►
From YouTube: .NET Design Reviews GitHub Triage
Description
We're looking at issues that are ready for review: https://github.com/dotnet/corefx/issues?q=is%3Aopen+is%3Aissue+label%3Aapi-ready-for-review
A
B
C
F
H
Edit
helpful,
you
should
maybe
working
doc,
it's
kind
of
its
kind
of
like
put,
but
generally
it's
used
when
you're
merging
information,
and
rather
than
replacing
it
myself.
So.
But
what
are
you
sending
for
that?
There
are
like
you,
absenting,
a
payload
that
has
like
yeah.
It
has
a
request
entity
body,
but
instead
of
having
say
say
that
a
like
I
want
to
update
your
dependence
in
an
HR
system.
Instead
of
sending
your
entire
employee
record
as
part
of
the
request,
I
would
just
send
only
the
dependents
field
or
something
longer
so.
C
H
I
J
J
C
I
M
N
C
Think
it
would
just
work
strange
so
personally,
like
I
would
say
when
in
doubt
always
follow
what
the
type
already
has
over
doing
something
new,
but
in
general,
like
I,
don't
I
can't
think
of
a
single
downside
to
doing
over
the
left-hand
side,
because
when
people
don't
really
see
the
API
is
like
this
right.
They
see
it
and
intellisense
collapse
by
name,
so
you
have
to
actually
do
a
little
them.
The
only
test
we
would
see
this
is
dogs.
I
only
say
that,
especially
for
dogs.
This
is
actually
somewhat
annoying.
If
you
see
them
twice,.
N
The
damage
has
been
already
done,
so
I
agree,
yeah,
you
see
them
twice
and
then
I
think
you
know
like
somebody
sooner
or
later
will
see
them.
Somebody
will
use
reflection
and
we
will
be
surprised
that
there
are
no
two
overloads
and
I
think
the
value
of
saving.
You
know
one
method
slot
all
right
sign
of
flow.
C
P
P
D
I
See
this
one
honorable
duel,
yeah
we're
trying
to
add
two
long
overloads
so
basically
there's
an
API
that
takes
a
12-part
file
path.
I,
don't
think
it's
correctly
Kim,
okay
Scott
case
it's
offset
into
the
file,
and
you
know
over
sockets.
You
just
send
that
power.
That
piece
of
file
there
are
seems
to
be.
Sometimes
there
can
be
large
files,
so
you
need
long
offset
modern
for
gigs
right,
and
the
only
question
is
how
to
expose
the
long
offset,
because
we
already
are
exposing
the
offset
as
int.
N
J
N
C
H
Yeah
we
should
just
put
the
long
and
the
future
proof
or
the
problem
is
I
agree
with
herself
again
like
if,
if
you
have
variables,
if
the
color
has
variables
which
are
starting
and
and
both
of
those
variables
are
long
they're
going
to
evaluate
to
a
long
once,
you
subtract
them,
which
means
you're
now
forcing
the
color
to
do
a
cast.
So
that's
what
I'm
saying
I
think.
M
L
N
E
J
N
N
N
H
So
her
MST
on
the
array,
dot
length,
counts.
The
number
of
elements
across
all
dimensions
of
the
array
and
each
dimension
can
have
up
to
two
gigabytes
of
up
to
billion
elements.
So
you
can
imagine
it
would
be
possible
to
create
a
array
with
you
know
three
times:
2
billion
dimensions
or
three
times
two
billion
elements.
N
C
L
N
I
O
N
Just
this
API,
so
you
know
it's
it's
just
like
they
array
and
we
talked
about
how
screwed
up
the
API
SAR.
How
do
you
use
an
API
where,
like
offset
and
count
can
throw
like?
Are
they
like?
Are
these
80s
now
obsolete
like?
Basically,
how
can
you,
how
can
you
safely
use
the
offset
property
after
this
change.
L
N
C
C
Weird,
because
now
you
basically
that
means
basically
what
you
want
is
this
right.
You
basically
say
I
call
an
existing
API
that
the
only
deals
with
ends
right,
so
that
means
I
can
only
accept
up
to
four
gigabytes.
So
I
can
refuse
this
thing
outright.
Let's
say
like
this
is
too
large.
I
cannot
handle
that.
N
C
C
N
N
C
C
N
C
C
N
C
I
N
N
C
I
C
I
see
something
we
call
the
so
they're
passing
the
fussin.
We
never
go
through
the
fuzzum
you
just
take
out.
Take
the
handle.
C
C
I
C
N
N
N
Chunk
because
you
take,
you
would
put
that
you
basically
just
do
one
windows,
API
call,
you
just
pass
the
handle
and
you
tell
it
like
sent
yeah,
but
then
you
just
increment
the
offset
for
the
chunk
and
you
call
the
API
again
for
the
same.
You
can
set
them
in
the
loop
with
the
same
handle.
He
said
that
the
handle
on
what
happens
with
the
ownership.
We
called
the
way
30-day
pians
I.
H
N
C
I
C
N
I'll
see
now
I
understand
what
they're
saying
so
there
are
so
many
when
you
open
the
file
stream
or
file
Hindu.
There
are
so
many
options,
so
we
would
have
to
either
put
all
these
options
into
this
type
or
it
basically
types
that
consume
this
type.
Are
you
talking
about
this
right
here
like
we
would
have
you
know
there
are
so
many
different
options
and
they
would
have
to
show
up
on
the
API
or
you
just
let
people
like
open
file
stream.
However,
they
want
I
think
whatever
options
they
want
and
then
we
just
stories.
C
N
N
C
N
C
So
basically
you
pass
it
here
is
the
socket
acing
event,
arcs
and
I
think
they
still
holds
the
element
right.
So
this
would
basically
your
one-off
thing.
No
I
mean
this
is
a
class
in
the
one
of
the
class.
You
have
two
locations
for
that:
I.
Don't
know
whether
that
actually
matters
when
you're
transmitting
large
file,
but
that's
one
of
the
things
it
may
matter,
but
that,
like
what
I
understand
like
what
do
we
have?
Why
will
we
add
this
API
rather
than
you
know,
take
the
eggy
element
directly
or
some
other
data
holder.
C
But
or
alternatively,
instead
of
us
adding
a
shit-ton
of
a
pianist
to
this
element
team,
we
could
add
a
struct
based
descriptor
that
you
can
design
to
be.
You
know
long
from
the
get-go,
so
we
don't
have
this
mode
problem
and
then
you
just
change
those
methods
to
accept
that
guides
that
they
don't
even
allocate
it
is
causing
instructor.
Then
you
just
have
the
same
information,
but
then
you
can,
you
know:
I
have
pal
stream
bass,
you
can
have
its
thin
base
and
you
just
don't
have
that
problem.
P
C
N
C
P
C
You
can
either
set
this
one
on
this
one
just
set
both
of
them.
We
throw
and
say,
like
that's,
an
invalid
request,
and
then
we
just
say
you
know
if
you,
if
you
only
specify
this
one,
but
not
a
long
one,
whether
or
not
one
can
be
from
this
one,
because
it's
a
implicit
conversion
and
otherwise
you
put
in
the
other,
so
the
cure
is
worse
than
the
disease.
In
this
case
the
other
hand
bill,
like
I,
don't
know
having
motes
on
this
type,
I'm,
not
sure
it's
better.
You
just
have
fewer
types.
C
I,
don't
take
your
federal
semantics.
We
have
modes
in
some
other
place
yeah,
but
the
most
the
other
place
are
I
would
argue
easy
turn
your
send,
because
basically,
this
other
thing
is
just
a
request
holder.
Well,
you
pass
in
the
later
and
you
call
a
method.
It's
like
you
know,
like
process
started
before
it
set
set,
set,
call
no.
F
R
F
N
So,
actually,
you
know
this
send
packet
element.
The
fact
that
it's
gonna
have
molds
I
think
we
cannot
avoid
it
to
some
extent.
It's
okay,
I'm,
not
concerned
about
slapping
more
api's
on
socket,
like
it's
already
kind
of
busy
and
I
think
it
would
be
nice
if
we
didn't
add
so
many
api's
for
advanced
scenarios.
C
Api's
are
there
I?
Did
they
I
could
see
them
I,
don't
know
what
they're
called
through
what
if
they
base
it
just
conveniences
to
call
the
other
API
I
would
agree
with
you
stretch
it,
because
if
you're
not
super
convenient
to
call
either
sitting
the
temp
of
arguments,
they
take
the
big
insight
if
I
like
the
old,
a
sink
right
so.
M
Q
C
N
Event
accessing
it
was
also
ginormous
type
so,
but
I
think
we
could
start
with
just
adding
the
socket
packet
element
api's,
and
if
people
complain
about
performance
which
I
kind
of
surprised,
I
mean
you're,
dealing
with
networking
and
sending
large
files
I
think
creating
one
object.
One
more
object
on
the
heap
is
not
such
a
big
deal
and
we
start
with
the
simpler
API.
If
people
complain,
we
add
more,
in
other
words
suckers
right
so
like
it
doesn't
have
to
go
over
the.
C
C
C
The
offsets,
and
so
those
are
the
ones
you
already
have,
so
we're
not
planning
on
anything
with
those.
So
those
are
the
ones
you
would
add
now
we
would
also
add.
N
C
This
cool
it's
true,
but
it's
the
only
way
you
can
pass
in
the
options
right.
Otherwise,
you
need
basically
an
API
that
takes
the
file
plus
everything
file
stream.
Constructor
takes
like:
what's
sharing,
do
you
have?
What
caching
do
you
want
to
use
right
like
this?
This
is
absolutely
like.
The
you
know
you
control,
essentially
how
Windows,
who
access
the
file
at
construction
of
the
hand
online?
C
O
L
H
C
Know
I
mean
the
answers
welcome
today,
by
the
way,
like
stream,
reader
stream
writer.
When
is
lose
deposit
positive
to
then
they
start
owning
that
guy.
So
it's
a
if
it's
an
established
pattern.
It
I
mean
it
is
a
great
pattern
to
be
clear
like
especially
for
file
readers
and
others
way.
You
sometimes
surprised
to
in
align
extreme
with
this
post
when
you
disclose
the
reader
but
I.
Think
in
this
case
I.
Don't
think
you
are
surprised
per
se
they're
just
like
is
he
all
you
all
you
would
do
with
the
call
site?
C
J
C
H
I
C
S
I
Okay,
I
guess
I
think
it's
probably
about
you
know
not
you.
This
flag
is
used
to
granular
control
of
the
content
of
each
message
on
a
Datagram
or
a
message.
Oriented
socket
so
like
send
the
message.
Basically
don't
wait
for
more
data
or
something
like
that,
I,
don't
know
why
we
don't.
Maybe
it's
like
from
the
you
know,
all
a
sync
style
timeframe
and
we
didn't
have
just
the
array
in
the
RX,
but
the
acts
as.
H
H
L
C
C
C
C
N
Like
that's
just,
but
if
you
not
access,
if
you're
not
accessing
these
properties,
you
just
pass
this
type
through.
You
don't
have
a
problem
yeah.
The
problem
is,
if
you're
actually
accessing
these
properties,
but
you
got
the
type.
You
know
the
instance
from
some
other
library,
then
you
may
be
surprised
that
suddenly
it
starts
growing
and
what
I
mean
basic.
N
The
only
thing
kind
of
against
obsoleting,
so
I,
imagine,
let's
say
I'm
writing,
sockets
and
or
my
own
wrapper
for
sockets
I
get
one
of
these
types.
I
read
the
buffered
property
is
not
know,
so
somebody
just
passed
me
in
the
right
cell
I
would
rather
access
the
offset
and
count
properties
because
then
I
don't
have
to
cast
an
apple
store
water
that
will
still
work
and
it
will
always
work.
Yes,.
I
I
K
N
J
N
Let's
leave
them
nope
saluted
because
it
would
be
very.
C
H
H
M
M
H
N
H
H
I
File
stream
over
to
do
so
to
construct,
we
have
actually
we
have
buffer
and
file
path
for
writing.
So
they
should.
It
should
be
the
same
thing
like
you
know.
If,
if
you
put,
if
you
don't
bounce
buffer,
if
you
pass
for
pot
for
pot,
it's
gonna
return
knowledge.
You
lie
just
check
it,
so
I
think
the
vision
becomes
a
stunt.
Now
the
question
is:
if
we
should
expose
the
file
stream
in
case,
you
are
passed
us
only.
The
file
path,
so
I
would
say,
I
would
I
would
basically
say.
I
M
N
C
C
Q
C
H
H
C
H
N
C
C
N
P
N
N
I
P
I
P
You
don't
have
mats
as
well,
so
was
he
said
he
tries
his
best
to
represent
that
I.
Actually.
R
C
Here
we
go
so
generally
speaking,
so
my
concern
with
this
API
is
pretty
simple.
Like
I
would
say
to
me,
it
seems
a
bit
odd
that
we
had
ideas
here,
that
we
don't
adhere
specifically
the
one
that
returns
these
guys.
It
seems
odd.
The
second
thing
is
the
pattern
itself,
I
think
borne
some
discussion,
whether
we
want
to
do
this
at
all,
but
it's
say
if
you
would
remove
those
ones
from
the
discussion
for
now.
The
remainder
looks
sensible
to
me.
No
problems
with
that.
I
N
Later,
if
you
will
do
at
this
pattern
or
not
so
I,
this
is
a
kind
of
comment
that
I
would
have
is
I,
don't
think
we
have
any
try,
API,
Sun,
link
and
I.
Think.
The
reason
is
that
people
like
to
change
these
8d
skulls,
so
most
of
these
ATMs,
a
I
could
imagine
them.
Instead
of
this
being
try,
we
use
Nala
boo,
because
knowledge
can
be
chained.
You
can
basically
do
you
know
value
or
default.
Then
you
end
up
with
a
default
value.
You
can
keep
changing
them.
There's
never.
First.
R
Internship
return,
zero
and
sort
of
this
pattern
in
general.
Doesn't
work
link
and
that's
the
reason
that
first
or
default
similar
evil
exists,
a
the
only
advantage
value
of
this
API
so
differentiating
case
in
which
you
got
a
null
value
from
a
case
in
which
you
actually
didn't
find
a
value.
So
why
would
we
not
return
Allah
boy
instead
of
doing
try
baby
is
so
novel
is
because
not
a
will
has
that
problem
that
you
cannot
two
difference
between
something
being
now
and
something
not
being
there
ago.
So.
R
C
N
R
What
is
Warren
had
a
comment
all
that
he
explained
that
if
you
let's
say
I
just
put
this
this
call
inside
a
nested
inside
a
linked
query.
Then
you
are
forced
to
share
that
out.
Parameter
is
with
in
all
the
calls
that
you
make
it
when
you
are
iterating
over
turns
fault,
yes
and
then
the
query
we
can
start
affecting
and
then
basically
the
whole
paradigm
of
thing
blows
up
like
you
cannot
execute
in
parallel
anymore,
but
basically
it
no
other
thing
in
Inc,
as
I
said
that
this
is
like
one.
C
R
Cleaner
problem
I
mean
I'm,
not
trying
to
argue
in
favor
of
the
adding
this
API,
but
they
problem
with
that
argument
is
that
if
you
make
couple
the
double
version,
the
only
one
that
we
add,
then
it
kills.
The
usability
that
they
were
trying
to
do
add
is
basically
the
whole
point
of
this
API
is
that
is
sure
to
do
this
thing
of
trying
to
get
the
first
value
with
the
option
to
know
whether
there
wasn't
the
first
value
there
right
in
just
one
step
and
efficiently.
R
C
C
H
C
R
R
I
might
not
have
been
there
and
there
are
other
ways:
I
mean
if
you,
if
you
really
want
to
do
this,
you
can
take
the
ienumerable
or
the
wearable
a
get.
The
numerator
and
iterate
iterate
once
I
mean
get
the
move
next
right,
Ethel
see
if
there
is
something
there
and
then
if
there
is
another
move
next,
for
instance,.
N
R
N
C
N
C
Like
the
other
thing
is,
you
cannot
enforce
consistent
naming
because
it's
purely
you
know,
convention
that
used
to
claim
its
success
and
way
you
name
it
like
forward
bar
or
not
aiming
at
all.
It
was
about
it,
in
which
case
it's
hard
to
document,
a
concept
that
is
literally
like
convention
right.
It's
very
easy
to
document
the
FBI
and
say
use
option
of
T
for
cases
X&Y
the.
R
C
I
C
N
Becomes
well
except
we
repeatedly
have
the
problem
that
no
level
of
T
is
constrained
to
struct.
Therefore,
you
cannot
basically
it's
a
type
that
cannot
be
used
in
genetic
code
and
like
if
you're
dealing
with
T's
and
the
T's
are
not
constrained
to
struct
you're
screwed.
We
don't
basically
have
a
type
for
it.
Sorry.
C
N
N
N
C
C
I
get
that
I'm
saying
is
that
we
have
to
like
I,
think
it's
very
weird
to
me
to
ever
tie
that
they
don't
use,
because
that
ends
up
usually
in
the
world,
where
nothing
interacts
with
the
type.
But
then
it's
like
a
I
think
you
have
to
have
semantics.
For
example,
do
we
want
to
have
an
implicit
cast
from
inaudible
of
T
to
an
optional
of
T,
for
example?
Or
do
we
want
to
have
a
try,
parse
pattern
that
returns
an
option
of
T
or
but.
N
Would
we
use
it
in
more
places
than
this
one
and
that
we
punt,
and
then
we
six
months
later,
we
in
the
same
spot,
yeah
so
you're
saying
if
the
virtual
add
type
would
use
it
I
will
just
add
it
and
be
done
with
it.
If
we
think
that
okay,
so
the
only
thing
I
would
do,
is
I
would
add
it
if
this
API
would
actually
use
it,
yeah
I
would
not
add
it
like
just
you
know,
just
for
the
sake
of
it,
because
we,
but.
M
C
C
That's
the
question
I
mean
like
you
may
have
to
be
careful
that
you
don't
like,
for
example,
one
common
problem
that
the
language
guys
have
is
depending
on
what
you
want
the
semantics
to
be,
for
example,
you
may
or
may
not
want
that
the
rules
participate
all
over
resolution.
If
you
want
to
do
that,
then
it's
certain
you
know
design
and
then
there's
other
things
where
a
cyclic
design,
like
I,
think
fundamentally
I.
Think
you
should
talk
the
language
guys
to
say
if
we
were
to
add
optional
of
G.
C
What
would
that
mean
like
how
would
a
director
of
reference
type?
Does
it
at
all?
How
would
interact
with
nullable
types?
Does
it
at
all
when
we
have
some
tactic
sugar?
If
so,
what
does
it
look
like?
Is
there
any
until
up
story
between
these
three
things
in
this?
So
you
know,
what
do
you
want
to
do?
I
mean
Johnny
speaking,
they
were
the
ones
pushing
forward
in
the
walls
at
timeframe,
because
you
know
for
the
wither
is
essentially
the
absence
of
a
video
yeah,
as
opposed
to
now.
R
C
C
At
this
discussion
of
crystal
view
time,
they
disagree
with
crystal
I.
Think
like
to
me
to
me.
Try
like
he
was
basically
saying
here,
everything
alone,
but
I
think
the
idea
was
try.
Was
we
base
it
on
for
an
exception?
Like
this
elf
story,
an
exception?
We
give
you
return
million,
you
can't
really
be
succeeded
or
not.
We
have
used
try
in
other
places
where
it's
not
really
an
exception.
C
C
R
C
C
N
A
C
H
C
N
C
C
R
R
R
R
R
P
A
K
C
For
the
person
filing
that
issue,
like
it's
me,
things
like
the
this
is
like
it's
mostly
an
action
item
on
us,
whether
we
believe
this
API
is
useful
enough
to
have
the
conversation
on
optional,
but
I
also
kind
of
repress.
Of
that
it
just
happens
to
be
the
current
issue.
A
option
would
have
been
venue,
but
we
all
you
know
the
optional
would
generally
pay
your
bowleg
to
me.
It's
not
like.
Oh,
we
have
to
invent
optionality
just
for
this
one
case
right.
It's
just
get
another
timer
that
will
be
ends
so.
N
I
yeah
I
would
say
that
maybe
we
should
do
the
discussion
that
you,
because
there
are
all
the
points
that
you
made
a
good
points
like
when
we
design
optional.
There
are
many
questions
that
would
need
to
be
answered.
I
think
if
we
answer
them
and
we
do
add,
optional
I
think
it
would
be
the
best
way
to
design
this
API.
Is
there.
I
G
F
O
C
In
there
as
well
like
this
one
doesn't
use
option,
this
one
only
uses
the
try
pattern
which
we
don't
like,
because
it
has
the
outer
arguments
right
and
then
this
one
has
both
of
them
like
I,
would
say
generally.
It
seems
very
weird
to
me
that
in
your
mobile
career
will
have
different
API
servers,
I.
I
S
Q
C
R
I
I
I
Q
I
About
the
Train
yeah,
because
I
owe
virtually
no,
it
was
the
resize.
The
thing
is
that
we
will
never
I
believe
we
will
never
actually
try
to
make
it
smaller.
That's
not
a
bother
it's
more
about
upfront
upfront
allocation,
because
you
know
that
I'm
gonna
be
inserting
ten
thousand
things,
so
please
don't
reallocate
over
and
over
again
the
way.
So
the
you
know
agreement
at
least
for
that
scenario
was
like
to
rename
it
or
you
know,
to
introduce
any
short
capacity.
N
N
I
Q
I
Bit
in
general,
and
we
don't
know
how-
and
we
thought
about
it
like
you
know
they
thought
about
it.
Like
you
know,
you
don't
know
how
to
make
the
resize
be
meaningful,
because
the
resize
kind
of
says
that
you
know
it's
going
to
shrink
as
well,
how
to
make
it
not
a
confusing
API
or
the
other
option
is
to
call
it
ensure
capacity.
Yes,
which
might
be
potential
if
we
one
day
find
out
some
beautiful
name
and
beautiful
behavior
might
be
suddenly,
like.
I
K
H
C
I
H
N
It
yeah
I
completely
agree,
but
this
is
basically
the
difference
between
our
team
and
customers,
like
customers
want
what
they
want
at
this
given
month,
and
then
we
keep
adding
ideas
that
are
not
general
purpose
and
then
we
end
up
having
like
moulds,
because
oh
geez,
we
had
this
API
two
years
ago
to
to
increase,
and
now
we
have
this
other
to
trim.
How
do
they
relate
and
blah
blah
blah.
I
N
So
I
know
that
resizes
a
bit
more
complex
but
I
would
say
the
semantics
should
be
the
following.
You
pass
resize
and
you
can
pass
a
smaller
value
than
the
current
size
of
the
internal
structure,
but
the
fact
will
devalue
the
semantics
of
the
value
that
you
pass
in
are
the
following:
I
will
be
able
to
add
as
many
items
to
the
dictionary
we
found
any
reset
Sagarika.
H
A
H
N
I
N
N
N
C
N
C
N
H
C
N
What
do
you
mean
by
her,
oh
by
ear?
Why
does
it
make
it
harder
I
think
that
some
people
commented
that
some
you
know
hash
based
data
structures.
They
may
have
problems
determining
the
absolute
lower
bound
to
not
resize,
but
the
value
of
a
margin
of
error,
not
on
this.
They
they
not
only
have
a
load
factor.
They
also
have
the
margin
of
error.
C
C
N
Q
C
I
mean
because
that
now
we
aren't
that
get
to
me
like
this
could
have
been
the
same
API.
We
just
decided
that
we
don't
want
to
do
the
same
API
and
that
seems
just
odd
I
think
it's
mariza,
just
at
180
I,
taste
capacity,
and
if
you
pass
away
the
larger
the
smaller
the
account
we
throw
and
anything
larger
than
count
is
fine
and
then.
N
I
C
C
The
size
of
the
array
to
me
is
like
recomendation,
ization,
specific
factor
of
plus
nears
five
I.
Don't
care,
like
you,
there's
some
algorithm
that
you
have
to
run
today
when
I
do
new
dictionary
and
pass
in
the
thousands
for
capacity
right.
So
somehow
you
took
that
value
and
turn
it
into
a
size
of
the
of
the
array
like
our
contract
is
not
that
when
you
pass
in
is
the
actual
size
of
the
array.
All
we
are
saying
is
that
the
contract
is,
is
the
if
I
up
to
that
number?
You
don't
have
to
resize.
C
I
C
C
C
N
Who
pick
some
name
that
conveys
the
concept
but
I
thought
that
the
argument
was
that
resize
implies
that
it
can
shrink
in
sure
capacity
does
not
say
anything
that
it
will
shrink
gray
and
we
need
a
name
that
says
it
can
either
shrink
or
grow,
which
resides,
in
my
opinion,
is
a
better
name:
everything.
Okay,
it's.
B
C
P
T
H
N
C
S
H
I
C
C
C
Don't
want
to
store
the
radio,
secondly,
they
I
think
for,
but
for
a
thing
that
wraps
the
buffer,
making
the
capacity
settable
I'd
argue
still
something
I
find
weird,
because
you
at
least
have
to
allocate
new
and
copy.
So
it's
not
what
I
would
consider
a
cheap
operation,
but
for
something
like
rehashing
and
like
that
seems
like
a
really
bad
game.
Yeah.
H
C
N
Q
C
N
N
I
M
C
C
Not
a
great
name,
but
on
the
other
hand
like
it's
an
established
name,
I
mean
it's
one
API
we
have
the
framework
much.
This
is
an
established
API,
but
I'm
kind
of
ok,
so
4
5,
so
I
can
kind
of
claim
fire
for
that,
so
that
would
basically
fuzz
before
I
just
set
the
stupid
wish
this.
The
proposal
that
David
had
here.
H
N
E
N
C
D
C
N
C
Today,
it's
going
to
be
simple.
Another
thing
I
mean
to
keep
in
mind
like
the
reason
they
want,
the
sapiens
probe,
optimizations
right
so
I
think
once
they
pass
the
fact
that
we
resize
things
all
over
again,
and
somebody
may
fight
over
these
long
running
things
over
here
and
all
we
want
to
do-
trim
excess
of
those
like
given
the
cost.
We
have
to
wrap
this
a
few
guys,
which
is
the
work
that,
at
the
same
time,
with
other
Miriam.
C
C
C
C
I
C
I
H
C
O
N
J
N
C
N
B
N
K
I
I
L
D
G
I
Well,
I'll,
stop
saying
that
word!
Oh,
it's
all
roughly
would
be
really
useful
if
we
can
do
that.
Do
you
know
if
it's
three.