►
From YouTube: 🖧 IPLD Bi-Weekly Sync 🙌🏽 2018-11-26
Description
A bi-weekly meeting to sync up on all IPLD related topics. It's open for everyone and recorded. https://github.com/ipfs/team-mgmt/issues/720
A
B
A
A
A
That's
the
one
of
the
steps
to
get
basically
dependent
of
the
ipfs
organization,
kind
of
and
being
our
own
project,
there's
still
some
things
thinking
to
activist,
but
probably
all
the
time
and
we'll
get
more
money,
pennant,
but
I
think
there's
no
rush!
Surely-
and
it
contains
all
the
details
about
this
meeting,
then
I
like
to
talk
about
the
Jas
API
specs
update
so
I
just
yet
and
again
think
in
the
agenda
I'm
just
two
issues
where
I
proposed
a
new
API,
but
it's
more
interesting
for
the
MJS
people.
A
A
C
Yeah,
this
isn't
so
much
of
an
update
from
last
week,
but
just
coming
into
this
week,
I'm
going
to
start
landing
a
bunch
of
UNIX,
FS,
B,
spec
stuff,
there's
a
bunch
of
different
PRS
for
different
parts
of
it,
and
just
the
sort
of
skeleton
will
be
landed
now
and
I'm
like
what
I'm
going
to
do
is
I'm,
going
to
start
on
any
open
discussions
that
we
still
have
going
will
turn
into
their
own
issues
or
their
own
PRS
and
will
actually
reference
them
in
the
review
from
spec.
A
So
should
we
do
a
quick
all
on
the
stuff
that
you
plan
to
work
on
the
next
two
weeks,
roughly
so
I
started
so
I
ever
worked
on
so
I
said:
did
the
API
draft
I
plotted
trend
to
F
planted
okay
with
them
a
bit
and
implement
them
on
the
current
stuffs
just
to
see
if
they
are
really
like
sound
or
if
they
actually
work
on?
Not
only
look
good
of
paper,
and
it's
also
connected
to
kind
of
graph
C,
because
it's
kind
of
is
the
high
level
API?
E
E
D
So
if
you
people
do
a
lot
of
comments
from
the
I
feel
a
little
bit
besides,
not
much
from
these
high
because
obvious
path,
though
they
like
one
thought
that
may
help
push
this
forward
is
like
this
would
happen,
liability
where
both
the
ideas
it
could
move
see.
Ids
from
the
ability
org
to
the
multi
for
mentored
and
basically
say
well
like
see
IDs
are
a
multi-format.
There
are
like
a
strong
dependency
of
IPL
d
in
practice.
This
makes
absolutely
no
difference.
It's
still
like
a
core
component
of
IPL
d.
D
F
Sure
my
quick
update
is
and
working
on,
go
IPL
d-prime
still
I
recently
gave
it's
immutable,
arrays
and
so
now
I'm
starting
to
write
some
tests
that
go
like
make
an
IPL
de
Graaff
out
of
the
mutable
interface
and
then
go
test
that
it
can
get
stuff
back
out
of
it.
It's
still
very
primitive,
but
I
hope
I'm
going
in
the
right
direction.
F
F
F
F
E
E
C
C
E
Thing
the
idea
is
just
I'm:
gonna
have
a
bunch
of
I
want
with
the
easiest
possible
thing,
which
is
the
graphs
are
represented
internally,
but
on
the
wire
all
I'd
on
the
wire
I,
just
sort
of
send,
updates
and
say:
here's,
the
previous
node,
here's,
my
parent,
here's
my
child
right
and
then
those
get
sent
over
the
wire
and
interpreted
I
didn't
I
I
figure.
There
must
be
something
going
on,
but
I
decided
to
just
make
a
protobuf
that
did
that
which
feels
like
the
wrong
direction.
C
I
mean
the
big
thing
would
be
to
you
see,
Boren's
that
have
put
up
up
there.
I
mean
that
was
a
little
bit
cleaner
and
important
that
the
new
direction,
that's
kind
of
the
primary
thing,
but
I
think
about
in
terms
of
how
you're
structuring
the
data,
it's
probably
relatively
similar.
The
only
thing
that
I
would
ask
is
like
how
are
you
deciding
what
to
send
over
the
wire
other
than
just
a
new
root?
C
E
So
I'm,
not
the
graphs
aren't
static
right.
The
idea
is
I'm,
just
sending
you
you
know,
I
I
wish
to
add
a
note
to
this.
I
wished
I
had
a
note
under
this
parent
and
then
there's
a
whole.
The
protocol
that
right
now
is
the
dump
protocols.
You
just
send
all
the
notes
right,
but
eventually
a
more
sophisticated
protocol
of
saying
which
knows
do
I
actually
need
to
send
you,
because
we
have
two
different
versions
of
this
graph.
E
C
Yeah,
so
so
we
have
a
new
replication
repository.
That's
that
I
peeled
these
flash
replication.
If
you
could
just
write
up
sort
of
like
this
use
cases
like
what
you're
trying
to
send
around
and
what
the
merges
look
like,
then
we
can
discuss
the
right
replication
strategy
and
document
sure
yeah
and
also
you
see.
B
B
Okay
I
was
just
go
ahead,
so
I
have
like
three
points
which
first
one
like
I'm
a
little
bit
surprised
or
increased
I.
Don't
really
understand
like
why
you
need
fast.
V2
is
kind
of
like
dragging
like
I.
Remember
that
it
was
like
a
huge
discussion
on
on
the
spec
then
Michael
like
over.
We
can
even
within
the
prototype
of
it
and
I.
Don't
know
right
now
like
why,
like
there's
like
three
people
assigned
to
UNIX,
it
has
to
be
like
drafting
this
back
in
multiple
card
sheets
and
I.
B
Think
the
conversation
is
not
happening
and
because
that
this
was
happening
like
it
is
just
getting
stuck,
then
the
the
other
two
things
and
I
was
just
like.
Oh
I'd,
the
tree
is
Oh
Jess
when
using
I
peel
the
we
do
ipfs
right
now.
It's
still
parabolic
I
quickly,
not
like.
Essentially,
we
cannot
write
because
the
way
that
kelp
sterilizes
things
into
HTTP
is
that
makes
everything,
JSON
and
so
I
kind
of
like
messes
up
with
like
buffers
and
so
on,
and
so
there
are.
B
C
C
B
C
Okay,
perfect
and
then
the
last
point,
so
you
were
asking
sort
of
like
what
is
the
holdup
on
unix
sp2
or
why
are
so
many
people
assigned
to
it?
I'm
not
entirely
sure
that
I
understand
all
the
context.
They're,
basically
like
the
spec,
was
stalled
out
for
a
while,
and
there
were
a
lot
of
unanswered
questions
and
then
we
sort
of
rebooted
it
a
little
bit
and
there
are
still
some
unanswered
threads
like
see
threads.
C
We
have
resolution
on
about
really
particular
points,
but
I
think
that
we're
ready
to
land
something
that
we
can
start
implementing
and
then
eventually
worry
about
some
of
those
other
issues
like
later
on,
for
instance,
there's
like
a
thread
about
right
now.
It
says
like
when
you're
encoding
the
chunk
data,
it's
always
a
link
to
the
binary
blob.
You
can't
do
it
in
line
and
there's
like
an
ask
for
smaller
files.
Can
we
just
inline
it,
and
if
we
don't
care
about
deduplication,
can
we
just
inline
it?
C
C
C
Last
point
is
that
we
are
currently
sort
of
pointing
to
a
hampton
plantation
that
is
not
drafted,
and
so
this
spec
will
be
in
front
of
that
and
look
we're
basically
pointing
to
an
implementation
of
it
that
is
used
in
POW
coin
and
go
that's
like
the
only
reference
that
we
have
for
that.
So
it
will
be
like
a
spec
that
points
to
a
non-existent
set,
so
that'll
be
a
little
bit
problematic
in
the
future.
I
think,
especially
for
the
jail
got.
F
You
yeah
+12
that
that
ham
thing
I
think
that
that
creates
a
lot
of
fear
for
me
when
I'm
avert.
Whenever
I'm
looking
at
things
in
this
area,
we
haven't
figured
out
how
to
specify
that
and
I'm
not
entirely
sure
we
figured
out
how
to
leave
the
right
shaped
hole
in
the
specification
for
either.
It's
just
was
really
scary.
D
F
A
D
Part
of
it
is
also
again
the
fact
that
we
did
try
to
come
up
with
solutions
head
and
it
stalled
because,
but
there
was
a
time
period
of
time
we
said:
okay,
finally,
Jason
visible
support,
binary
and
now
we've
gone
back
and
incited
it
well,
actually,
maybe
Jason
will
support
binary,
so
we
have
to
like
actually
just
something
someone
needs
to
own
it
and
change
this
back,
and
it's
not
that
difficult.
But
we
need
to
people
get
people
to
agree
on
it.
Wait.
C
C
D
D
C
D
C
D
So
we
have
this
block
if
you
guys
attends
around
the
blocks,
but
doesn't
get
resolution
stuff
like
that,
and
it's
kind
of
annoying
you
could
be
kind
of
nice
to
just
modify
the
current
die,
get
the
I
to
send
raw
data
I
wonder
if
we
could
add
an
encoding
cuz
like
the
way
we
currently
do.
This
encoding
things
to.
D
Like
encode
it
using
JSON,
but
we
really
mean
it.
They
kind
of
keeping
it's
like.
The
encoding
is
listless
if
your
response
encoding,
but
we're
actually
doing,
is
telling
to
convert
and
I
feel
the
object
to
something
else,
which
is
why
it
doesn't
make
that
much
sense
in
the
first
place,
I
guess:
well,
it's
have
to
sit
down
and
talk
to
the
JSP
boy,
like
you
figure
out
the
right
command
structure.
For
that,
because
I
think
the
current
documents
are
kind
of
like
broken
yep,
yeah
yeah.
C
So
typically,
so
this
is
with
an
HTTP
API
right,
yeah.
D
B
D
You
can't
the
current.
The
current
API
expects
you
to
send
like
it's
actually
a
bit
problematic,
because
errors
are
indistinguishable
from
the
parasites
to
use
be
insignificant,
non
errors.
Now
it's
slightly
less
bad,
but
we
haven't
like
charged
compatibility
reasons.
The
old
clients
still
like
can
still
make
very
still
confused
errors
and
not
errors.
B
We're
coming
to
saying
the
comments
like
so
this
is
like
one
of
the
interface
issues
that
I
was
trying
to
bring
up
earlier.
The
conversation
has
been
going
for
two
years
now
and
and
and
so
it
basically,
it
stumbles
upon.
That's
like
he's
having
when
he
says
it's
an
EVP
that
sounds
really
HTTP,
API
and
I
go
just
like
seized,
HTTP
sockets,
as
I
can
t
see,
P
socket
and
I
realized
everything
to
Jason
I.
Just
like
sends
it
true,
and
you
know
what
that
means
like
it
just
like.
B
If
there
is
a
buffer
or
any
special
like
pipe
like
you,
just
just
compares
it
to
Jason
and
just
like
it's
cruising
up
and
Ike
doesn't
give
any
other
information
for
that
probe
of.
We
have
been
able
to
survive
with
that,
because
we
know
the
schema
over
that
go
robot,
so
we
would
always
cast
the
data
key
to
like
a
buffer
and
like
it
would
work
right
like,
but
like
we
see
one,
we
don't
excuse
anymore.
So
we
cannot
like
yes
into
like
there
was
like
a
lot
of
like.
E
B
If
you
have
a
taxi
board
nailed
it
has
like
whatever
inside.
If
you
give
me
that
wrong
thing,
then
I
condition-wise
it
locally,
it's
kind
of
like
last
mile
resolve
or
that's
like
the
term.
We
use
at
a
time
and
they're
like
that,
preserves
the
the
types
and
like
yeah,
like
it's
kind
of
like
weird,
that,
like
sometimes
you
will
get
the
wrong
route.
Sometimes
you
will
get
the
property
there.
Actually
I.
Think
there's
like
some
really
like
still
some
rough
edges
to
polish,
but
it's
something
that
like
implemented
today,
will
enable
Jess
ipfs
API.
D
It
could
be
consider
breaking
dad
get
simply
because
it's
already
said
broken,
or
would
that
not
be
a
solution
like
if
it
could
do
that,
then
we
can
always
wrap
it
in
adjacent
object
and
like
have
a
base64
encoded
field
with
whatever
size
it
has
the
data
and
then
another
one
that
has
like
the
Eid
prefix.
So
that
would
give
us
all
the
information
we
need
and
it
was
very
consistent.
The
other
idea
this
is
to
use
the
Bach
if
you're
either
force
of
walk
API.
As
far
as
I
know
can't
do
resolution.
B
B
D
B
B
A
Okay,
we
are
almost
out
of
time
or
we're
out
of
time.
I
just
want
to
add
is
I.
Currently
we
say
if
we
talk
about
the
HTTP
API,
we
talk
about
its
check.
The
responses
are
JSON
encoding,
but
the
big
question
for
me
is
that
doesn't
always
need
to
be
chased
on.
Just
like
a
Stephen
basically
says
is
we
could,
just
as
he
send
over
C
Bohr
and,
for
example,
as
to
be
reasonable.
A
D
Actually,
we
could
send
back
whatever
we
want.
We
have.
Can
someone
mean
I'm
gonna?
Be
back?
Thank
you.
Sorry,
sorry
to
come
straight,
so
we
actually
have
an
encoding
parameter,
so
you
can
say
like
in
code
using
part
about
the
code
using
c
p--
or
whatever,
but
home
assembly
and
music.
I
need
to
know
the
encoding
of
the
thing
I'm
expecting
to
actually
tell
it
like
not
like
change
it
at
all.
D
Yeah,
so
my
idea
was
actually
just
just
used
Jason
because
it'll
of
send
back
other
data
problems,
we
just
send
back
raw
data.
We
can't
send
back
additional
metadata
easily.
We
could
put
in
the
headers
that
the
current
inspector-
sorry,
our
current
HT
API
spec-
allow
for
that.
But
if
we
said
back
an
actual
like
Jason
object
with
like
some
field
that
just
had
the
raw
object
encoded
in
it
and
it
would
work
there
may
be
other
ways
of
doing
this
type.
Just
don't
know.
A
Yeah,
it's
basically
like
those
odd
details.
I
would
say
so.
That's
like
that's
totally
fine
as
well
and
make
sense
to
have
basically
drift
in
a
JSON
object
and
then
we'd
be
hundreds
of
space
sleep
anyway.
All
right
is
there
anything
else
because
to
be
already
running
out
of
time,
I
mean
it's
over.
I
have.
A
F
Make
it
quick,
you
asked
why
stuff
slow
and
it
seems
to
me
we
don't
actually
have
a
ton
of
people
who
are
full-time
on
IP
LD
and
it
would
be
really
nice
to
get
more.
It's
like,
according
to
the
capacity
planner
anyway,
it's
Michael,
Volker
me
and
that's
it
and
Michael
is
part-time
and,
despite
that
doing,
an
awesome
job
of
like
trying
to
push
our
specs
through,
but
that's
one
of
each
of
us
for
the
two
languages
here.
F
D
Well,
that's
why
they're
blank,
because
I
try
to
get
myself
dragged
back,
but
at
the
moment
I've
been
dragging
on
directions.
I
think.
A
key
thing
here
is
better
coordination
that
we
notice
the
same
thing
on
the
go.
Express
team
we're
like
if
we're
all
at
coordinating
talking
a
lot
like
stuff
just
gets
done.
It
happens,
but
whatever
it's
kind
of
off
in
their
old
world,
it
can
be
very
hard
to
like
make
progress,
because
you
know
don't
know,
what's
happening.
It's.
C
B
B
Could
this
is
like
a
very
large
community,
like
we
have
like
even
like
they'll
wait
at
the
LV,
and
so
on
was
people
that
were
even
like
more
overbooked
like
we
had
way
less
capacity,
and
we
have
like
an
amazing
community
of
contributors
contributing
to
this
crash
and
doing
the
research
participating
on
these
trends,
and
so
that's
why
quick?
Why
also
encourage
it?
B
Oh
yeah
I
check,
you
know,
I'm
checking
with
all
these
people
shaking
all
these
threads
from
the
past,
because
we
might
have
now
solved
problems
that,
when
I
waited
in
the
past,
that
now
would
encourage
people
to
like
come
back
and
continue
the
conversation
and
so
yeah
I
guess
an
issue
and
I
killed
his
like
I,
be
if
I
just
like
points
to
the
notes.
I
actually
spent
some
time
on
that
weekend,
just
like
labeling
I
peel.