►
From YouTube: 🖧 IPLD weekly Sync 🙌🏽 2021-03-08
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
What
I
did
is
I
improved
again,
the
p
the
code
for
js
ipfc,
ipfs,
unix
fs.
A
So
that's
a
super
simple
interface
and
I
hope
that
also
other
parts
of
ipfs
can
then
just
use
this
thing
instead
of
like
a
full
block,
service
or
ipld
and
of
course
next
is
to
keep
integrating
it
into
j
into
js
ipfs.
I'm
almost
there
I
have
about.
I
don't
know
something
like
six
six
tests
that
don't
run
well
and
like
we
have
something
like
like
300
tests.
A
Also
so
it's
like
getting
there,
but
no
now
they
are
probably
harder
to
beat
debug,
and
the
idea
is
really
that
the
from
the
ipfs
side
of
things
you
still
interact
with
the
old
cids
or
basically
the
cds
you
are
used
to
and
then
on
the
lower
levels
they
get
converted
and,
for
example,
unix
fps
will
fully
use
the
new
cid
stuff,
but
on
the
outside
servers
it
will
still
be
the
outside.
So
if
you
use
js
ipfs,
it
will
just
work
as
as
usual.
That's
at
least
like
for
start.
A
B
Cool,
so
something
I
mentioned
last
week
is
that
I
I've
been
moving
us
to
the
codec
deck
pp
package
that
rod
wrote
a
while
ago,
which
is
the
dac
baby
codec
on
top
of
ipld
prime,
because
some
libraries
in
ipfs
and
ipld
already
use
ipld
prime
sort
of,
but
they
use
a
different
dac.
Bb
codec
called
go
ipld
prime
proto,
and
they
both
are
kind
of
compatible,
but
they
are
not.
B
They
can't
be
used
at
the
same
time,
so
we
have
to
choose
one
and
use
that
so
the
good
news
is
that
I
I
converted
gomerkaldag
gocar
and
gograph
sync,
which
were
the
only
pieces
of
existing
software
that
used
the
other
codec,
which
we
can't
use
at
the
same
time
and
that
all
worked
fine.
They
don't
have
a
lot
of
tests.
There
was
only
one
test
that
failed,
but
I
figured
I
can
you
know,
can
check
that
out
later,
because
we're
not
merging
code
just
yet
and
what
I'm
doing
now
is
wrangling.
B
All
of
that,
together
into
go
ipfs
because
hana
does
have
a
branch
in
go
ipfs
that
switches
to
the
go
merkle
dag
that
uses
go
ipld,
prime
and
she
has
four
failing
tests,
but
everything
else
works.
So
I
added
my
stuff,
which
is
moving
everything
to
rods,
codec
and
more
stuff
fails
and
I'm
trying
to
debug
that
so
yeah.
I
still
don't
know,
what's
what's
failing
or
why
it
might
be
a
mistake,
I've
made.
It
might
be
something
that
I
misunderstood.
B
It's
kind
of
weird,
because
I
get
the
same
cids
I
added
debug
prints
in
the
codec.
Nothing
is
failing.
There
are
no
errors,
so
I
still
have
no
clue
why
tests
are
failing
and
the
tests
that
are
failing
are
giving
me
very
useful
error
messages
like
failure,
nil
or
like
failure
not
found
stuff
like
that,
so
it's
very
fun
to
debug
and
yeah.
That's
that's
me.
I
I
wish
I
could
tell
you
what's
wrong,
but
I
think
it's
going
to
take
me
like
another
day
or
so
to
figure
out
what's
wrong.
A
C
So
I'm
still
in
the
trenches
on
trying
to
upstream
some
stuff
from
a
couple
weeks
ago
already
into
go
multi-hash
and
surely
it's
close
to
landing
now
we're
going
to
have
a
nice
plug-in-ish
registration
e
system,
so
you
can
add
multi-hashes
to
the
multi-hash
library
without
having
to
like
patch
the
library
itself
and
you'll
be
able
to
get
hashes
out
and
they'll
stream
and
all
this
stuff,
and
I'm
pretty
sure
I
talked
about
the
why
in
the
last
week's
recording,
so
I
won't
do
that
again,
but
there's
a
pr
we've
fought
through
what
seems
like
hundreds
of
small
details
in
that
vr
already,
and
I
think,
there's
four
left
and
surely
they
will
be
done
in
a
day
or
so
after
that
the
lynx
system,
pr
is,
is
still
out,
and
it's
pretty
much
just
chilling
out
waiting
for
that
multi-hash
stuff.
C
C
So
basically,
if
you're
reading
things,
sometimes
you
have
a
storage
pool
where,
if
you
have
total
control
of
the
storage
pool-
and
you
trust
that
you're,
the
only
one
that
can
touch
it
and
so
on
and
so
forth,
and
you
put
data
into
that
storage
pool
and
you
hash
that
time
then,
should
we
be
able
to
read
that
out
and
de-serialize
it
and
not
bother
hashing
it?
Maybe
sometimes
that's
actually
reasonable
right
now
we
just
don't
have
an
api
for
that.
C
So
the
question
is:
where
should
we
put
that
api
there's
more
than
one
possible
thought
on
that
and
there's
a
pr
with
one
option
on
it
and
there's
some
other
discussion
on
that
pr
other
ways
we
could
do
it
so
if
anyone's
interested
in
that
you
can
take
a
peek
and
yeah
chris
is
cool
yeah.
I
think
to
give
a
quick
rundown
of
what
the
options
are.
C
Is
we're
considering
having
like
one
big
flag
for
the
entire
link
system
like
okay,
if
you're
dealing
with
this
whole
thing,
then
don't
check
things
or
if
we
should
complicate
the
api
of
the
storage
functions
to
let
them
say
whether
they
should
be
trusted.
It's
just
kind
of
deciding
like
who
should
have
this
responsibility.
D
Cool
I
got
busy
with
a
bunch
of
filecoin
issues
debugging
last
week,
so
not
quite
as
much,
but
I
figured
I
would
update
some
of
the
other
ipld
related
things
that
they
were
both
us.
Looking
at
ipld
and
ipfs
got
through
last
week,
there's
a
ipld
prime
git
codec.
D
The
interfaces
that
are
coming
for
thinking
about
what
fetching
ipfs
objects
in
a
ipld
prime
native
dialect
looks
like
is
that
interface
refined
some
last
week.
You
can
take
a
look
at
what
it
is
now,
but
I
think
it's
actually
a
much
more
easy
to
use
ergonomic
interface
yeah.
It
just
has
a
callback
instead
of
two
channels,
and
this
makes
life
somewhat
easier
in
terms
of
trying
to
understand
who's
in
charge
of
closing
channels,
and
why
are
there
two
channels
and
etc?
D
The
other
exciting
thing
I
think
it
was
like
on
friday,
hannah
was
able
to
get
together
a
go
unix,
fs,
node
ipl,
the
prime
representation
adl
of
what
a
unix
fs
node
looks
like,
which
is
exciting,
because
that
gets
us
another
step
towards
having
sort
of
a
bunch
of
these
mutation
type
things
or
this
is
especially
used
for
pathing,
because
that's
the
the
view
of
your
data
that
you
want
when
you're
trying
to
do
various
traversal,
empathic
style
things,
so
things
are
falling
into
place
there.
D
Yeah,
okay,
so
so
the
actual
data
is,
in
some
sense,
dag
tv
data
right,
so
you've
got
these
things
with
links,
but
when
we
talk
about
the
paths
that
we
want,
the
paths
that
we
have
are
going
to
be
based
off
of
ipfs,
sorry
of
unix
fs
pads
and
which
thinks
of
them
as
maps
with
like
name
to
sub
child
of
directory
right
and
if
we
just
have
dag
tb
data,
which
is
probably
the
codec
that
we
initially
would
want
to
load
this
data
out
of
blocks,
as
we
don't
know
how
to
path
over
it.
D
That
way,
so
we
need
to
interpret
it
as
unixfs
and
what
we
currently
have
in
the
ipld
format
version
of
ipfs
is
we
have
this
sort
of
extension,
where
we
take
our
sort
of
view
of
a
dag,
pv
protobuf,
and
we
have
this
other
protobuf
thing
that
we
that
that
we
call
a
file
system
node.
That
is
the
unix
fs
sort
of
view,
and
then
we
have
this
custom
sort
of
pathing
logic.
That's
like!
Oh,
when
you
want
to
think
of
me
as
the
map.
D
I
will
expose
my
directories
if
I'm
a
directory
and
sort
of
looks
in
it
and
does
some
special
pacing
stuff,
so
we
need
some
equivalent
of
that
in
ipld
prime
to
be
able
to
still
have
the
ability
to
do
this
seemingly
natural
thing
of
walking
a
directory
tree,
which
is
the
thing
that
we
do
all
the
time,
and
so
the
question
is:
how
does
that
work
right
and
I
think
we're
still
figuring
out
exactly
how
that
works,
because
we've
got
you
know
the
dag
pb
node?
D
That
is
the
thing
that
we
are
loading
out
of
disk,
but
then
we
also
have
this
unix
fs
node.
That
is
the
thing
we
want
to
interact
with
for
pathing
style
things,
and
so
is
the
deal
that
we're
going
to
do
an
adl
like
thing:
are
we
going
to
just
sort
of
wrap
it
when
we
do
the
pathing
and
get
the
back?
Are
we
going
to
have
one
thing
that
is
some
horrible
hybrid
amalgamation
of
both
of
these
things
or
something
else,
so
that
repo
is
a
first
pass?
D
C
We've
also
had
some
really
outlandish
ideas
in
the
last
week.
Okay,
I'm
calling
it
outlandish,
it's
my
idea-
and
I
think
it's
outlandish
so
so
there
are
we'll
just
talked
about
how
we're
probably
we're
looking
at
using
the
adl
concept,
to
redefine
the
way
that
we're
doing
this
unix
best
stuff
in
a
much
simpler
and
more
coherent
way.
We
hope
so.
C
C
And
in
taking
care
of
those
things,
it
does
a
bunch
of
the
work,
but
something
else
that's
interesting
about
unix
fs
is
that
it's
got
when
you're
walking
over
the
directories.
You've
got
these
hemp
like
things
and
then
you've
also
got
a
bunch
of
attributes
and
then
you've
got
the
jump
over
the
next
content.
C
So
that
means
actually,
when
you're
doing
the
pathing
in
ipfs,
like
over
the
wherever
your
address
is
slash,
ipfs
files
path.
Here
thing:
if
you
chunk
that
all
up
each
one
of
those
segments
is
doing
several
things,
it's
jumping
up
through
the
hand,
and
so
that's
something
we
might
do
with
this
adl
and
it's
doing
the
additional
like
dag,
pb,
semantic
translation
thing,
and
then
it's
stepping
through
the
little
attribute
structure
and
then
it's
going
forward.
C
A
Thanks
next
one
is
rock.
F
One
simple:
I
I
I'm
trying
to
help
get
some.
F
This
is
ultimately
about
getting
some
typescript
network
landed
in
the
jsi,
bfs,
stuff
and
and
part
of
that
volca
is
doing
some
integration
of
our
newer
ipld
javascript
components
over
there
and
there's
there's
just
stuff
going
on
with
that
sort
of
in
that
area
and
and
it
means
consuming
some
of
our
newer
pieces.
F
But
there's
compatibility
issues
and-
and
we
we
knew
this
would
be
a
problem
because
it's
it's
part
of
this
es
modules.
Migration,
that
the
javascript
ecosystem
is
undergoing,
where
we
have
tried
with
our
new
stuff
to
straddle
that
line,
much
more
intentionally
and
be
much
more
future
looking
with
it.
But
but
we're
still
in
the
middle
of
pain
with
it.
F
And
so
we've
gone
one
path,
but
the
ipfs
team,
with
the
edgier
tools
that
they
use
for
all
the
all
the
things
they're,
testing
and
bundling,
and
everything
have
chosen
to
use
this
tool
called
es,
build
which
is
a
go,
javascript,
compile
thing
and
it
doesn't
support
export
maps
for
ds
modules
and
so
and
we
rely
on
export
maps
to
solve
all
our
problems
in
the
es
modules
area.
F
So
so
our
new
staff
doesn't
work
with
like
the
rest
of
the
pl
javascript
stack,
so
we're
just
trying
to
address
that,
and
and
and
it's
just
it's
just
a
big
whack-a-mole
game,
and
we
knew
this
would
be
the
case
because
you
fix
it
for
one
bundler
and
then
another
bundle
of
breaks.
F
But-
and
so
I
thought
I
had
a
solution
last
week-
merged
in
yesterday
tried
it
out
and
it
wasn't.
A
bundle
of
the
broker
was
actually
the
typescript
compiler
that
broke
so
fixed
one
thing
and
the
typescript
compiler
broke.
But
I
think
we've
got
a
path
to
well
fixing
that
one.
But
who
knows
what
else
will
pop
up?
So
it
might
be
a
case
of
we'll
get
this
fixed
and
then
someone
will
report
hey.
F
This
is
now
broken
in
my
who
knows
what
tool
chain
that
they're
using
it's
yeah,
it's
a
bit
of
chaos,
but
we'll
move
forward
internally
and
see
what
breaks
out
accidentally.
That's
all.
D
A
Yeah,
so
one
one
comment:
I
I
forgot
to
post
on
the
issue
that
if
you
want
someone
to
test
it
with
common.js
because
they
got
like
like
as
I
could
do,
js
ffs
work.
I
can
test
it
because
I
actually
do
those
things.
So
I
want
to
do
it
today
but
forgot
about
it,
but
I
would
really.
F
Gonna,
I
think,
because
the
way
michael
set
up
publishing
for
these
things,
I
can't
actually
publish
pre-releases
for
them,
because
michael
has
the
keys
and
everything's
published
through
github
actions
and
there's
no
way
to
do
like
a
pre-release.
But
I
think
what
I'll
do
is
I'll
get
I'll,
get
a
solution
that
I
think
works
actually
just
published
of
js
multi
formats
and
then
throw
it
over
to
you
to
test
on
common
js,
because
that
is
the
missing
piece
of
of
the
test
coverage
we
really
need
is.
F
A
Yeah,
I
can
also
yeah.
I
guess
I
can
also
just
point
you
if
you
like
to
the,
because
I
have
a
pr
for
unix
fs,
which
which
should
do
the
trick
basically
to
just
basically
take
this
br
check
it
out,
compile
it
and
see
if
it
works.
I
think
great.
Okay
yeah,
we
can
catch
up
after
the
call,
yeah
cool.
Let's
see
anything
else,.
A
Oh,
I
have
some
notes
down
there
from
someone
that
we
have
plenty
of
repositories
who
wants
to
comment
on
this.
One.
C
Yeah,
this
is
just
a
fun
fact
that
I
wanted
to
make
available
somebody
else
working
within
the
entirety
of
all
the
stuff
that
protocol
lab
supports.
C
So,
like
all
the
ipfs,
all
the
ipld,
all
the
lib
p2p,
all
the
other
things
tried
to
just
do
a
summary
and
an
exploration
of
like
how
many
repos
are
there
across
all
of
these
orgs
that
all
these
people
are
contributing
to,
and
the
answer
is
in
the
hundreds,
it's
not
even
in
the
low
hundreds,
it's
like
770
ish
and
then
I
think
we
discovered
a
whole
other
organization
that
wasn't
actually
fully
counted
under
that
report.
C
So
it's
probably
eight
or
nine
hundred
or
maybe
more
multifulness,
yeah
yeah.
I
think
that
one
was
in
there,
but
yes,
big
numbers.
F
So
I
was
looking
at
an
hour
old
recently
and
still
discovering
repos
that
I
was
like
they're
not
on
my
mental
stack
at
all,
but
I
I
think
I
think
collectively,
like
even
this
group
we
could.
We
could
identify
things
in
the
ipld
org
that
that
we
want
just
to
archive
or
deprecate.
So
if
stewards
is
coming
up
with
a
strategy
for
how
to
do
that,
I
think
we
could
pretty
quickly
make
that
happen
across
the
ipl.org.
It's
just
that
we
don't
have
a
strategy
right
now,
so
they're,
just
in
limbo,.
A
D
Gen
really
a
missing
piece.
I
think
there
there
are
surprising
numbers
of
heavily
relied
upon
repos
and
varying
states
of
disrepair
that
are
still
in
people's
personal
accounts.
That
are
what.
D
F
A
Cool
didn't.
C
Mean
to
give
you
more
work,
no,
it's
actually
a
really
good
observation.
You're
right,
I
like
there
are
some
things
with
my
name
on
it.
That's
in
the
transit
depths.
There
are
some
rod,
things
that
I
know
are
in
our
transitive
depths
in
various
places.
There's
some
juan
things
like
there
are
plenty
of
people
name
things.
A
Cool
is
there
anything
else,
people
want
to
chat
about
or
have
comments.
F
No
cool
yeah,
no,
I
I
was
a
a
little
bit.
I
didn't
quite
catch
the
details
today
and
I
haven't
really
figured
it
out
from
the
github
traffic
about
the
link
system
stuff,
like
the
hold
up
on
link
system
and
go
multihash
mentioned
it
eric,
but
there
seems
to
be.
Is
it
a
conflict
or
just
a
priority
problem
or
a
speed
problem?
What's
going
on
here
between
you
eric
and
will
team,
and
is
that
on
path
to
be
resolved.
C
Was
I
wished
it
would
have
been
done
a
while
ago,
but
in
the
process
of
trying
to
straighten
that
out?
I
drafted
that
whole
thing,
assuming
that
the
go
multihash
library
was
going
to.
Let
me
get
a
standard
library
hash
with
this
interface.
That's
been
in
the
go
standard
library
since
literally
forever,
and
I
found
that
it
doesn't
and
that
actually
like
really
screwed
with
the
abstractions
and
so
at
first
I
just
like
absolutely
bulldozed
my
way
through
it.
F
C
There
are
a
lot
of
interesting
things
like
every
time.
I
think
anyone
in
this
team
has
discussed
anything
about
multi-formats,
we're
kind
of
surprised
at
how
quickly
it
gets
out
of
hand
right
like
those
magic
numbers
they
just
so
some
of
the
stuff
and
go
multihash
is
like
oh
yeah.
We
have
a
function
that
checks.
If
this
magic
number
is
the
right
magic
number
like
is
this
one
that
we
know
to
be
a
multi-hash?
C
C
Suddenly
we
have
to
ask:
what
is
this
thing
supposed
to
mean
and
does
it
actually
provide
value?
And
when
you
ask,
does
this
provide
value
of
a
function?
That's
been
around
for
seven
years
and
nobody's
ever
questioned
it?
Well,
you
know
it
doesn't
matter
whether
the
answer
is
yes
or
no.
The
question
is
tetchy,
no
matter
what.
D
It's
it's,
this
figuring
out
the
reality
of
the
tension
of
what
is
it
that
you
know
like
if
product
wants
a
prototype
of
something
in
three
weeks
and
the
reality
is
we're
just
like
it's
tough
to
merge
stuff
and
actually
do
good
release
management
and
and
there's
just
different
timelines
for
things
and
we're
gonna
have
to
figure
out
how
we
balance
those
and
how
we
talk
about
that,
and
I
think
you
know
we'll
be
able
to
point
back
to
that
at
some
point
of,
like
you
know,
is
the
reality
that,
as
we
identify
these
things,
that
we
need
to
have
projects
that
we
can
flip
back
and
forth
between
to
pick
up
unblocked
threads.
D
I
think
we've
been
lucky
that
for
this
ipl,
the
an
ipfs
one
there's
enough
different
things
that
we're
not
blocked
like
there
are
other
pieces
that
need
to
happen,
but
especially
for
smaller
scope,
things
as
we
end
up
in
there.
I
think
it's
very
reasonable
to
say.
Okay,
so
you
know
we
did
the
engineering
work.
D
But
now
we
need
to
expect
two
or
three
weeks
to
to
get
this
through
steward
review,
which
is
you
know,
ten
percent
the
author's
time
answering
comments,
but
is
just
you
know
a
workload
thing,
and
so
how
do
we
talk
about
that?
When
we're
talking
about
the
three
weeks
that
a
project
takes?
So
I
think
there
will
be
iteration
on
figuring
out
what
that
means,
but
I'm
not
unhappy
that
this
came
up
in
the
beginning
as
a
thing
that
we
then
go
back
and
point
to
of
like
okay.
What
does
unblocked
mean.
F
I
think-
and
I
hope
I
hope
that's
working
out
well
eric
that
you
have
other
people
to
lean
on
to
get
that
resolved.