►
From YouTube: 🖧 IPLD weekly Sync 🙌🏽 2019-10-28
Description
A weekly meeting to sync up on all IPLD (https://ipld.io) related topics. It's open for everyone and recorded. https://github.com/ipld/team-mgmt
A
A
So
I
worked
on
getting
IPL
the
rest,
I
only
work
on
surgery,
and
so
the
sad
news
is
it
looks
like
we
can't
do
assert
it,
because
it
would
need
an
upstream
change
and
although
the
author
of
C
bore
so
the
sri
siva
stuff,
isn't
favorite
and
the
maintainer
is
not
so
I
think
the
comment
I
think
in
the
in
the
notes
and
it's
it
looks
like
it's
not
going
to
happen.
I
first
posed
another
comment,
so
I'm
still
fighting
for
it,
but
it
looks
like
it
won't
happen.
A
Yeah,
I,
I,
totally,
don't
agree
with
this
decision,
and
but
there's
not
much
I
can
do
I
guess
so
for
now,
I
would
probably
just
yeah
I'm.
So
there
is
already
a
rest.
I
pulled
the
implementation
by
someone
else,
which
is
not
using
survey
and
I
probably
just
use
this
for
now.
So
we
can
just
keep
going
and
then
yeah
see
what
the
future
brings.
But
that's
yeah
kind
of
unexpected,
so
I
expected
I
can
convince
the
maintainer
but
I
can't
but
we'll
see.
B
A
Yes,
I
read
so
busy
I
also
told
them.
I
totally
agree
that
if
you
have
a
library
you
don't
accept
like.
If
someone
comes
in
and
says
I
wanted
to
do,
X
Y,
you
totally
don't
have
in
this
case.
It's
really
let
there
is
basically
an
agreed-upon
heck
to
do
to
add
text
to
to
survey,
and
this
is
basic.
What
I
did
for
the
zero
stuff
and
then
this
is
already
implemented
in
Tamil
and
in
message
page.
So
basically,
three
different
authors
use
the
agreed-upon
Hank
to
work
around.
A
Yes,
so
let's
the
Qin
state,
yeah
I,
keep
you
updated
if
anything
changes
but
I,
don't
really
have
it.
Yeah
I,
don't
see
big
changes,
they
did
not
change,
but
this
also
good
news
and
because
totally
different
thing
am
I.
Like
a
few
months
ago,
I
was
working
on
the
stack
prototype
on
the
stack
spec
specification
and
I
found
out
that
they
are
doing
something
with
checksums
and
they
wanted
to
do.
It
basically
have
clearly
appears
with
the
hedge.
A
They
are
using
it
in
the
check
sum
and
I
said
well,
you
could
use
multi
hi
Floyd
and
they
said
great
idea.
That's
what
we're
gonna
do
and
so
I'm
gonna
get
to
write
quickly,
a
spec
for
them
to
use
multi
hash,
it's
digital
mixer
tension
and
the
good
news
is
so.
This
is
a
specification
which
is
currently
done
by
independent
people,
but
it
will
go
to
the
OGC.
So
the
urgency
is
the
matrix
thing
as
for
refer
to
your
standards
and
often
also
or
she
sees
ten
has
become
ISO
standards.
A
B
That's
just
been
a
niche.
That's
been
needing
to
be
scratched
so
currently
working
through
that
I've
got
introductory
material
like
the
material
that
I've
got
up
so
far
is
okay,
I've
shown
up
at
this
thing.
What
is
it
and
then
this
would
help
you
get
that
in
your
head.
What
do
you?
What
is
this
thing?
And
then
it's
I'm
I
have
a
version
on
my
machine
that
is
restructuring
a
little
bit
of
the
extra
the
additional
content
to
guide
through
more
advanced
materials,
and
that's
where
I'm,
that's.
B
B
No,
it's
jeaious,
I,
peeled
easy
cache
is,
is
busted
because
the
upstream
decoder,
basically
no
one
in
the
JavaScript
implementations,
have
been
abandoned
because
they
were
all
just
Forks
of
Bitcoin,
with
some
hacks
to
make
it
work
for
Z,
cache
and
and
no
one's
really
kept
them
up-to-date
or
maintained.
So
that's
left,
as
he
cache
see,
DAG
thing
whatever.
It
is:
no!
It's
national
bag.
B
So
the
ipfs
needs
to
make
this
optional,
because
it's
something
that
doesn't
need
to
ship
to
people
mostly
unless
they
really
want
to
use
the
cache.
But
I
wrote
an
implementation
a
while
back
a
block
decoder
and
I
got
that
hooked
up
last
week
so
that
it
can,
it
replaces
what's
there
and
it's
does
it
have
any
dependencies,
I
don't
even
know
if
it's
got
any
dependencies,
so
it's
not
any
risk
there
of
it
getting
ahead
of
date
and
having
dependency
security
problems.
B
C
Well,
I've
had
a
fun
week
again,
as
the
last
couple
of
weeks
have
been.
I
have
just
learned
so
many
things
about
the
go
compiler
and
about
how
performance
works.
So
a
couple
pieces
of
small
good
news
in
case
we
have
anybody
else.
Try
to
join
us
on
this
I've
learned
a
bunch
of
new
things
about
getting
the
go
compiler
to
tell
me
very
clearly
how
it
feels.
So.
There's
like
this
one
GC
Flags
incantation
that
I
discovered
this
week.
C
C
It's
also
wildly
undocumented,
so
GC
flags
M
gives
you
some
short
information
about
what
the
compiler
thinks
about
escape
analysis
and
where
memory
allocation
should
go
like
keeper
stack
and
so
on
and
GC
flags
quote
it
M
M
again
does
something
completely
different.
Well,
okay!
No,
it
does
the
same
thing,
but
it
does
it
with
enough
information
to
actually
be
useful.
The
single
M
gives
you
like.
It
just
says
a
bunch
of
things
that
are
facts,
but
it
doesn't
give
you
any
understanding
of
how
the
compiler
came
to
these
conclusions.
C
C
Yes,
we
want
to
use
it
embedded
strokes
most
of
the
time,
because
that's
going
to
allow
us
to
amortize
down
the
number
of
distinct
memory
allocations
that
we
use
so
I
mean
that
was
kind
of
known,
but
I've
learned
a
couple
more
things.
On
top
of
that
which
make
the
whole
thing
work
much
sooner
than
this
by
the
fact
we
want
to
use,
embedded
structs
so
not
pointers.
C
C
C
So
some
of
the
best
abstraction
ideas
I've
come
up
with
now
is
lots
of
embedded
struggles
and
and
doing
kind
of
a
similar
thing.
On
the
note
builder
side,
if
we
have
a
node
builder
implementation
that
actually
has
a
bunch
of
reserved
space
inside
of
it
for
the
node
builder
types
that
we
know
we'll
need
for
the
children,
then
we
can
have.
You
know
the
node
builder
for
one
struct
can,
when
it's
asked
for
the
node
builder,
for
it
the
member
types
inside
of
it.
C
C
C
Don't
know
if
I
want
to
just
read
this
whole
thing
aloud.
I
would
probably
have
some
more
writing
to
do
about
like
what
exactly
the
limits
as
engineering
time
approaches.
Infinity
are
here,
but
I
think.
Basically,
it
turns
out
there's
still
a
lot
of
trade-offs
where,
for
the
most
part,
we
want
to
embed
structures
and
not
use
pointers
in
order
to
amortize
down
the
a
lock
account
that
much
we
already
knew,
but
then,
when
we
start
looking
at
the
ways
in
which
people
might
use
the
data
structures
that
we've
unfolded
into
trees.
C
If
somebody
wants
to
reshuffle
around
a
bunch
of
the
interior
of
the
tree,
then
they
probably
start
to
wish
that
there
were
pointers
around
again
in
whatever
specific
parts
of
the
tree
they're
reshuffling,
because
we
don't
want
pointers
everywhere.
We
failed
at
memorizing
analogs.
We
do
want
pointers
wherever
you're
about
to
do
a
shuffle
right,
and
so
it
pretty
much
turns
out
that
wherever
you
want,
the
pointers
to
be
is
determined
by
the
transformations
that
you're
about
to
do.
C
But
this
isn't
actually
information
that
we
can
use
another
way
that
we're
designing
our
code
generation
right
now,
we're
just
designing
code
gem.
It
spits
out
helpers
to
do
a
bunch
of
tree
manipulation
in
general
and
if
we
don't
know
exactly
the
parts
of
the
tree
that
the
user
is
about
to
write
further
code
to
manipulate,
we
just
don't
have
that
information
at
the
right
point
in
our
pipeline.
In
order
to
use
it.
C
C
We
were
produced
a
whole
new
programming
language
and
probably
something
that's
built
on.
Like
the
category
theory,
recursion
schemes,
morphisms
sort
of
stuff
got
a
highly
histah
whoo,
and
if
we
had
a
programming
language
like
that,
then
we
could
determine
where
we
want
all
the
pointers
and
stuff
to
be
able
to
do
the
optimal
truffles
that
exactly
match
the
actual
logic
that
we
want
to
perform,
but
that
programming
language
doesn't
exist
right
now.
So
we'll
probably
don't
want
to
build
it.
So.
C
That's
frustrating
but
I
I
guess
I
appreciate
having
this
information
available
now,
because
now
we
can
definitely
know
about
whatever
solutions
we
pick
are
going
to
be
compromises
and
I.
Guess
that
at
least
makes
me
a
lot
more
confident
that
we
can
move
forward
with
a
side
compromises
so
yeah,
fun
and
one
other
piece
of
good
news.
C
Hana
started
using
some
of
the
cut
of
Jen's
stuff
to
do
some
work
and
filled
in
a
bunch
more
of
stuff
for
like
cranking
out
integer
types
and
stuff,
and
so
we
have
a
little
bit
of
usage
exercise
their
deformity,
like
feedback
loop
to
make
sure
we're
not
screwing
up.
That's
really
exciting,
that's
it!
That
was
my
week.
B
No
I
am
used
to
call.
Apparently
someone
is
thinking
at
somebody
with
that
much
background
in
what
we're
doing
listening
into
this
might
think.
That
sounds
like
some
really
intense,
premature,
optimization,
getting
right
and
down
to
that
level
for
something
that
doesn't
even
do
anything
yet
so
so
to
clarify.
B
C
Bit
a
little
bit
more
than
I
was
like
it's
determining
where
pointers
are
ending
up
in
all
the
maybe
types
and
a
bunch
of
stuff
like
that
and
and
those
are
unfortunately
things
that
do
leak
all
the
way
through
to
the
user
facing
API
I
wish.
There
was
a
way
to
make
them
not
do
that,
but
there
is
not
side
note.
C
One
of
the
other
things
I
considered
is
hiding
another
whole
set
of
stuff
under
under
interfaces
in
the
go
code
and
being
able
to
switch
out
more
of
where
pointers
are
and
are
not,
then
it
turns
out.
If
you
do
this,
then
you
lose
the
ability
for
the
compiler
to
inline
a
bunch
of
methods
and
I'm
fairly,
certain
that
would,
in
the
details,
turn
out
to
be
a
huge
mistake.
So
putting
that
aside,
the
other
place
it
shows
up
is
oh
right.
C
So
I
was
kind
of
frustrated
to
discover
this
one,
but
I'm
seriously,
considering
an
alternative
to
the
note
builder
abstraction,
because
the
note
builder
abstraction
kind
of
builds
stuff
up
from
the
inside,
and
so
you
build
a
thing
of
any
wrapped
around
this
keep
going
out,
and
that
is
the
opposite
of
what
I
want.
If
I'm
trying
to
do
my
memory
allocation
and
one
hunk
first
and
then
fill
it
in
as
I
go
they're
completely
opposed
directions
are
flow.
E
E
C
If
you
want
to
come
Splunk
performance
with
me,
I'm
kind
of
having
the
ball
over
here,
wait:
I'm,
I'm,
sorry
just
write
shell
scripts
around
the
compiler
to
like
dump
some
benchmark,
I'm
gonna
be
like.
Does
this
have
a
feature?
Yes
or
no?
And
then
it's
that's
one
way
to
get
hard
answers
about
this
stuff.
I'd.
A
No
I
thought:
okay,
yes,
as
this
civilizations
that
was
complicated
as
I
mentioned,
I
guess
like
just
getting
it
work.
Somehow
shouldn't
be
a
big
deal
like
we
speak.
Getting
the
data.
Someone
who
seaboard
bank
shouldn't
get
beat
you.
The
point
was
I
wanted
to
use
surname,
but
if
I
don't
use
it,
it
shouldn't
be
like
yeah
hydroid,
so
so
so
anyway,
it's
all
bits.
So
maybe
what
I'm
saying
is
the
silver
sages
stuff?
Wouldn't
shoot,
wouldn't
and
shouldn't
be
a
blocker
on
any
other
work.
I
guess
so.
A
Yeah
but
I
actually
have
any
plans-
okay,
so
yeah
so
I
having
planned
in
much
detail
in
my
next
steps,
but
the
idea
was
to
so
we
photo
into
selectors
probably
is
to
look
like,
so
the
plan
is
to
look
into
the
Prosecco
system.
What
we
currently
have
and
what's
out
there
like
the
whole
Mikey
for
my
stuff,
he
what
what
we
truly
have
for
this
order
to
be
there
and
then
build
the
missing
low-level
pieces
and
then
go
into
those
things.
The
basis
I
want
to
do
this
to
look.
C
This
stuff,
I'm
gonna,
be
really
interested
in
when
you
get
to
it
is
I,
have
a
feeling
that
you're
gonna
discover
most
of
the
same
stuff
that
I'm
pulling
my
hair
out
right
now
about
like
this
memory,
Mart
ization,
stuff
and
rust
is
probably
gonna,
throw
it
in
your
face,
because
it's
rust
and
so
I'm
just
I
really
excited
to
see
how
that
shapes
up.
Yeah
cuz
I
have
good
lessons
for
me
over
here
in
another
language,
hopefully
yeah.
A
Yeah
so
I'm
still
like
I,
am
a
big
and
concerned
that
my
skills
are
not
good
enough
to
find
the
most
optimal
solution,
but
we'll
see
so
it's
like
I'm
still
discovering
like
like
the
whole
study
stuff,
was
certainly
like
a
good
step
to
learn
the
deeper
into
stuff,
so
I
mean
I
will
see
it's
amazing.
What
bus
can
do
in
like
how
me
like
you,
can
do
so
many
things
in
so
many
different
ways,
so
I'm
pretty
excited
to
see
like
how
we
can
get
I
produce
stuff
working.
A
C
One
more
thing
that
I
didn't
write
down,
but
we
I
think
we've
talked
about
this
a
little
bit
before
in
some
place
or
another,
but
we
should
put
some
more
research
points
for
all
of
us
into
recursion
schemes
and
that
sort
of
stuff
cuz
a
much
shorter
way
to
describe
some
of
my
update
this
week
is
I
basically
discovered.
Oh
I
want
this
data
record
kind
of
abstraction
for
anamorphisms
and
for
almost
any
other
kind
of
more
complicated
morphism,
then
I
need
a
totally
different
abstraction
that
has
different
memory
problems.