►
From YouTube: 🖧 IPLD weekly Sync 🙌🏽 2019-09-16
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
Don't
think
that
live
stream
works
but
doesn't
matter,
the
recording
is
one
welcome
everyone
to
this
week's
IPL,
the
team
sync
meeting.
It's
September
16th
2019.
As
every
week
we
tell
everyone
what
we've
working
on
and
what
we
do
next
and
also
discuss
any
open
items
we
have
as
every
week
is
also
start
with
myself.
A
Library
using
survey
form
message
pack,
which
I
think
so
they
have
a
similar
thing
to
takes
in
Seymour
I
think
they
found
a
pretty
smart
way
so
I'm
trying
this
one
to
see.
If
I
can
convince
the
author
to
perhaps
much
my
version
of
it
and
but
I
ongoing
work
and
yeah
we'll
see
how
this
goes
all
right
and
then
next
on
my
list
is
Eric.
C
C
So
this
week,
I
spent
a
bunch
of
time
trying
to
enumerate
just
all
the
design
choices
that
we've
got
coming
up
about.
How
generated
code
should
look
and
feel,
and
so
there's
a
whole
bunch
of
interesting
questions
going
on
there.
This
stuff,
that's
dumb,
is
basically
regarding
the
generic
LeMond
amorphize
that
you
know
interfaces
that
we
are
pretty
used
to
dealing
with,
but
those
still
aren't
like
the
end
of
the
story.
C
C
For
strongest,
especially
there's
a
ton
of
their
own
beans,
because
that
is
not
a
concept
like.
Oh
maybe
I
can
do
it
in
many
different
ways,
but
getting
economics
questions
so
lots
more
to
go.
There
I've
got
a
document
with
some
decisions
outlines,
but
I
think
is
probably
more
to
discover
them
before
it's
any
procedure
on
coloring
it.
D
So
my
wakers
will
been
specs.
Oh
sorry,
schemas.
Here's
a
basic
list
just
from
looking
back
over,
mainly
the
specs
repo,
which
is
telling
me
most
of
what
I've
been
doing,
string,
drawing
string
pairs
and
list
pairs
and
now
in
jeaious
go
and
the
specs
they
were
in
the
docs.
But
now
they're.
It's
fully
expect
and
that's
in
the
implementations.
D
D
D
We
lifted
that
from
the
far
coin
specs,
including
Jason,
and
go
X.
Let's
say
that
were
either
a
union
of
different
kinds
of
bytes,
so
you
got
to
you
know
you
get
two
or
three
things
that
are
of
byte
kind,
but
are
named
types
you
have
to
put
them
in
the
union
with
bike
prefix
and
you
give
them
each
an
integer
value,
then
that
integer
will
differentiate
between
what
the
trailing
bytes
are.
D
D
It
is
actually
in
the
in
I
think
it's
in
the
implementations
already,
but
it
needs
to
be
merged
in
the
specs.
There
was
another
thing
in
here:
threat
advanced.
D
So
the
advanced
data
layout
stuff
we
boiled
down
last
week
to
some
syntax
and
then
that's
still
import
request
in
this
box.
Repo
Erik
raise
some
interesting
challenges
with
that.
Yesterday
was
regard
to
like
thinking
through
the
encryption
news
case
and
whether
it
even
makes
sense
to
have
it
as
a
representation
and
how
these
things
compose
together
and
unpack,
and
that's
he
left
me
with
that
thought.
As
I
was
going
to
bed
so
I
haven't
I,
haven't
fully
I
started.
Writing
something
figure
out.
How
that
would
look
and
I
could
see
some
other
problems.
B
D
But
that
sort
of
that's
a
bit:
that's
one
of
the
big
blockers
right
now.
It's
getting
that
syntax
merge,
so
I'm
actually
wondering
whether
there's
a
short
path
here,
because
I
would
really
like
to
get
to
the
point
this
week
of
being
able
to
validate
the
file
coin
specs,
as
with
some
minor
changes
to
their
syntax
and
the
whole
string
key
thing.
D
So
that's
why
it's
a
priority
and
then
the
other
thing
is
the
they're
using
aliases,
so
type
alias
is
where
you
know,
type
foo
bar
so
you're,
saying
foo
is
the
type
of
bar
Eric
and
I
were
talking
about
that
yesterday
and
how
one
might
concede
of
that
through
the
hole,
a
schema
process,
belly
flows,
truth!
If
we
can
figure
those
two
things
out,
then
that
would
satisfy
the
goal
of
being
able
to
validate
far
coin
specs,
with
some
other
changes
to
them,
that
that
would
be
a
nice
thing
to
get
to
this
week.
D
E
Just
gonna
say
like
the
events:
layouts
is
the
last
thing.
That's
blocking
me.
Actually
so
I
mean
aliases
like
once.
They're
in
I'll
probably
use
them
at
some
point
of
time,
but
there's
actually
like
a
lot
of
updates.
They'll
have
to
make
to
the
to
my
schema,
gen
stuff,
won't
aliases
land,
because
right
now,
I'm
speaking
to
Erik's
kind
of
carnality
concept.
The
encoder/decoder
paths
are
basically
the
same
because
they
take
decoding
block
data
or
data
that
you
had.
You
want
to
give
to
create
a
block
and
those
are
identical.
E
D
I
was
thinking
that
maybe
what
I
could
do
is
get
a
get
a
form
into
the
J
esparza
that
conforms
to
what
we
agreed
to
last
week
with
regard
to
advanced
layout
syntax.
But
every
time
you
use
it,
it
spits
out
a
warning
to
standard
error
and
says
this
is
this
is
not
necessarily
going
to
be
the
final
form,
but
it's
usable,
and
then
that
would
work
for
the
file
coin
and
your
use
case,
but
give
us
that
out
which
there's
we
haven't.
D
D
That
would
give
some
breathing
room,
so
we
don't
rush
something
through
that
we
might
end
up
regretting
not,
but
not
that
it's
a
big
deal
to
change
these
things
right
now.
We
don't
know
with
any
consumers
of
this
stuff.
So
it's
not
so
breaking
changes
and
not
the
worst,
but
it'll
be
good
to
get
this
one
run.
C
E
D
I
should
note
that
in
here,
I'll
go
back
to
that
when
I'm
done
speaking.
So
the
last
thing
was
this
progress.
1
line
2,
which
is
a
distillation
of
180
and
184
I,
think
they
were
so
the
hashmaps
back
the
hemp
spec
I
it
currently
encodes
keys
as
bytes.
So
whether
you
provide
string,
keys
or
byte
keys
in
the
interface,
it
will
encode
them
onto
the
block
as
bytes.
D
This
proposal
changes
that
to
store
it
to
to
storing
a
union
of
bytes
or
strings.
So
if
you
provide
string
keys
in
the
implementation,
it
will
store
string
keys
and
if
you
provide
bytes
keys,
it
will
store
bytes
now,
according
to
our
data
model,
if
you
use
this
as
a
dermal
replacement
like
an
advanced
layout,
then
you
probably
should
only
be
using
strings,
but
the
implementation
that
allowance
for
bytes
is
no
it's
by
2,
in
fact,
more
efficient
than
strings,
because
you
don't
have
to
convert
them
to
do
the
hashing
and
there
are.
D
D
E
E
E
Then
the
leader,
all
of
a
sudden,
we
didn't
do
multi
block,
because
when
you're
writing
you
can
think
about
a
block
at
a
time
previously
put
when
you're
reading
you
have
to,
or
you
can
produce
a
block
at
a
time
where
there,
but
when
you're
reading,
like
especially
reading
through
these
schema
generated,
ap
is
the
whole
point
of
them.
Is
that
you
can
define
where
the
block
boundaries
on
that
sort
of
like
easily
occurs
through
them.
E
So
doing
that,
like
we're
cursing
through
the
union's
in
the
fairly
transparent
way,
that
kind
of
stuff
worked
on
that
have
some
pretty
basic
stuff
working
now
on
the
reader
side
of
it,
the
initial
sort
of
API
generation
stuff
that
I
did
was
a
little
messy
and
it's
gotten
a
little
bit
cleaned
up.
But
it's
now
starting
to
show
where
it's
really
really
messy
and
it'll
probably
need
to
be
factored
a
bit
at
some
point.
E
A
E
I
mean
it's
well
we're
not
even
really
parsing
markdown
like
we're
parsing
code
blocks
out
of
the
markdown.
We
never
actually
like
you
and
throw
it
into
a
markdown
parser,
because
it's
just
unnecessary,
like
we
do
something
very
similar
to
what
broker
posted
with
those
awk
scripts.
It's
just.
We
needed
the
the
type
of
the
code
block.
So
you
know,
Brittany
I
ended
up
earning
my
own
thing
and
then
rod
changed
a
bunch.
So
I
don't
even
know
what
it
looks
like
now,
but
it's
great
actually
yeah.
D
No,
it's
the
so
yeah
Michael,
mutters
initial
version,
just
stripped
them
out
and
and
then
treated
each
block
as
a
different
part.
I'm
wrote,
I
rewrote
it
so,
instead
of
doing
that,
it
just
it
took
the
markdown
and
then
replaced
everything
that
wasn't
spec
with
blank
lines
and
then,
when
you
pass
that
through
apasa,
you
get
errors
that
tell
you
the
line
numbers
saying
they're,
applying
your
markdown
dog.
So
that's
okay,.
D
A
C
C
C
Sometimes
rod
you
and
I
have
been
chatting
about
like
how
it
would
be
simultaneously.
It
seems
useful,
but
I'd
also
just
incredibly
frustrating
Papworth
like
okay,
if
I
type
in
communication,
yes
t
type
of
mutation.
This
is
the
comments.
Are
none
of
those
things
it
makes
me
wonder
if
having
the
separate
ast,
one
is
actually
a
good
idea,
so
that
stuff
can
be
persisted
there.
D
A
C
D
I
mean
is
that
that's
one
of
the
nice
things
about
having
the
JSON
for
me,
so
it's
this
canonical
thing.
That
is,
that
is
good,
although
that
say
something
well,
no,
it's
both
it's
it's
in,
but
although
I
mean
Eric,
here's
another
thing
about
that
which
is
the
ordering
I
guess
with
Jason
that
sort
of
helps,
but
that's
something
that
this
go
map
ordering
thing
has
been
just
crazy
for
making
Jason
these.