►
From YouTube: wg-incr-comp mtg 12, 2021-02-09
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
And
then
we'll
go
from
there,
okay,
so
this
is
the
12th
meeting
of
the
incremental
violation.
Working
group
wesley,
unfortunately,
cannot
attend.
David
has
been
so
kind
as
to
at
least
get
us
started
with
an
agenda,
that's
linked
from
the
zulip
channel,
and
with
that,
let's
just
do
a
quick
round
of
you
know
status
reports.
I've
done
very
I've,
been
almost
nothing
related
to
this
group
working
group
and
I
have
been
doing
things
really.
A
The
competition
as
a
whole,
which
I
think
will
relate
to
the
working
group,
but
we'll
we'll
talk
about
that
more,
I
think
after
we
go
through
status
stuff,
so
let's
just
go
around
and
see
what
how
everyone
else
is
doing
david.
C
Okay,
yeah
right
in
santiago.
I
think
I
was.
I
was
fortunate
enough
that
I
was
working
on
a
traits
when
I
traced
pr
and
ended
like
impacting
some
stuff
on
incremental
compilation,
but
I
haven't
been
like
I
haven't,
been
doing
anything
specifically
about
incremental
compilation.
Basically,.
A
A
It
yep
by
the
way
I
don't.
I
suspect
this
is
a
holdover
from
whatever
previous
agenda
was
there.
I
do
want
to
mention
like
there's
an
issue
with
the
date
where
the
dates
are
formatted.
This
is
striking
me
because
we
had
a
thing
at
my
job
where
people
were
like
going
lam
going
off
about
dates,
date,
formatting
and
stuff.
You
know
how,
like
nine
slash,
two
is
ambiguous
about
whether
it's
denotes
september,
9th
or
february,
9th
or
september,
2nd
or
february.
A
I'd
say
I
mean
I
think
the
answer
here
is
to
not
use
numbers
for
the
months
instead
use
three
letter.
You
know
jan
feb
bar,
because
then
it's
unambiguous.
What
you're
talking
about,
I
think,
but
anyway,
let's
quickly
run
through
the
items
then
they're
on
here,
just
to
make
sure
we
review
them
and
keep
track
of
them.
So
we've
got
79560
the
diesel
crate
thing
and
yeah
we're
still
talking
about
trying
to
add
diesel
to
the
ci,
but
I
haven't
done
anything
there.
A
Yet,
despite
saying
I
would
but
you're
right
your
pull
request,
lance
santiago.
So
that's.
A
Right
right,
so
we've
got
the
split
dwarf
pr
and
I
know
that
you've
yeah,
I
know
david
you've
been
asking
for
feedback
here.
I
think
the
heart
of
this
is
that
we
we
don't
typically
have
time
to
dig
in
to
something
like
this,
and
the
question
is:
is
it
worthy
of
like
a
a
design
meeting
or
or
something
we
gotta
figure
out
some
way
to
make
progress
here?
Do
you
david?
Do
you
have
any
like?
A
So
I
suspect
we're
if
we
don't
have
stakeholders
now
that
are
you
know
known
to
us
they're
going
to
be
known,
soonish,
that's
my
way
of
saying
like
even
though
you're
not
getting
feedback
right
now
I
think
you
might
see
people
say
why
was
I
consulted
if
we
just
take
action
without
you
know
some
sort
of
process.
B
Yeah
I
mean
that's
stuff.
That
is
really
it's
not
user
facing
it's
just
how
we
want
to
persist.
The
dwarf
information
between
dependencies
to
the
later
compilations.
B
Yeah,
so
we've
output
dwarf
objects
and
when
you
then
link
them
together
a
package,
it
expects
the
dwarf
objects
from
dependencies
to
still
exist
and
they
don't
right,
and
so
the
question
is:
how
do
we
make
sure
they
still
exist?
Do
we
put
them
into
the
our
lab?
Do
we
like
other
solutions
like
that?
Do
we
have
a
flag.
A
B
A
If
that's
the
case,
I
think
we
should
take.
We
should,
instead
of
just
waiting
on
people
to
get
feedback,
I'd
say
we
should
make
an
informed
decision
or
on
an
or
even
uninformed
decision
and
and
just
move
forward,
because
this
is
like
a
real
bug
that
needs
to
be
fixed
and
it
seems
like
if
it's
truly
just
an
internal
detail,
then
even
if
we
make
the
wrong
decision,
we
can
go
back
and
fix
it
right.
A
So
that's
that's
my
inclination
there
then,
given
that
we
haven't
really
gotten
much
forward
progress
over
in
the
last
19
days.
I've
actually
have
feedback
on
it.
Okay,
we've
got
seven
nine
five,
one
nine
story
here.
Actually
it's
in
a
side
table.
A
This
is
an
example
of
something
where,
like
you
know,
someone's
asking
me
like
how
like
what
I,
what
low-hanging
fruit
do
we
have
for
for
making
incredible
better-
and
I
was
thinking
of
something
like
things
like
this,
but
I
didn't
at
the
time
say
things
like
this
that
I'm
talking
about
so
okay,
there's
been
some
progress.
Perf
has
been
run,
pr
is
conflicts
and
it
sounds
like
this
basically
wesley's
been
owning
this
in
terms
of
reviewing
it.
So
I
don't
know
what
else
to
do
there
besides
just
letting
it
keep
moving
forward.
A
It's
a
long,
that's
a
lot
of
commits
on
here,
but
I
think
they're
factored
well.
It
seems
like
hopefully,.
A
Well,
the
beach
there
was
there
was
something
about
changing
to
a
b
tree
method.
That
was
that,
at
the
time
of
the
last
meeting,
let's
see.
B
A
C
A
See
I
see
there
was
commentary
from
wesley
17
days
ago.
I
think
what
it
was
in
the
meeting
is,
we
sort
of
said
we
still
want
to
go
back
and
try
the
thin
index
vectors
and
like
like
instead
of
the
b
tree
bath,
I
think,
is
what
the
takeaway
is
there
I'll
go
back
and
review
the
last
meeting,
which
you
both
got
right.
I
shared
that
with
you.
I
believe
yes,
okay
and
so
I'll
add
an
action
for
myself
to
put
to
say
something
here.
A
A
A
A
So
the
compiler
team
sprint
was
that
that
was
something
I
think
we
really
were
intending
on
talking.
We
did
talk
about
the
last
thing.
That's
why
this
item
is
on
here
yeah
and
then
in
the
actual
planning
meeting
for
the
sprint,
we
decided
to
change
focus
or
you
know.
Basically,
there
was
enough
feedback
saying
no
focus
shouldn't
be
incremental
violation;
it
should
be
on
memory
usage,
so
that
meant
we
sort
of
scrapped
a
fair
amount
of
stuff
that
we've
been
thinking
about.
A
Regarding
the
sprint
in
particular,
there
have
been
talk
about
trying
to
do
a
survey
amongst
our
users
to
like
get.
You
know,
performance
profiles
of
machines,
basically
marks
simulacrum
convinced
me
like.
A
That
was
not
a
good
use
of
effort
right
now,
because
there's
enough
stuff
that
we
can
do
internally,
based
on
our
own,
like
observations
of
what's
bad,
that
we
shouldn't
be
wasting
survey
effort,
but
especially
from
the
people,
be
filling
out
the
survey
or
doing
it
on
that
until
until
we've
really
taken
care
of
the
low
hanging
fruit
ourselves,
which
I
was
convinced
by,
we
should
wait
to
do
it
until
we
really
need
the
information,
though
I
think
that
we
could
still
benefit
from
some
investment
there,
maybe
just
in
terms
of
having
a
good
process
in
place.
A
The
other
thing,
though,
is
in
terms
of
the
actual
sprint
in
march,
I
still
am
curious.
Whether,
like
you
know,
memories
memory
usage
is
the
focus,
how
much
of
that
could
be
in
our
wheelhouse
in
terms
of
the
memory
usage
of
the
incremental
compilation
structures
themselves,
I
have
no
idea
like
what
the
depth
map
and
stuff
like
that
ends
up
consuming
with
the
runtime
profiles.
I
have
never
really
looked
that
carefully
at
the
like.
I
I
I'm
aware
of
the
whatever
the
tool
is
called
on
the
memory
profile.
A
B
A
Yes,
thank
you
massive.
Yes,
that's
right
with
an
if
at
the
end-
and
I'm
just
I
don't
even
recall,
inspecting
like
the
massive
profile
for
incremental
versus
non-incremental,
but
that's
the
kind
of
thing
where
there
might
be
information,
that's
useful
there,
I'm
hoping
or
no
I'm
more
than
hoping.
A
I
I'm
actively
working
on
trying
to
make
tooling
for
also
getting
instrumentation
from
within
the
compiler
about
memory
usage,
basically
based
on
some
ideas
that
idp
had
posted
years
ago,
and
so
I
think
that
I'm
going
to
try
to
use
employ
that
myself
locally
as
part
of
my
upcoming
work.
This
is
all
me
a
way
of
me
saying,
like
I
think
that
we
still
have
stuff
to
offer.
That's
stick
to
this
sprint,
and
so
it's
worthwhile
just
keeping
the
back
of
your
mind.
A
Like
you
know
what
are
some
things
that
we
could
ask
we,
you
know,
don't
worry
necessarily
about
fixing
the
memory
usage
issues
right
now,
although
you
could,
but
rather
like
okay.
How
would
I
even
like
figure
out
where
the
problems
are
and
better
ways
to
identify
and
investigate
them?
A
Okay,
so
that's
that's
all
I
want
to
say
there
and
then
we
have
the
same
issue
with
issue
triage.
Where
I
haven't
done
anything,
I
don't
think
we
want
to
do
it
live
right
now.
Do
you
all
have
anything
else
you
want
to
sort
of
add
to
the
oh?
Oh,
there
was
one
more
thing
I
want
to
make
sure
we
talked
about
did
either
of
you.
Neither
of
you
were
part
of
the
group.
That
said,
oh
yeah,
let's
help
with
the
before
the
perf
rlo
website,
right
on.
A
Busy
right,
yeah:
okay,
that's
fine!
I'm
not
expecting
your
input
on
that
in
terms
of
active
work,
but
I
do
think
that
a
huge
part
of
the
incremental
violation
or
sorry,
a
huge
part
of
the
perf
site
is
dedicated
to
stuff
about
incremental
violation
for
better
or
worse
and
so
again
that
kind
of
falls
into
our
wheelhouse
about.
A
A
If
you're
using
perf,
assuming
you
do
use
perf
rlo,
if
you
could
just
keep
some
in
the
back
of
your
mind,
if
you
think
about
okay
here
the
things
that
actually
I've
found
useful
and
things
I
haven't
find
useful,
especially
with
respect
to
the
data
regarding
common
violation.
A
Either
have
them
back
your
mind
or
take
notes
or
whatever.
I
think
that
it'll
be
useful
to
have
some
sort
of
feedback
there
as
we
go
into
revising
the
site,
because
I
personally
am
very
worried
about
us
over
the
current
site,
I
think,
is
over
biased
towards
incremental
compilation
data,
and
I
worry
about
that,
meaning
that
the
cases
that
some
people
care
the
most
about
are
underweighted
in
terms
of
the
presentation
of
the
data.
B
C
So
if,
if
I
were
to
pick
a
like
a
some
issue
or
something
like
given
that
I
haven't
been
around
like
what's
on
vacations
and
all
that
do
you
consider
there
is
an
issue,
an
important
issue
or
something.
What
would
you
suggest?
I
don't
know
if
or.
A
A
A
I
don't
know,
I
think
that
any
I
think
my
personal
guess
is
that
any
of
the
things
in
the
things
tagged
with
wg
ink
comp
could
be
interesting
to
look
at
there's
a
link
in
the
issue
triage
section
of
the
meeting
notes
for
this
right,
and
it's
just
a
question
of
I
think
I'd
be
more
rather
than
I
mean
if
you
want
to
try
to
talk
about
identifying
the
most
the
highest
profile
thing
or
the
things
that
mice
that
might
you
know,
yield
the
best
benefits.
A
I'm
not
sure.
I
think
that
the
idea
that
you
know
looking
at
things
like
dropping
unused,
unreferenced
external
creation,
I
triggle
recompile
that
the
fact
that
that
industrial
recompile
is
interesting
and
might
be
a
sign
of
like
something
deeper
that
needs
to
be
fixed.
I
don't
know
so
I
I
don't,
but
I
don't
know
if
trying
to
identify
that
the
highest
bang
for
buck
issue
is,
is
as
important
right
now
versus
just
like.
What
do
you
think
you
have
the
best
chance
of
like
right.
C
C
So
justin
did
you
say:
wg,
oh
yeah,.
A
Or
a
incorrect,
no,
I
said
wg
I
would
mostly
say
wg
because
I
was
like
there's
only
seven
of
them
and
so
it'll
be
easier
to
get
through
to
look
them
quickly.
That
was
my
own
bias
there
in
terms
of
like
okay,
if
I'm
going
to
say
an
answer
right
now,
let's
look
at
like
the
short
list.
Yeah,
the
other
one
is
too
long
and
that's
we're
still
the
process.
There
is
we're
just
trying
to
triage
those
right,
yeah
yeah
they
there
might
be
a
lot
of
there.
Are
there.
A
In
there
there
might
be
a
lot
of
junk
in
there.
So
if
you
want
to
spend
time,
triaging-
hey,
that's
that's
fine
too
yeah,
there's
lots
of
possible
things
that
might
be
interesting
here.
There's
a
lot
of
stuff.
A
Okay,
it
sounds
like
we're
basically
done,
then
I
will.
My
plan
is
to
take
the
same
approach
with
this
meeting
that
I
took
with
the
last
one
in
terms
of
putting
it
up
on
youtube
and
making
it
unlisted
and
then
sharing
the
link
with
whoever
wants
it
from
this
working
group
right.
I
I
debate
I'm
still
debating.
A
Maybe
maybe
we
can
go
back
to
a
model
where
we
post
the
meetings
openly
again,
just
because
I'm
not
sure
that
the
the
reasoning
that
I
had
for
you
know
making
the
meetings
unlisted,
I'm
not
sure
it's
going
to
bear
fruit
in
terms
of
like
why
rightly,
the
the
the
people
that
I
was
trying
to
get
involved
so.
C
C
I
think
one
important
input
that
you
need
is
that,
for
instance,
for
me,
I
don't
have
any
problem
if
you
want
to
like
publicly
share
the
thing
you're.
A
A
Right
because
to
me,
if
nothing
else,
it
makes
me
less
stressed
about
the
link,
like
you
know,
trying
to
protect
the
link
like
saying
it's
going
to
be
public
means
I
can
do
things
like
put
the
link
in
the
chat
room
and
not
worry
about
it.
Okay
sounds
good
thanks.
You
thanks.
You
too
talk
to
you
later
bye.