui
depths.
A
So
there's
gui
dash
full
depth,
which
is
kind
of
a
test
suite
that
sits
next
to
the
ui
test
suite
I
have
never
seen
it
being
run
and
according
to
roxanne,
might
be
failing
on
master
as
well.
B
I
don't
know
what
that
is.
Okay,
oh,
I
think
those
are
tests
that
require
the
like,
require
you
to
have
built
everything
from
the
stage
a
given
stage
like
they
usually
would
be
relying
on
the
compiler
libraries
for
example.
So
I
don't
run
those
you
can't
just
build
standard.
You
have
to
build
all
this
stuff.
A
B
A
Okay,
I
guess
that
makes
sense.
Also
this
pr
had
a
three
percent
regression
of
the
tuple
one
had
a
three
percent
regression
and
wg
gamma.
I
am
assuming
this
is
because
of
the
added
obligation
and
like
the
branch
branches
that
we
do,
but
I'm
not
sure
what
all
is
part
of
the
what's
it
called
what
all
gets
accounted
for.
Wait:
okay,
that's
yeah!
So
when
it
says
full
incremental
and
incremental
change
in,
like
the
comparison
and
on
the
proof
test,
is
that
just
the
number
of
instructions
generated.
A
B
And
full
versus
incremental
fall
and
all
that
stuff
full
means
it
just
did
a
regular,
build
incremental
full
means
you
did
a
regular
build,
but
in
incremental
mode,
which
adds
a
little
overhead,
okay
and
incremental
unchanged
means.
You
did
a
second
run
after
the
first
incremental
build
where
nothing
changed,
which
in
principle
should
be
much
faster,
and
so
those
are
the
different
modes
and
then
there's
sometimes
some
more
incremental
like
stuff
like
that.
So
yeah,
I
don't
know
why
it's
a
2.7
regression
but
like
simulacrum,
I
am
not
overly
concerned
about
that.
B
A
B
B
B
Random
reasons
that
are
hard
to
pin
down-
I
don't
know
if
that's
a
little
bit
of
a
poor
answer,
but
I'm
not
sure
why
I'm
looking
at
wg
grammar
in
the
wall
time
not
sure
why
it
looks
like
it
also
regressed
wall
time
or
at
least
those
are
slower,
the
wall,
the
wall
time
measurements
in
general
are
all
over
the
map.
So
I
don't
know
what
to
take
from
that.
But
you
know
if
I
had
to
guess
why
it's
doing
a
little
more
work,
it
would
be.
B
B
This
parser
generator
framework,
actually,
which
may
well
use
a
lot
of
closures
and
generate
some
distribution.
A
All
right,
another
thing
I
think
we
should
do
is
update
our
tracking
issues
so
as
part
of
the
adding
a
feature
gate
we
kind
of
list.
This
is
the
tracking
issue.
Go
check
progress.
A
I
don't
think
that,
like
the
main
thing
on
the
tracking
issue
has
been
updated
in
a
while,
so
we
should,
I
think,
just
point
it
to
either
the
project
board
and
kind
of
leave
a
summary
that
saying
we
are
about
done
with
the
capture
analysis,
part
and
then
yeah,
because
right
now
I
think
this
is
last
update
on
may
or
other
than
when
I.
B
Yeah
we
can,
I
can
update
the
tracking
issue,
went
over
at
the
at
the
repository.
A
A
B
A
So
just
share
this
on
the
rust
chat.
It's
to
do
so.
I
don't
know
if
you
want
to
share
this
right
away,
but
I
don't
know
if
it
hurts
either
it's
on
the
zoom.
B
C
A
B
I
don't
know
if
you
I
mean
this
is
a
great
doc.
I
wouldn't
invest.
You
know
we
have
the
issues
and
other
things
so.
A
A
No,
I
don't
particularly
some
of
it
might
be
required
for
stabilization.
I
think
when
we
put
the
peers
that
were
involved,
but
at
least
that's
one
of
the
based
on
one
of
the
things
that
looked
at,
but.
C
A
Think
we
can
track
it
from
the
project.
What
is
also
that's
fine,
this
I
also
kind
of
wrote
when
we
did
a
sink
back
in
early
september,
just
to
get
everyone
back.
B
A
Yeah,
so
all
right,
this
is
done.
Okay,
let
me
I
can
just
point
you
to
the
pr
is
also
minimum
capture
is
almost
cleaned
up.
The
two
things
that
are
kind
of
there
left
are
sort
of
the
logging
part
and
the
pattern
issue
that
we
discussed
both
are
in
works
right
now.
So
let
me
just
okay.
A
So
again,
I
just
sent
this
on
zoom
chat.
This
kind
of
is
a
pr
and
works
that
shows
how
the
logging
things
are
being
done,
how
the
more
like
the
what
the
logging
now
looks
like
it
still
needs
some
cleanup
and
the
minimum
capture
part.
A
But
basically,
I
think
this
is
kind
of
what
we
had
talked.
A
B
A
Yeah,
so
it's
starting
to
start
to
get
there
and
yeah.
I
think
wait.
Did
I
list
anything
else
hold
on
yeah,
okay,
so
one
one
more
thing
would
be
so
couple.
One
thing
that
we
also
do
part
of
fourier
that
we
haven't
kind
of
updated
to
place
is
to
use
closure
or
we
use
closure
origin
to
point
to
things
where
that
the
closure
is
an
f
and
one
square
and
f
mute
closure,
and
why
did
we
ended
up
picking
that
particular
trade
for
that
closure?
A
It
currently
uses?
I
think
it's
a
variable,
just
a
variable
name
or
it
might
just
be
an
item
and
what
is
it
called
and
like
just
the
span
of
it,
I'm
gonna
pull
up.
Do
we
want
to
move
that
part
to
you?
Be
that
part
to
be
you
to
use
places
or
or
do
we
want
to
keep
it
as
is
for
a
bit
and
when
we
start
moving
start
using
places
post
type
check,
then
we
move
that
over.
B
It
seems
relatively
low
priority.
I
guess
it
would
be
better
for
it
to
use
a
place
okay,
but
we
can
print
a
nicer,
a
more
precise
message
like
and
not
misleading.
A
That
that
again
is
a
little
concerned,
because
I
I
think
we
discussed
this
a
while
back
that
you
might
need
to
start
keeping
track
of
two
expressions
primarily
because
there's
an
expression
that
resulted
in.
So
we
have
two
things
going
on.
We
have
the
path
length
and
we
have
the
capture
kind,
so
yeah.
A
Yeah,
so
it's
possible
that
we
say:
oh,
we
might
need
to
like
need
better
investigation
on
how
we
do
edit
reporting
around
that.
So
that
was
also
one
of
the
concerns
like
maybe
this
before
I
complement
this,
maybe
we
should
just.
B
I
would
defer
that
if
it's,
it
seems
like
a
reasonable
deferable
item
yeah
and
for
now
you
can
just
generate
you
know
some
variable
like
truncate
to
the
variable
name
and
pick
pick
a
span
and
it'll
be
good
enough.
Yeah.
A
A
A
Yeah,
I
I
I
think
we
can
leave
it
for
later
as
well,
because
this
won't
be
too
big
a
change
to
update.
I
think.
B
Right,
if
I
remember
it's
a
pretty
it's
a
relatively
isolated
structure,
that's
only
used
yeah.
A
Yes,
okay
and
let
me
just
confirm
this
is
sort
of
the
structure
we
were
going
for
for
our
changed
to
expert
users
or
delegate.
So
let
me
just
where's
the
second
I'm
finding.
A
Oh
wait.
I
think
one
of
the
files-
oh
no,
I
think
I
found
it
yeah
so
copy
the
link,
that's
that
yeah.
So
this
is
kind
of
like
now
we
have
a
dedicated
place
and
an
expression
id
that's
used
for
diagnostics,.
C
A
That
that
looks
fine,
so
basically.
B
A
A
It's
still
used
within
memory,
sorry
memory
categorization
to
pass
around
information,
which
I
think
is
fine,
because
otherwise
there's
too
much
decoupling
that
not
for
memory
categorization
is
required.
I
mean
that's.
A
Give
me
one
second
I'll
had
a
question
about
something
in
clippy,
I'm
trying
to
find
that.
Let
me
just
quickly
bring
him
again.
He
it's
basically
around
what
some
certain
variable
names
mean,
and
it
wasn't
entirely
evident
from
the
name.
A
A
Found
it
so
there's
this
lint
escape
lint
and
clippy
and
there's
this
block,
which
is
okay.
Let
me
just
share
screen,
I
guess
or
oh
I
wanted
to
do
that
I'll.
Just
send
you
the
file
language,
if
that's
easier,
so
there's
this
blocker
on
line
140,
where
the
place
with
her
id
is
called
cmt.
Is
that
I'm
not
sure
what
that's
supposed
to
be.
A
Mostly
curious
has
to
like
having
an
idea
what
we
are
editing
around
and
if
some
of
these
things,
don't
just
just
making
sure
that
still
makes
sense
to
separate
out
in
all
places
based
on.
A
It
does,
if
you
look
at
line
127.
B
A
A
I
guess
we
can
turn
that
into
that.
We
keep
the
place
with
variety,
but.
B
A
I
I
guess
my
point
behind
that
one
was
it's
possible
that
I
I
actually
don't
know
how
many
times
we
keep
a
track
of
expression
id
for
not
diagnostics.
So
if
we
do
it
for
genuine
purposes
it
like,
and
then
it
kind
of
trickles
down
into
diagnostics,
then
it
might
be
an
interesting
question
like.
A
B
A
Okay,
so
I
think
yeah,
that's
all
I
have
for
this.
I
yeah
I'll
discuss
the
ice
with
roxanne
and
make
sure
that
we
go.
A
We
we
go
with
the
error
solution
and
see.
Hopefully
this
doesn't
break
things
everywhere
and
like
we
don't
need
to,
we
need.
We
don't
need
particular
information,
also,
actually,
if
it's
an
error
kind
and
we
need
a
particular
value
within
that
tuple,
it
will
just
not
work
anyway,
okay,
but
we
need
to
make
okay,
but
we
need
to
make
sure
that
it
doesn't
just
completely
throw
extra
outer
so
that.
B
C
A
Let's
see
so
that's
not
too
many
honestly
there's
one
in
pretty
printing
in
which
I
guess
we
can.
If
we
get
an
error
there,
we
can
just
print
thai
is
not
available.
So
that's
not
hard.
A
There
is
some
stuff
in
the
borough.
Checker.
A
A
Okay,
that's
good
to
know
hold
on
I'll.
Just
add
that
I
don't
know.
C
A
Yeah,
I
think
all
places
we
do
is
just.
B
Some
of
them
yeah,
some
of
them
like
the
mirror,
yeah
mirror
shim,
probably
probably
can't
actually
get
there
if
there
have
been
errors
reported,
but
I
would
prefer
not
to
hard
code
that.
A
Yeah,
that's
that's
fair!
I'm
assuming
same
with
code
gen
that
we
would
have
probably
failed
like.
A
I
think
most
of
our
issues
would
be
like
traits
and
type
check
and
type
check
is
the
one
creating
this.
So
that
wouldn't
be
too
bad,
so
really
trades,
because
that
one's
both
before
and
after
type
check.
A
A
Okay,
I'll
look
into
the
other
reported.
Is
it
part
of
russia
or
the
part
of
sorry
or
is
it
part
of
the
rus
standard.
A
Okay,
it's
part
of
rusty,
okay,.
A
Okay,
yeah,
so
I
think
this
is
all
I
have
do
you
want
to
discuss
anything
else
or
yeah?
What
20-ish
minutes.
A
C
A
So
hopefully,
if
everything
goes
well,
there
would
be
at
least
a
pr
for
number
20
by
monday
and
once
that
gets
merged
in
I'll
rebase,
and
then
the
other
ones
that
are
open
should
be
open
by
next
week.
Ideally
week
after
we
have
manual
capture
in
the
compiler.
A
It's
ready
the
like
right
now.
It
has
a
hack
in
place
to
deal
with
the
pattern
where
it
just
changes
the
her
id
and
then
within
the
place
of
the
variety
structure,
an
expert
user,
but
other
than
that,
like
I've
commented
on
it
like
logan,
like
all
of
us
kind
of,
went
and
just
sat
together
and
kind
of,
did
a
bunch
of
added
a
bunch
of
comments.
A
B
Okay,
either
way
this
looks
great
yeah
yeah,
you
have
to
fix
the
ice,
so
I
guess
that
should
take
some.