►
From YouTube: 2020-03-10 :: Ceph Crimson Meeting
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).
C
A
B
B
A
B
Thank
you
for
a
very
and
tried
it
out
make
a
new
system
I
think
to
win
I
I!
Don't
need
to
to
to
to
mimic
the
Invitational,
but
I
was
in
the
sense
that
we
need
to
track
the
path
that
which
means
that
the
note
the
reference
to
or
note
when
when
when,
when
going
down
into
the
leaf
node.
So
we
can,
though
we
can
be,
had
merge
the
the
parent
notes
without
without
going
back.
D
I
reversed
some
back
fixed
patch
to
verify
the
the
promise,
but
it
seems
the
following
back:
fixed
patch
not
influence
the
promise
without
a
still
carried
the
same
result
and
but
I
can't
get
the
result
Acosta
in
December,
so
not
sure
what
the
reason
and
the
code
is
not
changed,
masked
match
for
any
stock
and
so
I
think
I
can
put
it
away
for
some
time
and
the
since
the
any
star
is
not
a
high
priority.
It
it.
B
D
D
A
A
D
A
I'm
working
on
the
scrubbing
going
over
the
code
and
the
since
yesterday
I
have
a
loosely
chart,
license:
I'm
writing
I'm,
creating
a
diagram
of
the
exact
state
chart
state
charts
for
rubbing
I
hope
to
serve
as
documentation
to
make
sure
that
I
understand
everything
and
I
hope
to
finish
this
in
a
day
or
two
and
send
thanks
for
the
answer.
Yesterday.
B
A
E
B
E
E
You
can
look
at
the
way,
I've
used
it
in
the
LB
a
tree,
but
it's
Oh
a
lot
of
the
I'm
currently
writing
that
bed
code.
So
it's
a
little
sparse
right
right
now,
but
the
exam
I
think
was
asking
last
week
about
how
to
provide
buffer
lists
and
how
to
interpret
it.
There
are
methods
now
for
that
or
interface
methods
that
you're
attended
to
a
whole
bunch
of
em's.
But
what
comments
at
least
indicating
what
I
sort
of
expect
them
to
do?
E
C
C
F
Looking
until
I
last
way,
I
looked
at
the
Google
beat
fee
that
current
confuse
information
referencing
it
there
are
many,
for
there
are
many
useful
tricks
about
memory,
management
and
temperance,
but
the
implementation
will
be
very
differently
because
it
is
not
a
B+
tree
and
we
will
have
therapist
sized
keys
and,
most
probably
I,
think
it
will
be
hash
to
a
fixed
sized
key
for
indexing
performance.
But
we
will
still
do
do.
Is
hash
collisions
and
I'm
I'm,
not
very
sure
about
the
parent,
pointer
implementation
them
all.
F
F
B
E
E
B
E
If
you're
gonna
erratic
three
will
be
the
other
way
around
right,
so
radix
tree
Hanif,
it's
from
six
eyes
keys
with
the
uniform
distribution.
That's
the
description
of
Oh
notes,
at
least
the
32
bit
of
bro
map
keys.
You
have
no
control
whatsoever
over
the
length
of
the
key.
You
don't
have
any
control
over
how
long
the
prefixes
are.
You
don't
have
any
control
or
boothy
over
the
distribution
either
they
could
all
have
the
same
prefix
for
it.
For
instance,
I.
E
Saying
all
of
those
things,
I
just
said
are
very,
very
bad
for
radix
race.
So
no
I
think
that
the
Oh
map
tree
is
almost
certainly
a
v3.
The
Ono
tree
could
conceivably
be
erratic
Street,
but
you'd
have
to
do
something
to
deal
with
collisions
and
you
can't
assume
there
won't
be
very
many
of
them
because
to
remember
snapshots
are
collisions
and
people
can
have
a
lot
of
snapshots
of
a
particular
object.
E
B
E
B
E
E
Common
trick
and
beating
that,
if
you're
trying
to
avoid
an
explosion
of
space,
is
to
do
prefix
compression
right,
if
you
know
all
of
the
keys
in
a
particular
node
start
with
the
same
prefix,
you
write
the
encoding
so
that
that
prefix
is
in
the
beginning
of
the
note.
And
then
you
don't
repeat
it
for
the
rest
of
the
note
looks
like
like
like
that,
but
that's
a
little
tougher
with
a
sort
of
uninterpreted
topple
concept.
E
So
I
actually
do
have
six
live
streams
or
six
left
keys.
I
can
do
pretty
straightforward
stuff,
it's
still
taking
for
as
complicated.
My
my
suggestion
actually
is
that
you
consider
it
as
a.
E
That
is,
it
just
happens
to
be
the
case
that
string
sorting
gives
the
same
ordering
that
H
object
sort
it
currently
does
you
can
do
something
similar
with
a
not
necessarily
principal
byte
encoding
and
used
some
trickery
to
ensure
that
maybe
to
ensure
that
you
can
quickly
interpret
it
without
being
expensive
to
decode,
but
I
think,
ultimately,
what
you're
looking
at
is
a
substring
I'll,
be
at
a
string
with
some
rules,
but
a
string
like
you
always
know
the
first.
However,
many
bytes
are
going
to
be
the
hash.
E
E
E
Before
is
cheating
they
get
to
not
they
don't
have
to
list
things
in
lexicographic
order.
We
do
or
they're
not
cheating
with
POSIX.
It
doesn't
require
them
to
list
elements
in
lexicographic
order.
That's
what
so
butter!
That's
is
the
same
thing.
It
houses
things
128
bits
and
then
does
b3
based
on
that
right.
That
works
for
them,
because
if
you
put
10
files
into
a
directory,
you
don't
have
control
over
what
order
they
come
out
in.
The
only
thing
you
have
control
over
is
that
you
can
page
it
right
in.
E
You're,
looking
right
at
you're
working
on
the
scrub
code,
the
scrub
code
relies
very
heavily
on
being
able
to
iterate
you
in
order
official
e,
because
every
O's,
when
you're
iterating
through
the
space
and
all
do
Steve's,
they
object,
need
to
come
out
in
the
same
order.
And
you
don't
want
to
be
doing
you
don't
want
to
have
to
you.
Don't,
for
instance,
want
to
have
to
read
the
entire
OS
Eve
store
to
find
the
next
H
object
for
the
poor
PJs
same
with.
Back,
though,
that's.
F
A
E
E
Let's,
let's
say
that
hypothetically,
it
might
be
possible
to
get
away
from
this,
but
other
components
are
simpler
because
viola
T
eat
this
complexity.
Does
that
make
make
such
this
in
exchange?
For
that
other
components
get
to
assume
that
objects
will
come
out
in
order
and
the
OS
DS
get
to
do
recovery
more
quickly?
I
mean
it's
comparatively
very
difficult
to
get
like.
Let's
say
you
have
two
different
collections
of
unsorted
objects,
possibly
different
versions.
How
do
you
find
out
whether
two
different
OA
C's
have
the
same
collection
of
objects?
E
B
B
E
E
E
There
are
well
and
there's
there's
some
other
stuff
working
against
this
too.
We
don't
really
expect
true,
collect
objects
that
truly
collide.
We
don't
expect
them
to
have
same
name
or
even
remotely
related
names.
E
We
do
expect
things
that
are
that
have
two
snapshots
of
the
same
object
to
differ
only
in
the
snapshot
field,
though
so
you
can
do
something
like
use
a
run
length
encoding
deal
where
the
first
time
you
encounter
a
particular
hash.
You
include
the
hash
of
the
key,
but
the
next
time
you
you
see
that
house
you
omit
up
and
so
on,
giving
you
a
sort
of
a
basic.
What's
that
going.
B
E
And
that's
that
sort
of.
Actually,
you
know
what
that's
a
pretty
good
medium
I
would
go
ahead
and
try
that
that
basically
gives
you
prefix
compression
for
free.
It's
not
exactly
optimal,
because
adjacent
adjacent
nodes
under
the
same
great-grand
current
probably
share
the
same
things.
But
even
then
you
can
probably
finesse
it
already.
B
E
Asking
is
random,
but
let's
say
you're
you're
far
enough
down
the
tree
so
that
you're
just
talking
about
collisions
on
a
particular
hash,
you
might
have
multiple
legit
or
you
might
have
multiple
sibling
nodes
that
have
the
same
hash
that
have
keys
that
have
the
same
hash.
But
it's
true
in
general
right.
You
might
have
adjacent
keys
that
have
the
same
ash,
namespace
name
but
differ
only
in
the
snapshot,
in
which
case
you
don't
have
to
include
any
of
that
Freedia
stuff,
just
the
snapshot.
B
But
it
because
inspiration
from
my
view,
any
questions.