►
From YouTube: 🖧 IPLD weekly Sync 🙌🏽 2020-06-15
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
So
in
the
past
week,
I
haven't
done
that
much
I
made
sure
what
I
actually
did
I
think
I
did
a
bit
of
Charles
fruit
stuff
as
well,
but
most
notable
for
us
is
that
the
block,
if
hypr,
is
almost
done
still
in
review.
It
gets
a
bit
distracted
in
the
review
by
random
things,
but
I
hope
that
eventually
I
will
get
there
and
so
slowly
getting
there
and
the
block.
A
If
I
should
be
there
soon
and
then
it's
basically
what
I
really
wanted
to
get
in
and
then
I
really
have
to
look
like
what
I'm
doing
next,
probably
something
like
look
into
schemas
or
something
MOC.
A
B
In
a
bit
so
last
week,
wrote
and
Michael
gave
me
a
little
simple
project
to
add
something
too
low
to
CLI,
which
turned
out
not
to
be
a
simple
little
project,
so
it
turns
out
that
we
basically
have
to
shorten
specs
for
selectors.
Now
one
is
what
we
had
in
the
shorthand
or
schema
specification.
The
other
one
is
the
one
the
two
most
working
on.
Neither
of
them
is
very
CLI
friendly
at
all
number
one
number
two.
B
What
the
hold
ups
were,
but
long
story
short
basically
will
end
up
not
doing
selectors
at
all
for
the
consideration
of
CLI
evidence
that
will
do
basic
passing.
We
had
with
a
deliberately
limited
character
set
and
a
little
bit
further
in
the
meeting
who
will
have
to
circle
back
to
that.
Basically,
if
I
am
not
using
characters
that,
for
example,
Eric
wants
to
use
for
something
else
based
on
this
write-up
earlier
today.
B
C
Basically
correct
so
good
job,
so
I
spent
a
bunch
of
time
off
in
the
corner,
tinkering
around
with
cogent
stuff
and
also
doing
a
little
bit
of
a
change
of
pace.
It's
been
some
time
poking
around
on
experimental
projects
for
what
it
would
be
like
to
actually
be
sumer
of
some
of
this
code.
Gen
stuff
I've
been
working
on
and
I,
don't
know
if
I
have
a
ton
of
finished
thoughts
to
report
from
that
other
than
it
was
certainly
a
thought-provoking.
So
I'm
going
to
be
doing
a
little
more
review
of
how
these
systems
work.
C
The
Cochin
Megadeth
branch
that's
been
out
for
quite
a
while
now
had
accumulated
over
100
commits
by
the
end
and
I
have
now
landed
that
on
master.
There
is
also
a
nice
table
and
read
me
for
future
completeness.
It
contains
check
marks.
It
also
contains
X's.
There's
still
lots
more
to
go
there,
but
it's
in
single
earth
master
again,
so
like
small
sigh
of
relief
there,
and
that
is
my
main
progressions
for
the
week.
C
A
E
Ok,
so
I
got
to
write
a
lot
of
code
last
week.
That's
awesome,
okay,
so
just
multi
formats,
the
interface
work
is
done.
It's
complete
now
other
than
moving
to
USM.
That
literally
normal
thing
that
happened
in
moving
the
ESM
there
were.
E
There
was
some
some
side
questing
involved
in
getting
coverage
working
turns
out
that,
because
in
because
of
nodejs
ESM
is
actually
provided
by
v8,
you
can
no
longer
hook
into
the
module
system
and
just
transpile
all
of
the
code,
two
random
things,
which
is
how
covered
was
implemented
before
so
have
to
use
the
native
v8
tools
for
coverage,
tool
called
c8,
and
there
are
some
corners
of
that
that
aren't
quite
up
to
par
yet,
but
it's
working
enough
for
us
to
rely
on
it
and
to
use
it
so
I
moved
the
the
hundreds
tool
that
I
wrote
and
a
lot
of
our
cover
checking
to
see
eight
the
nice
thing
about
it,
because
it's
native
it's
way
faster,
then
than
the
Istanbul
and
yce
coverage
tools.
E
That's
great!
That's
all
working
now
I
improve
the
github
action
quite
a
bit.
So
now
we
do
matrix
for
all
of
our
tests
and
tests
on
112.
That
gauge
is
12
and
forum
stato
point.
Well
there
appreciate
that
so
anyway,
that's
all
working
now,
it's
a
great
piece
of
github
automation.
I
have
now
decided
that
I
am
I.
E
E
E
We
also
have
plugins
for
every
hashing
function
that
we
had
in
multi,
hashing
async
now
so,
but
is
multi-format
stuff
is,
is
all
there
now
and
it's
totally
usable
as
long
as
you're
comfortable
migrating
to
es.
Mr,
so
that's
good
I
started
working
on
migrating
GS
block
over
to
it.
That's
almost
done
the
test
passing
we'll
coverage.
There's
just
a
few
atoms
to
work
out.
I
need
to
expose
some
some
endpoints
that
already
have
some
codecs
loaded
and
I've
got
updated,
Doc's
and
a
couple
other
things,
but
other
than
that.
E
G
I
was
late;
okay,
it's
so
a
shouldn't
week
for
me,
probably
holiday
and
and
yeah
just
that
little
things
again,
lonely
those
weeks,
but
the
main
highlights
were
Bitcoin
docks.
There's
a
pull
request
with
my
work
in
progress.
It's
tedious
work,
busy
progress
and
the
other
thing
I
encountered
was
see
boards
noticeable
strictness
thing
which
I
thought
we
were
done
with,
but
it
was
an
issue
in
bork,
a
suitable
person.
Somebody
filed-
and
that's,
if
you
know,
47
turns
out
that
book
has
a
will
pass.
G
Multiple
top-level
objects
in
a
row
as
an
implicit
array
and
yes,
I
feel
the
taxi
boy
is
doing
this
and
only
fetching
the
first
object.
So
if
you
have
a
dag
seat,
bought
object
block
then,
and
you
and
you
can
have
multiple
top
level
objects
just
smush
together,
we
without
a
container
so
I
introduced
poor
quest
for
the
specs
to
say:
that's
not
allowed
for
the
XE
board
and
poor
request
the
to
jeaious
take
Siebel
codex
to
fix
that
that
was
that
was
me.
A
H
A
So
so
in
case
you
start
looking
to
rust.
I
fell
d,
surely
check
out
the
kernel
only
open
PR
because,
like
that's
the
let's
see
a
major
change
that
gets
in,
and
so
it
just
makes
sense
to
basically
base
it
on
that
one
because,
like
it's
you
actually
getting
in
and
but
this
will
hopefully
lend
this
week
so
yeah.
A
Cool
yeah
so
I'm
happy
to
get
feedback
because
I
get
still
like
it's
like,
based
on
my
usage
and
usage
from
from
David,
the
other
orphans
kind
of
like
yeah,
pretty
vague.
If
it's
a
good
idea
or
not
what
we
are
doing,
and
so
it's
good
to
get
feedback
and
just
check
to
see
how
it
really
works
in
real
world
applications.
So
that's
it's
great,
so
I'm,
good
feedback,
end,
and
certainly
so
certainly
in
Russ
I
could
lead.
Nothing
is
set
in
stone.
So
I'm
like
I'm,
happy
to
change,
basically
everything.
B
B
B
That's
my
point
basically
I
want
to
make
sure
that
we're
not
going
to
end
up
with
something
that's
specific
for
a
foul
coin,
because
we
didn't
have
the
the
forethought
to
just
say.
You
know
this
thing:
we're
not
allowing
right
now
and
if
you
need
it,
come
talk
to
us
because
we
have
the
library
yeah.
E
So
something
that
came
up
in
my
college
air
today
that
is
probably
actually
shared
more
broadly
than
just
my
phone
call
with
Eric
I,
think
I
already
told
Peter,
though
we
so
we
need
to
work
on
a
path
spec.
We
need
a
spec
for
pathing
I
want
to
pull
this
out
of
the
conversation
about
selectors
like
entirely
for
a
minute
and
just
think
about
paths,
because
we
continue
to
like
ship
around
paths
and
talk
about
paths,
and
we
actually
don't
have
a
specification
anywhere
for
what
they
are.
E
E
H
E
So
it's
really
just
a
way
to
say:
like
okay,
I
have
some
kind
of
field
view
graph
somewhere
and
I
want
to
traverse
through
it
to
hit
a
particular
point,
and
so
here
is
like
a
string
that
gives
you
the
path
to
that
thing
and
there's
there's
like
we
should
talk.
We
we
should
think
about
I.
E
E
So
I
mean
like
this.
This
came
up
like
this,
like
in
the
last
few
weeks,
just
because,
like
we,
we
need
in
the
Lotus
command
line.
We
need
a
way
to
say
in
the
retrieval
market:
go
and
get
me
this
file
out
of
a
car
file,
and
you
need
to
be
able
to
pass
at
a
path
which
is
basically
relative
to
the
content
route
of
the
car
file
that
pads
into
that
car
file
to
resolve
to
an.
G
E
Fs5,
and
so
like
this
is
lecture
like
relatively
simple
in
in
the
car
files
that
we
wrote
for
Dumbo
drop,
because
it's
always
just
like
an
offset
in
an
array
and
that's
it.
But
you
can
imagine
like
other
structures
where
you
you
have
like
more
complicated
graphs
on
top
of
a
bunch
of
files
and
servers
into
little
bit
more
yeah
I
see
that
yeah
Peter
just
posted
like
a
good
example
in
the
chat
I'm
like
what.
What
a
really
basic
path
looks
like.
G
E
This
is
something
that
like
actually
came
up
in
this
work
right
is
that
the
way
that
selectors
are
designed
is
that
they
are
broad
pattern-matching
over
a
graph
that
resolve
to
any
number
of
nodes
right.
So
once
you're
looking
at
a
selector,
particular
parse
selector,
it
doesn't
have
a
logical
end
like
it's,
not
linear.
You
can't
like
up
to
end
something
to
a
selector,
because
there's
no
end
right,
like
it's.
E
It's
very
similar
to
grass
QL
right
like
if
somebody
was
like
handed
you
a
graphical
query
and
you
were
like
oh
I
want
to
append
something
to
it
like
there's
no
place
to
append
it
to
like
the
selection
criteria.
Various
places
throughout
the
graph
right.
So
selectors
are
like
this
just
like
kind
of
mega
thing
in
fact
like
what
what
I
would
like
to
see
actually
is
selector
is
somewhere
once
we
have
a
good
selector,
which
we
sorry
once
we
have
a
good
spectra
paths
for
the
selector
spec
to
leverage
the
path
back.
E
So
you
can
put
the
pads
in
selectors
to
make
them
a
little
bit
easier
to
write,
but
selectors
yeah,
they're
they're
not
constrained
enough
for
us
to
use
them
in
certain
places
like
like
in
this
case
this
has
to
resolve
to
a
single
file.
So
we
have
to
know
that
this
is
owned.
This
is
a
linear
traversal
that
resolves
to
one
point,
and
we
don't
have
that
like
anywhere
else,
whether.
E
H
C
C
That's
not
a
definable
block
is
one
major
distinction
right,
whereas
you
clearly
can
append
to
a
path
and
I
actually
will
end
up
using
paths
when
I'm
implementing,
selectors
and
traversals
because
say
I'm
doing
some
traversal
over
a
large
piece
of
graph
and
I'm
evaluating
a
selector
as
I.
Do
this
right
at
every
point
in
my
traversal
I
will
have
on
my
stack
a
path
and
just
in
half
right.
C
So
it's
like
it
is
arguably
a
subset,
but
it's
a
very
useful
subset
and
at
any
point,
I
think
it's
particularly
useful
to
be
able
to
take
that
path
and
serialize.
It
like
I,
want
to
be
able
to
printf
debug
my
way
through
these
situations
and
hopefully
that
string
that
comes
out
of
my
printf
debugs.
That
should
also
be
a
reasonable
thing
for
me
to
reinterpret.
C
As
a
path
again
in
a
way
that
I
can
use
that
as
an
instruction
to
make
a
specific
traversal
to
a
specific
point,
so
some
of
the
parts
that
we're
missing
right
now,
like
all
of
our
libraries
that
exists,
they
do
this
thing
where,
like
you've,
got
a
path
on
the
stack
and
something
you're
products
we
haven't
defined.
All
the
escaping
basically
and
I.
Think
that's
something
that
we
all
agree
that
we
need
to
do
soon.
A
E
E
E
E
Gonna
want
to
use
this
command
line
flag
to
pad
into
it.
There's
there's
no
point
at
which,
like
you
you
hand
like
a
directory
to
the
Lotus
CLI,
and
it
like
does
all
that
stuff
for
you,
like.
It's
actually
really
really
minimal
right
now
and
the
more
advanced
stuff
it's
like
people.
People
are
actually
like
exporting
a
car
file
and
ipfs
loading
up
that
car
file
and
then
like
taking
that
car
home
and
they
get
back
to
IP
address
yeah.
So
it's
it's
pretty
clear
and
I
mean
we're.
E
This
also
just
needs
to
be
documented
well
across
style
coin,
because
even
beyond
the
little
CLI,
the
retrieval
market
is
taking
selectors
in
order
to
do
selection
into
these
car
files
and
if
they
do
contain
big
UNIX
nest
directories.
The
they're,
like
that
the
algorithms
for
traversing,
like
ipfs
directories,
is
not
actually
in
file
coin
in
Lotus.
So
that's
not
going
to
be
supportive.