►
From YouTube: 🖧 IPLD Bi-Weekly Sync 🙌🏽 2019-01-21
Description
A bi-weekly meeting to sync up on all IPLD related topics. It's open for everyone and recorded. https://github.com/ipld/team-mgmt
A
A
A
D
C
Yeah
I
mean
the
main
thing
that
we
need
to
figure
out
in
in
this
one
and
started
B
rail
like
any
other
sort
of
actually.
So
before
the
end
of
the
meeting,
we
will
do
a
quick
update
actually
just
from
everybody,
but
I
just
want
to
get
this
out
in
front
because
we
have
new,
okay,
ours,
that
I
want
to
try
to
land
today
and
it
basically
sort
of
shifts
around
a
bunch
of
the
priorities
for
the
quarter.
So
I
want
to
just
make
sure
that
we
have
alignment
on
that
before
we
move
forward.
C
So
the
main
thing
is
that
we're
adding
four
okay
ours
that
are
basically
identical
for,
go
and
Jas
to
implement
a
IP,
LD
selector
grammar
parser
and
to
implement
a
dag
traverser
for
get
for
any
given
selector
ast.
Essentially,
though,
the
parsed
version
of
those
selectors.
So
if
you,
basically,
you
can
give
it
sort
of
a
block,
store
and
say,
hey,
go
get
me
all
the
blocks
that
match
this
particular
selector
and
it
will
go
and
do
that
essentially,
and
and
obviously
the
grammar
parser
is
pretty
self-explanatory,
but
yeah.
C
C
Okay,
cool
there's
been
some
discussion
in
the
PR,
but
it
sounds
like
it's
about
Radle,
and
that
was
the
main
thing
that
I
was
worried
about
next.
Actually,
all
it
Eric
go
next
because
I
by
the
end
of
the
meeting,
I
want
to
figure
out
how
we're
kind
of
staging
the
work
in
terms
of
like
what
fuller
is
doing
and
what
Eric
is
doing
cuz.
It
sounds
like
it's
been
doing.
Some
work
on
the
on
a
schema
engine.
C
D
Yeah
I,
don't
know
how
much
the
code
will
be
portable,
but
at
least
like
my
brain
state,
so
I've
been
doing
a
lot
of
research
on
the
schema
stuff
and
long
story
short.
There
is
a
draft
now,
so
if
people
want
to
check
that
out
and
make
comments,
this
is
a
draft
that
covers
all
of
the
basic
stuff
that
we've
described
in
past
conversations.
So
there
are
types
based
on
the
IP
LD
data
model
stuff,
which
you
know
like
that
list.
D
That
would
actually
be
good.
Otherwise
it's
just
way
more
work.
So
there's
a
PR
link
in
the
in
the
pad,
that's
shared,
and
if
people
want
to
check
that
out
and
make
comments
like
it's
big
called,
if
this
schema
should
completely
self
described.
So
there
is
a
kind
that
is
called
like
struct
and
there
is
a
type
in
the
schema
called
type
struct,
which
describes
what
a
struct
can
have.
D
So
it
reads
a
little
funny:
you'll
see
some
JSON
elements
where
it's
like
the
type
of
the
fields
called
type
and
like
this
is
what
self
description
is
so
yeah
go
check
that
out,
there's
a
bunch
of
work,
bringing
that
close
to
being
implemented
and
usable
in
the
go
side
already.
There's
a
type
stop
node
implementation,
which
is
still
in
go
terms,
the
same
interface
as
the
other
notes.
D
It's
just
going
to
apply
more
rules,
so
if
you
have
a
type
violation
than
it
will,
depending
on
exactly
how
you
load
up
the
library
either,
do
it
in
advance
when
you're
putting
data
in
or
do
it
at
runtime.
When
you
try
to
access
it,
both
paths
should
be
supported
and
in
the
far
future,
I
want
to
have
code
generation
for
this.
But
I
won't
talk
about
that
today.
D
A
Yeah.
That's
my
update,
but
I
have
already
a
thing.
I
want
to
ask
Eric,
because
so
we
will
do
these
selectors
thing
for
go
and
just
probably,
and
will
you
be
okay
with
like
PCAs,
you
work
on
the
schema
stuff
currently
that
I
start
on
this
Travis
crypt
side
with
the
ast,
pausing
kind
of
things
and
then
yeah.
We
work
together
on
basically
making
sure
that
it
works
on
both
sides,
but
as
I've
already
spent
some
time
on.
A
D
A
C
So
one
thing
that
would
be
super
helpful
I
think
if
we
can
try
to
have
it
by
the
end
of
the
quarter,
is
so
at
the
bottom
of
the
selector
specs
that
one
wrote
it's
a
bunch
of
sort
of
implementation
notes
for
go
if
those
could
at
some
point
get
replaced
with
the
actual
ast
definition.
That's
not
dislike
language
control.
C
C
E
These
are
more
raw
thoughts
that
actually
any
proposal
to
make
a
plan,
but
I've
I'm,
starting
to
think
that
it
would
be
interesting
to
consider
making
IP
LD
as
something
that
can
be
used
stand
alone
without
ipfs
and
by
that
I
mean
I'd,
be
making
ideally
usable
on
top
of
my
peer
to
peer
directly,
so
that
then
I
PLD
can
govern
the
API
of
the
API.
E
That
ipfs
focus
really
hard
on
governing
and
and
stabilizing
the
api's
on
files
in
the
pewter
peer
can
I
focus
on
the
networking
part
again
like
this
is
like
very
raw
thoughts
and
I
haven't
as
a
chance
to
write
all
this
down
but
like
it
seems
as
a
good
path
moving
forward
to
create
what
we
can
then
call
like
stable,
api's
we've
without,
like
maybe
like
from
the
user
perspective
like
if
an
IP,
fast
user
comes
I
profess
C's
are
the
files.
Api
is
and
I
know
for
sure
about
this
concept.
C
C
Okay,
let's
look
like
create
this
new
replication
API,
but
not
take
as
part
of
that
context.
In
that
development,
like
everything
that
I've
give
us
is
already
doing,
it's
just
a
little
bit
more
difficult.
I
would
definitely
like
to
get
there,
but
I
see
it
as
like
a
slightly
longer
term
thing
right.
Then
then,
the
storage
layer
stuff,
if
that
makes
it
got.
E
C
Specifically
on
the
JavaScript
side,
though,
and-
and
part
of
this
is
like
I-
think
one
of
the
reasons
why
these
conversations
get
lost
is
because
it
was
also
included
in
some
of
the
very
early
talks
about
moving
to
async,
await
api's
and
generators
and,
like
we
were
talking
about
like
oh
well,
we
need
we
want
to
do
an
update
of
our
API
anyway,
let's
make
it
less
specific,
and
all
of
that
so
yeah
I
can
see
why
things
got
very
confused
there.
Eric
would
have
to
speak
to
it.
C
D
Yeah
I
was
nodding
intensely
to
the
separation
of
concerns.
Description
like
I've,
actually
been
very
carefully
resisting
importing
any
packages
and
as
much
as
possible
to
go
IPO
D,
prime,
because
my
hope
is
that
it's
going
to
be
this
like
intensely
reusable
thing
and
so
I
want
us
dependency
surface
to
be
absolutely
minimal.
D
It
should
be
something
that,
like
anyone,
can
put
into
anything
with
a
slightest
chance
at
the
time
of
dependency,
hopefully,
and
so
it
Yanks
in
the
Sydney
libraries
and
and
nothing
else,
the
rest
of
its
going
to
like
be
decorated
on
top
rather
than
I'm,
actually
really
hoping
like
I,
think
in
a
perfect
world,
it
would
be
also
if
we
could
get
the
IPL,
the
tooling
and
schemas
and
stuff.
In
addition
to
serving
ipfs
needs
really
well
and
Merkel
web
in
general.
D
D
Road,
but
like
that's
kind
of
where
I'm
mentally,
aiming
like
I,
have
a
bunch
of
other
projects,
for
example,
where
I'm
like
doing
some
message,
passing
over
standard
and
standard
out
between
it
process
tree
and
like
I
want
some
type
system,
because
the
parents
in
JavaScript
human
childs
and
go
don't
ask
me
how
that
happened.
Did
though,
and
like
I,
basically
want
I
peel
D?
For
that
too,
it
would
be
so
useful.
D
E
D
Don't
know
this
is
exactly
the
time
replace,
but
from
the
go
side,
so
I
haven't
figured
out
exactly
how
to
write
this,
but
I
suspect
part
of
the
solution.
I
might
try
for
go
is
since
we
have
structural
typing.
You
can
do
this
thing
and
go
where
you
define
the
interface
of
one
place
to
find
it
another
place.
If
it's
the
same
methods,
they're,
basically
equal,
you
can
just
toss
them
back
and
forth.
F
C
Yeah
I
mean
that's
pretty
similar
to
what
we've
been
in
the
past.
Actually
on
the
JavaScript
side.
I
think
that
I
think
that
I
would
actually
like
to
move
towards
is
it
is
away
from
bigger
interfaces
and
to
smaller.
So,
for
instance,
we
have
a
bunch
of
if
you
guys
where
you,
you
handed
an
entire
block
service
service
API.
But
many
of
these,
like
a
deck
traverser
only
needs
to
get
blocks
by
cid.
And
if
that's
the
case,
then
it
should
just
take
a
single
async
method.
Yeah
called
yet
they
get
blocks
by
cid.
C
And
then
you
can.
You
know
it's
one
line
of
code
to
to
map
a
block
service
object
on
to
that
right,
but
yeah
yeah
I
see
what
you're
saying
okay
cool
before
yeah.
We've
only
got
five
minutes
so
Ellen
and
Stephen,
and
he
he's
coming
with
us:
okay,
cool
cool,
okay,
one
last
question:
Davida
I,
wanted
to
in
the
ipfs
project.
Wt
g
talk
about
some
IP
LD
prioritization
stuff.
What's
the
best
way
to
add
agenda
items
to
the
project
working
group
meeting
so.
E
The
notes
talk,
he's
a
Google
Doc
that
is
open,
like
I.
Typically
am
the
one
creating
the
agenda
so
typically
I'm
the
one
for
siding
like
just
copy
placing
the
template
like
they
had
in
debate,
but,
like
anyone
should
feel
comfortable
like
if
it's
not
there
already,
you
should
just
add
it.
Let
me
link
it
to
you,
and
so
you,
like
the
the
calendar
event,
is
on
Google
Calendar.
D
C
Again,
that's
awesome,
awesome,
I
edited
also.
It
said
that
the
meeting
was
today,
but
it's
not
today
tomorrow,
so
I
updated.
That
date.
F
A
The
the
only
things
that
already
be
my
work
on
the
new
so
to
answer
that
question
earlier
that
my
rug
on
the
IPL
dapi
certainly
is
to
make
them
less
dependent
on
ipfs
and
just
being
like
yeah,
a
toy
for
IPS
to
really
have
like
proper
API.
So
there's
also
part
of
the
API
work
to
make
it
the
usable
as
a
standalone
thing.
Yeah.
That's
all.