►
From YouTube: 🖧 IPLD weekly Sync 🙌🏽 2020-04-27
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
Welcome
everyone
to
this
week's
IPL
d/deaf
syn-coll
is
April
the
27th
2020
and
as
every
week
we
go
over
the
stuff
that
we've
worked
on
the
past
weeks
and
then
discuss
any
open,
I
notice
we
might
have
everyone
was
attending.
Please
put
your
name
on
the
attendees
list.
If
you
haven't
attended
last
week,
as
your
name
will
already
be
there.
A
A
But
now
everything
is
fine
again
and
we're
in
a
good
shape,
also
things
to
who
from
the
IPS
team,
to
basically
help
me
with
those
things
and
yeah
you're
working
again
and
then
I
also
spent
finally
time
I
get
a
ILD
still
in
the
block
api
and
now
I'm,
truly
porting
a
small
tool.
I
used
javascript
to
build
eggs
from
a
text
file
and
it
also
uses
dick
chase
on
in
the
night
on
the
rust,
IP
LD
take
chase
implementation
had
some
weird
error
and
then
I
found
the
problem
and
I
was
like.
A
Why
is
it
implement
like
this
and
then
I
thought?
Oh,
our
spec
is
pretty
bad,
so
you
can't
really
implement
de
JSON
based
on
the
spec,
because
it's
totally
I'm
not
specified
so
I
spend
time
looking
at
the
spec
and
see
that
there's
P
I
was
open
from
August
last
year
and
I
tried
to
push
it
through
so
I'm,
pretty
close.
A
So
if
I
think
Michael,
if
you
approve
this
PR,
then
I
will
just
merge
and
then
we
will
be
good
and
I
have
another
PR
open
for
even
more
country
stuff
and
then
Dec
choice
and
should
be
fine
and
then
I
will
fix.
The
roster,
notation
and
probably
I
would
expect
oscillator
JavaScript
and
perhaps
even
Co.
Implementation
needs
to
change
a
few
things
to
be
spec
compliant
but
again,
like
deck.
A
Chase
is
mostly
used
for
testing,
so
I
don't
see
a
real
urge
in
like
being
fully
spec
compliant,
but
eventually
we
should
and
also
in
regards
to
this
I
plan,
to
do
some
test
vectors,
probably
starting
with
take
chase
on.
So
basically
all
implementations
were
just
reading
files.
A
Encoded
decoded
see
if
the
CID
that
you
it
comes
out
matches
to
make
sure
you
are
compliant.
I
think
this
would
be
a
huge
step
forward,
because
chrome
is
really
hard
to
for
new
language
is
to
implement
stuff,
because
you
need
to
look
at
the
code
of
JavaScript,
augmentation
and
probably
then
don't
agree.
And
then
you
pick
one
and
you
can
tell
if
that's
according
to
the
spec,
because
there's
no
space,
so
if
we
have
test
vectors
I,
think
that
will
improve
the
story.
A
B
B
So
this
is
a
lot
of
stuff.
Unions
are
still
missing
some
of
those
other
fancy
types,
so
those
are
still
future
work.
But
a
lot
of
the
essentials
are
now
there's
one
other
major
caveat
which
doesn't
quite
prevent
this
from
being
used
in
some
cases,
but
like
needs
to
fix
in
the
representations,
nodes
formats
and
lists
do
not
recurse
correctly
as
you're
stepping
through
it
will
sort
of
forget
that
you're
stepping
over
representation
knows
and
movers
return,
type
level,
note
interface,
and
so,
if
you're
traversing
over
some
structure.
B
Where
those
happen
to
be
the
same
thing,
then
you
won't
notice
that,
but
if
you're
doing
it
over
some
structure
where
you've
got
like,
oh
I'm
sure,
I'm
over
a
map
and
then
I
reaches
struct
on
this
truck
has
a
spring
join
representation.
So
it's
a
spring
and
autumn
now
then
you'll
definitely
notice
because
you'll
end
up
getting
map
in
the
mouth
when
you
should
have
gotten
string,
so
that
needs
to
be
done
soon.
B
This
is
easier
said
than
done,
because
I
don't
know
starting
to
balance
a
lot
of
different
things
like
minimizing.
The
amount
of
code
in
the
templates
is
totally
different
than
code
that
it's
generated
and
that
in
turn,
sometimes
has
surprisingly
little
correlation
with
how
big
of
assembly
and
like
binary
compiled
size,
is
generated.
B
In
order
to
be
able
to
really
go
forward
effectively,
so
we'll
see
how
that
goes,
I'm
not
sure
how
I'm
gonna
prioritize
that
either
I
might
just
put
that
on
the
table
until
later
and
just
do
the
naive
thing
has
a
lot
of
extra
line
will
come
just
to
be
able
to
demo
we'll
see
and
I
did
start
trying
to
deduplicate
some
of
the
template
code
as
well
in
places
where
it
was
obviously
possible
got
some
nice
progress
on
that
this
week,
a
lot
to
go
because
I
have
been
intentionally
conferring
it
to
all
of
the
aforementioned
reasons
of
it
just
being
complex
and
it'll
work
on
pushing
the
note,
assembler
generation
of
interfaces
the
whole
giant
winter
rework
into
the
graphs
and
levers
which
consume
it,
and
that
should
almost
be
done,
but
I
hadn't,
quite
I
got
some
test
failures.
B
That
I
can't
quite
make
heads
nor
tails
on
so
I'm
requesting
some
fire
support
from
Anna
there.
Hopefully,
that
all
ends
promptly,
assuming
it
does
I'll
go
back
to
Marco
Jenna.
Hopefully
you
have
a
really
big
demos
of
that
in
some
other
projects,
possibly
the
next
week
or
so.
Weird.
Very
exciting,
no
promises,
that's
my
dream
in.
C
Yeah
yeah,
so
a
couple
things
lest
we
get
started
in
on
the
J
s,
multi
formats,
refactor
stuff,
the
approach
that
I
ended
up
having
the
take
was
just
rewriting
all
of
it.
It
turned
out
the
rewriting.
All
of
it
was
actually
the
easiest
way
to
go.
If
you
weren't
listening
last
week,
we
need
to
move
to
a
model
where
we
don't
ship
every
code,
a
cover,
every
base,
encoding
library
ever
function
ever
we
you
need
to
be
able
to
pass
in
the
ones
that
you
want
support
for
when
you
instantiate
the
block
API.
C
When
you
don't
have
the
full
table
of
multi
floor
mats,
you
just
can't
support
a
lot
of
the
things
that
we
used
to
do
like
you
can't
do
certain
detection
if
something
was
a
multi
based
screen,
because
you
don't
know
all
of
the
possible
prefixes,
because
you
only
know
the
ones
that
were
loaded.
You
can't
get
the
string
of
the
codec
because
you
don't
know
the
screen
from
any
codecs
and
rather
than
trying
to
support
these.
When
you
have
that
information,
but
not
when
you
don't.
C
C
They
understand
that
it's
a
link
to
a
thing
so
anyway,
this
is
all
going
to
turn
into
a
lot
of
changes
down
the
line,
but
it's
a
really
kind
of
important
and
necessary
move
forward
to
get
out
of
using
all
this
crazy
stuff.
Also,
the
implementation
of
all
of
these
things
is
going
to
be
smaller
than
any
of
the
individual
old
libraries.
So
we
need
to
know
they're
all
combined
in
one
library.
Now
it's
actually
like
not
that
much
code.
C
C
You
can
also
replicate
a
database
to
a
property
rather
than
merging
the
entire
database
in
because
it's
a
it's
a
die,
so
you
can
just
lust
databases
indefinitely.
So,
if
you
want
to
have
a
database,
that's
a
collection
of
a
bunch
of
other
people,
databases,
you
can
add
them
all
as
remote
and
then
plug
them
into
all
these
different
properties
and
everything
just
works
and
it
does
smarter
merging
as
a
value
like
it
does
the
full
database
mode
rather
than
just
trying
to
swap
out
the
root
Rev.
So
yeah,
it's
it's
pretty
cool.
C
C
C
Okay,
what
are
we
not
gonna
do
so.
This
has
been
sort
of
surfacing
and
some
of
the
conversations
that
we've
had
lately
around
encryption
and
identity.
Although
I
think
that
we
all
know
that
we're
not
going
to
add
identity,
but
encryption
starts
to
get
into
some
hairy
areas,
especially
like
and
I,
didn't
realize
this
when
we
started
it,
but
selectors
are
a
very
particular
way
of
doing
leads
and
because
they
can
read
an
entire
dag
if
selectors
sort
of
represent
the
IP
LD
way
of
reading.
C
Things
are
like
a
prescribed
way
of
reading
things
like
encryption
and
figure
out
how
they
work
with
selectors
to
apparently
and
I.
Don't
think
that
we
anticipated
that
or
thought
that
that
would
be
a
thing
that
we
were
doing
when
we
took
on
the
selectors
work,
so
I
think
even
important
to
our
selective
work
is
to
figure
out
what
the
boundaries
there
are
as
well
so
yeah.
C
This
is
just
a
min
to
like
keep
in
mind
and
we
should
have
the
discussion
eventually,
but
I
want
to
figure
out
what
we
aren't
and,
as
we
take
on
more
ecosystem
projects
that
are
like
the
IDB
they're
like
using
AI
Kildee.
There
we
have
a
clear
understanding
of
like
these
are
not
part
of
the
project,
and
this
whole
kind
of
area
of
innovation
happens
outside
about
guilty
and
I.
Peel
D
is
just
shoring
up
and
solidifying
what
it
does
that
make
sense.
A
D
Still
going
on
with
the
multi
codec
discussions,
as
Volker
said,
that
is
both
extremely
frustrating
and
also
very
educational.
For
the
same
time,
and
it's
you
know
wading
into
things
that
where
we
have
opinions
about
how
things
should
be,
but
not
very
deep
knowledge
about
how
they
need
to
be
used,
so
I
think
this
is
a.
This
is
a
good
process
for
us
to
go
through,
but
it
is.
It
is
frustrating
and
yeah
yeah.
D
This
yeah
there's
some
really
interesting
context
that
was
posted
yesterday
that
haven't
been
out
to
absorb
yet,
but
that's
it's
still
on
the
go
we'll
get
there
somehow,
mostly
of
them
that
have
been
doing
Bitcoin
stuff,
so
I've
dumped
the
entire
block
chain
onto
my
desk,
two
terabytes
worth
of
binary
files.
The
block
blocks,
coupled
with
compact
Jason
form
of
the
block
and.
D
Yes,
so
that's
just
this:
that's
this
as
part
of
the
process
of
being
able
to
pull
pull
it
in
so
gets
just
getting
it
out
of
the
client
to
have
it
on
disk.
It
makes
us
make
some
easier
to
work
with
I
have
a
complete
block
printer
working,
so
I
can
replicate
the
exact
Jason
that
the
Bitcoin
client
prints
out,
which
is
nice
and.
D
Which
is
actually
no
mean
feat,
because
there's
a
bunch
of
other
so
much
stuff
in
there
there's
very
peculiar
about
how
Bitcoin
represents
certain
fields
and
pulls
them
apart
into
their
own
little
thing.
But
it's
nice
to
be
able
to
verify
that
I
can
take
a
block
and
what
the
Bitcoin
client
client
tells
me.
It
is,
and
I
can
do
the
exactly
the
same.
Javascript
and
say
I
know
exactly
what
this
is
in
watts
in
every
field.
I
can
also
go.
D
D
So
my
next
steps
with
all
these
are
I'm
I
need
to
generate
a
set
of
IP
lb
blocks
for
each
Bitcoin
block,
it'll,
be
a
large
string
of
blocks
and
then
make
sure
I
can
do
the
reverse,
where
I
can
take
the
blocks
and
turn
them
back
into
the
negative
Bitcoin
format.
So,
given
a
series
of
blocks
or
a
navigatable,
IP
LD
structure
can
I
form
it
back
into
that
the
instantiated
format
of
the
entire
block,
and
can
I
navigate
it
with
with
paths
as
well?
A
D
A
But
I'm
happy
to
do
a
review
or,
like
probably
really
don't
be,
like
I
skimmed
through
your
changes
but
yeah
I'm
happy
to
do
this
and
I'm
also
happy
too
busy
hand
over
the
mint
in
a
row
and
which
is
basically
also
the
same
thing.
What
Michael
is
working
on
with
you,
three
multi-format
stuff
and
cremate
him
off
like
also
the
CED
stuff
and
so
on,
so
I'm
basically
happy
to
be
slowly
get
face
down
and
so
I
can
concentrate
on
the
rest
stuff
and
you
take
over
that
childhood
stuff.
So
yeah,
it's
cool.