►
From YouTube: incr-comp meeting #6 (2020-11-03)
Description
B
A
little
hectic
over
here-
that's
a
little
late
here.
Let's
see
if
I
can
adjust
these
things
be
a
little
bit
more
nope.
Okay,.
A
A
Wesley
I
did
yeah,
we
can
talk
about
the
meeting
itself.
I
guess
we
I
did
reach
out
to
to
michael
worster
and
he
said
he
would
like
to
help
out
that
his
availability
might
be
pretty
limited.
So
I'm
not
sure
if
I
even
told
him
when
the
meeting
time
was
now
they
think
about
it.
I
thought
I
tried
to
but
I'll
reach
out
to
him
again
and
at
least
let
him
know
if
he
wants
to
join
it's
it's
friendly
to
europeans.
So
let's
look
over
the
agenda
quickly.
A
A
B
B
C
I
just
copied
the
issue
triad
section
from
last
week
so
and
updated
the
numbers,
so
those
first
three
high
issues
we
already
talked
about
last
week,
but
I
figured
I'd,
leave
them
just
for
reference.
So,
okay,
okay,
if
we
want
to
remove
them,
that's
fine.
A
Them
at
least
this
week
yeah,
I
figured
I'd,
leave
them
there.
Just.
A
I
gotta
figure
out
some
way
to
make
my
setup
here
more
comfortable
for
I've
got
like
too
many
computers
on
my
desk
right
now,
and
so
it's
and
I'm
trying
to
keep
certain
things.
What's
the
word
siloed?
Okay,
hopefully
this
is
gonna,
be
okay,
nope.
The
one
thing
I
wanted
to
do
is
up
there.
It
goes
okay,
I
want.
I
want
my
keyboard
and
mouse
to
work
and
I
was
like
they
stopped
working.
Okay,
so
yeah,
let's,
let's
go
ahead
and
get
started
a
little
more
officially.
A
A
C
Okay,
so
we
so
I
approved,
I
think,
a
week
and
a
half
ago
and
wrote
a
summary
of
the
perf
results
for
the
performance
review
team
and
then
it
failed
consistently
in
boards
on
the
I-686,
no
optimization
builder,
with
a
fairly
strange
error
and
we're
not
really
sure
why
so
building
just
on
I-686
like
as
a
target
works
locally.
It
has
something
to
do
with
that,
the
specific
configuration.
So
I
need
to
set
up
docker
and
try
to
run
the
sorry.
A
Do
do
we
have
no
op
builders
for
like
the
64,
you
know,
for
I
x86
64.
A
Okay,
I
can
take
a
look
real,
quick,
okay,
but
you
and
you
sorry
in
terms
of
the
reproductions,
have
you
said
it
didn't
reproduce
for
targeting
I-686
yeah?
Have
you
reproduced
locally,
just
self-hosting
or
you
because
you
have
no
time
to
to
investigate.
You
haven't
even
had
a
chance
to
like
reproduce
locally,
so
I've.
C
I've
tried
pulling
the
pr
and
running
the
incremental
tests
on
x86
64
and
targeting,
and
they
both
work
right.
I
haven't
had
time
to
set
up
docker
and
actually
wanting
that
particular
builder.
So,
okay.
A
Okay,
that's
you
know
we'll
just
keep
on
keeping
on
keeping
on
yeah
with
it.
It's
and
I
see
that
there's
been
some
interaction
with
ollie,
but
that's
just
because
of
basic
like
there
was
some.
I
guess
some
ideas
being
put
back.
A
Okay,
all
right
next
up,
78702,
add
cgu
estimated
size
info
to
self
profile
data.
So.
C
I
just
opened
this
this
morning,
there's
a
little
screenshot,
if
you
click
through
to
the
pr,
but
basically
for
the
cogen
module
activity
now,
there's
both
the
name
of
the
module,
as
well
as
the
estimated
size
cool
which
is
fairly
nice
to
try
to
get
that
info
in
there
and
just
clicking
through
the
results.
I
spotted
some
stuff
that
looked
a
little
fishy
like
that.
A
Do
we
also
get
the
amount
of
actual
like
lvm
bit
code
that
was
generated
for
the
module,
or
is
it
something
that
doesn't
make?
Is
this
pre-monomorphization
or
something
that
doesn't
make
sense.
C
A
Okay,
I'm
trying
to
get
like
other
other
things
to
measure,
but
you
know
obviously
at
this
point:
if
we
had
things
other
things
to
measure,
we'd
probably
incorporate
them
into
the
cgu
estimate.
Possibly
okay,
then
well
clock
time.
It
is.
I
guess
that,
in
terms
of
you
know,
it's
the
sanity
check,
that's
being
employed
here
and
that
is
you're
already
getting
interesting
stuff
out
of
it.
So
that's
good!
A
Okay!
Does
this?
Need
it
just
needs
to
be
reviewed.
I
guess
is
the
point
yep
it
just
yeah.
I
literally
just
posted
this
morning,
so
yep
great
great
okay.
Next
up
I
put
on
pr
long-standing
pr
74967.
I
finally
had
a
chance
to
look
over
this.
This
is
the
implement
lazy
decoding
of
death
path
table
during
incredible
incremental
compilation,
and
so
I
I
can't
remember
when
I
looked
at
it,
but
I
definitely
oh,
I
guess
it
was
around
five
days
ago.
A
I
started
looking
at
it
and
I
put
down
some
concerns,
or
at
least
some
questions,
and
I
haven't
heard
the
the
author
is-
was
active
pretty
recently,
so
I'm
expecting
to
hear
back
from
them,
but
a
reason
why
I
wanted
to
bring
up
here
is
because
you
you
look
through
this
as
well
what
as
well
wesley,
so
I
feel
like
as
confident
as
I'm
gonna
be
like.
A
A
A
C
A
I'm
tempted
to
maybe
just
land
it
nonetheless,
like
it's
a
huge
win.
What
I
understand
in
incremental
mode
and
oh,
that
was
a
thing
also.
That
was
what
aaron
said
they
were
going
to
look
into.
Is
that
someone
reported
slight
regressions
in
non-incremental
mode,
but
they
had
those.
It
was
bjorn
iii
who's
reported
like
previously.
When
we
run
benchmarking
on
this
there
it
was
a
clear
win
for
non-air
control
mode.
A
So
the
slight
regression
that
in
a
criminal
mode
is
surprising
and
so
aaron
said
they
were
going
to
look
into
resolving
that,
and
that
was
10
days
ago.
So
I
guess
the
right
thing
to
do
here
is
not
just
land
it
blindly,
because
if
nothing
else
aaron
it
sounds
like
aaron
may
look
into
something
and
we
don't
want
to.
You
know,
introduce
necessary
regressions.
Even
if
they're
tight,
relatively
small,
okay
I'll
go
through,
and
just
you
know,
ping
erin
yeah
and
see.
C
We
might
want
to
consider
in
terms
of
landing
at
is
figuring
out
a
good
time
to
land
it
like.
We
should
probably
make
sure
that
it
if
we're
going
to
land,
if
we're
just
going
to
sign
off
on
it,
we
should
probably
try
to
get
it
in
right
after
the
beta
split
yep,
there's
the
most
amount
of
time
it
can
sit
on
nightly
before
it.
A
Hits
me
yes,
I
I
think
that's
an
excellent
point,
one
that
I've
made
on
other
things
in
my
life
in
the
past,
but
forgot
to
here.
So
that's.
That
is
a
very,
very
good
point.
Oh
we
started.
We
started
writing
the
beta.
No,
no!
I
don't
know.
Okay,
I'm
looking
forward
now
to
find
out
what
the
timing
is
and
the
beta
cut
the
beta.
A
A
Okay,
so
I
think
yeah
we
say
I
think,
that's
a
good
good
I'll
ping.
Oh
I'll,
write
on
the
ticket,
then
that
given
our
relative
level
of
like-
let's
we'll
put
it,
let's
sorry
basically
put
this
stuff.
You
know
just
trying
to
be
safe,
we'll
plan
to
land
this
early
in
the
cycle
after
the
next
beta
cut
and
then
I'll
say
that
they'll
roughly
be
around
november
19th,
so
that
gives
aaron.
You
know
a
solid
a
little
over
two
weeks
to
investigate
the
thing:
okay,
good,
good,
good
and
next
up
is.
A
I
assume
this
is
david
yeah
talking
about
split,
dwarf
77117,
you
and
you
have
nothing
report.
As
you
said,
david.
A
Yep,
I
totally
understand
that,
okay,
so
how
did
we
do
some
issue
triage
before
we
get
into
the
issues
that
are
open
all
right?
I
had
planned
to
look
into
the
queue
that
we
had
built
up
last
week
and
I
didn't
do
anything
I
I
really
just
just
couldn't
in
terms
of
what
I've
got
going
on.
So
I
was
curious,
though,
how
we
did
in
terms
of
like
we
made
a
list
of
ten
things
and
there's
no
longer
ten
things
on
this
agenda.
A
C
Not
quite
I
didn't
have
enough
time
either
to
get
through
the
whole
list,
but
I
did
triage.
I
triaged
four.
I
changed
labels
on
three
items.
There's
a
topic
in
the
zulu
stream
called
issue
triage
where
I
was
originally
just
writing
what
I
planned
to
do
and
I
got
a
thumbs
up
from
somebody.
So
I
just
went
and
did
it
this
morning.
A
Okay,
I
guess
what
I
would
suggest,
then
is
we
I'd
still
like
to
try
to
use
this
approach.
It
didn't
quite
work
out,
but
that's
you
know
we're
all
busy.
So
you
can't
you
know
slap
ourselves
up
or
beat
ourselves
up
too
much
about
it.
I
I
might
just
and
then,
as
we
just
take
the
six
issues
that
you
that
we
that
didn't
get
triaged
and
put
them
and
re-enqueue
them,
but
I'm
trying
to
understand
here
so.
C
A
You
added
the
template,
like
just
added
that
section
in
acute
issues
to
be
triaged,
but
we
don't
have
a
list.
Yes,
okay,
that's
that's
what
I
was
sort
of
trying
to
figure
out.
So
let
me
look
at
the
so
we
can
just
take
the
pass
list
and
copy
over
the
six
issues
and
maybe
add
four
more
to
you
know
round
it
out
and
still
set
our
heights
set
our
sights
on
actually
doing
10
a
week.
A
So
if
I
look
in
the
issue
triage
zoolip
that
you
have
on
zulip,
I
see
that
right.
You
know
what
I'm
gonna
do.
First,
I'm
gonna,
I'm
gonna,
just
re-enqueue
all
10
right
now,.
C
A
Okay,
so
they're
seven
right
now
and
then
seven
five,
nine,
nine,
seven
right,
you're
saying
that
that
there's
a
general
thumbs
up
that
it
probably
shouldn't
be
reprioritized,
which
is
an
interesting
question
because,
like
I
would
say
like
we,
I
don't
know
if
we
at
times
we've
prioritized
feature
requests
in
the
compiler.
But
it's
not
something
we've
done
recently.
A
Wow,
this
is
a
pretty
specific
thing
here.
This
person's
trying
to
do
income
inflation
on
an
android
device.
Am
I
reading
that
right.
C
Incredulous,
there's
there's
actually
some
pretty
decent
terminal
terminal
emulators
and
you
get,
I
don't
say,
full
access,
but
it's
much
less
restricted
than
ios
is
so
okay.
No,
I.
A
To
be
honest,
I
was
it's
more.
My
concern
is
more
about
the
limitations.
The
device
hardware
in
terms
of
like
rusty
is
just
kind
of
you
know
a
little
bit
bloated.
I
don't
know
how
much
memory
they
have
on
their
phone.
I.
A
A
Okay,
where
is
this
cup,
so
is
this
happening?
Is
this
even
right?
Okay,
okay,
because
we
hard
right
right.
I
understand
now
we're
doing
hard
linking
and
really
it's
a
sign
of
something
that
could
happen.
Other
platforms,
I
suppose-
and
they
just
want
to
wait
it
to
they
just
want
some
way
of
flag
or
something
where
they
could
suppress
it.
A
Yeah!
That's,
oh!
No,
oh
no!
That's
not.
They
say
the
warning
is
obsolete,
but
that's
I
don't
know
if
I'd
say
it's
obsolete,
like
the
warning
still
seems
relevant
right.
They
just
can't
they
did
this
because
they
can't
express
this
doesn't
make
you
get
rid
of
the
warning
entirely
right.
They
just
want
to
be
okay,
they
just
want
us,
they
probably
want
to
suppress
an
android
entirely.
I
guess.
A
C
Anyway,
I'm
spending
too
much
time
thinking
about
this,
so
I
so
I
did
start
an
issue
in
our
repo
collecting
bugs
related
to
file
system
limitations
and
things
and
linked
to
it
from
there.
So,
okay,
it
might
be,
I'm
wondering
maybe
if
there's
a
project
here,
I
think
we
talked
about
a
couple
meetings
ago,
where
maybe
we
should
not
be
trying
to
use.
C
I
don't
want
to
say
such
advanced
but
like
at
least
on
some
platforms
like
hard
linking
kind
of
a
niche
thing.
So
maybe
we
should
try
to
use
a
smaller
subset
of
of
file
system
functionality.
That's
that's.
B
C
A
Okay,
all
right:
well,
let's
yeah:
let's
keep
it
as
a
feature
request
or
whatever
or
and
not
reinkew
it,
and
the
only
question
is
whether
to
put
a
priority
label
on
it
or
not.
I
personally,
I
I
don't
think
it
hurts
to
put
a
priority
label
on
it.
Personally,
I'd
call
it
like
p
medium.
A
If
anyone
else
wants
to
object
to
that,
that's
fine,
but
I
mean
do
you
think
it's
I'm
curious.
Do
you
think
it's
misleading
to
the
priority
labels
on
enhancement,
requests.
A
Yeah,
it
might
be
I
the
thing
about
me
with
p
low
is
like
sometimes
in
my
head,
I
think,
of
pilots
being
like
on
track
to
be
closed,
which
isn't
what
we
do
but
like,
given
that
we
have
a
four
tiered
prioritization
scheme.
I
I
really
do
wonder
a
lot
about
the
p
low
versus
p
medium
distinction.
Okay,
let's
leave
it
on
tag
for
now.
I
don't
want
to
like
commit
to
something
and
then
and
then
be
like.
Oh
I
was
that
done.
A
Okay,
let's
add
four
more
issues,
then,
to
the
list
of
things
that
are
of
the
things
to
to
look
at
I'm
trying
to
remember
how
I
even
like
chose,
which
ones
to
add
before,
because
they
sort
of
from
the
middle
of
the
issue
numbers
seems
like.
Oh
no,
it's
because
there's
a
relatively
new
one
was
added
okay,
so
we've
got
a
new
bug
actually
open
12
days
ago,
I'm
looking
at
so
I'm
looking
at
you're
only
posting
the
zoop
stream.
A
So
we've
got
enough
time,
let's,
rather
than
cueing
the
new
bug
I'd
like.
Let
me
just
talk
about
it
directly.
I
think
the
new
bug
is
78236,
which
is
an
ice.
C
Well,
sorry,
so
it's
getting
called
from
the
middle
from
rusty
middle
depth
graph!
It's
it.
A
A
We
should
look
at
this.
I
guess
it's
gonna
be
a
pain
to
figure
out,
but
we
should
at
least
try
to
see
if
we
can
reproduce
it
locally.
I
suppose
I'm
gonna
send
this
to
myself,
even
though
I
know
I've
had
like
such
little
time
to
investigate
things,
I'd
like
to
reach
out
to
simon
and
try
to
get
a
little
more
information
about
this,
so
yeah,
okay,
I'm
gonna,
summon
I
sell
myself
to
this
and
then
in
terms
of
prioritization
of
it.
A
Well,
I'm
gonna
tag
it
working
group
incremental
compilation
as
well,
and
the
only
question
is
what
priority
to
assign
it.
I'm
gonna
go
pee
medium
right
now
I
don't
know
or
oh
I
don't
know
it's
it's.
It's.
C
A
C
B
C
Combination
of
it
being
some
sort
of
an
incremental
loading
issue
and
being
super
easy
to
reproduce
meant
that
it
was
something
we
should
actually
try
to
tackle
soon
right.
It's
right
that
makes
sense.
A
Okay,
all
right,
so
that's
that's
fine
with
me
and
now
now
that
we've
done
that,
let's
look
at
the
the
backlog
now
that
that's
been
added
on,
so
I
yeah
what
I
was
doing
before
was.
I
was
starting
from
the
l
oldest
first,
I
believe,
that's
why
the
numbers
are
relatively
low.
No,
that
can't
be
right
either.
I
don't
remember.
I
really
don't
remember
how
I
chose
this
set
of
issues.
A
I
I
queued
earlier
I've
I
can't
be
oldest
first,
the
oldest
first
ones
are
like
really
old
all
right.
Well,
the
reason
why
I'm
bothered
by
this
is
because
I'd
like
to
like
you
know,
use
the
same
consistent
system
in
enqueuing
new
issues,
but
did
I
sort
by
like
at
least
recently
active
or
something
I
really
at
least
recently
updated?
Let's
look
at
that
nope.
That's
super
old,
too
any.
Does
anyone
have
any
idea
what
I
might
have
used
last
like
two
weeks
ago
to
make
the
list.
A
Yeah,
I
guess,
but
I
still
had
some
sorting
metric,
that
I
was
using
to
end
up
with
like
things
that
were
in
the
seven
five.
Seven
four
range.
I
think
it's
just
the
most
recent
stuff,
but
that's
a
bit
I'm
seeing
things
that
are
like
seven,
six,
seven,
seven
for
the
top
three,
the
top
three
things
not
including
okay,
the
top
two,
the
tough
rings,
three
things
not
including
the
simon
bug.
We
just
looked
at
it's
like
seven,
seven,
one,
eight,
seven,
six,
seven,
twenty
seven,
six,
two
five
one:
seven,
six,
zero!
C
What
you're
saying
I
don't
know.
A
C
Okay,
here's
the
new
search.
Can
you
paste
that
into
the
hacker.
A
Okay,
so
all
right,
so
let's
I'll
just
eq,
you
know
I
really
should
be
sharing
my
screen
or
something
here.
I
guess
no,
I'm
not
going
to
share
my
screen.
It's
not
prepared
to
be
shared,
so
I
gotta
prepare
for
things
like
that.
Okay,
then,
I
will
just
add
in
the
bugs
that
are
relevant
now,
so
there's.
A
A
Seven:
three:
eight
five,
nine
and
one
more
right,
so
seven,
three,
six:
five,
nine,
okay,
so
yeah,
that's
a
new
set
of
ten
issues
and
we'll
see
how
we
do
getting
through
this
week.
Thank
you
wesley
for
the
work
you
put
in
yeah
doing
the
four
like
whenever
it
was
two
days
ago
or
whatever.
C
A
I
guess
doing
the
hackmd
didn't
really
keep
us
on
top
of
it.
So
yeah.
I
think
I
like
yeah,
I
don't
know
the
hacking
didn't
work,
so
I
guess
we
could
just
post
in
the
zula
the
issue
triage
zulu
and
see
if
that
works
better.
I'm
okay
with
that
idea,.
C
I
meant,
in
terms
of
like
figuring
out
what
to
do,
because
I
didn't,
I
didn't
want
to
just
start
applying
labels
and,
oh
other
people,
be
like.
Oh
no.
This
is
this
is
a
pee
high
issue
or
whatever.
So
I
I
posted
what
I
thought
in
there,
which
is
sort
of
what
the
prioritization
working
group
does.
I.
A
See
your
points
I
see
my
plan
had
been
to
actually
tag
when
you,
my
plan
had
been
that
for
each
issue
to
tag
it
with
I
prioritize
when
it
was
enqueued
and
then
go
to
that
dual
stream
that's
created
and
do
work
in
there
and
that
way
the
prioritization
working
group
is
aware
of
what's
going
on
and
can
give
feedback,
but
then
to
go
ahead
and
tag
it
like,
and
you
know
really
make
it
official
and
if
you
get
you
get
feedback,
you'll
get
feedback.
A
A
Let's
try
that
system.
Let's
still
try
to
use
that
system
again
this
this
this
cycle
and
see
how
it
goes.
Obviously
it
didn't
happen
because
none
of
them-
I
don't
know
I
it
was
what
was
in
my
head,
but
I
didn't
act
on
it.
So
we'll
try
again.
Okay,
sounds
good
all
right!
Thanks
david
thanks
wesley
yeah,
okay,
we'll
see
you
I'll
talk
to
you
all
in
on
the
online
and
we'll
meet
again
in
two
weeks
says,
or
is
there
a
reason
not
to
meet
in
two
weeks?
A
I
don't
think
so.
There's
like
the
the
the
holidays
and
thursday
the
thanksgiving
holiday
might
be
around
that
time,
but
that's
thursday
not
tuesday
is
this.
Is
this
time
working
out?
Okay
for
you
both
that
works
for
me,
yeah,
okay,
let's
keep
it
where
it
is,
then,
all
right
great
thanks
to
youtube
bye.