►
From YouTube: 🖧 IPLD weekly Sync 🙌🏽 2020-07-20
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
desync
meeting
edge
tool,
either
20
2020
and
as
every
week
we
go
over
the
stuff
that
you've
worked
on
in
the
past
week
and
then
discuss
journey
items
that
we
might
have.
So
if
you
have
any
items,
please
add
them
to
the
agenda
and
I
start
with
myself,
and
it
was
again
one
of
those
weeks.
I,
don't
know
what
I've
spent
my
time
on
really
so
it
probably
wasn't
mostly
fyke
related
and
but
on
the
iPad
ISA
I.
A
It
was
I
looked
into
the
details
of
tiny
multi-edge,
which
is
another
Martinez.
Implementation
rests,
which
used
to
be
a
PR
on
Martinez.
It's
this
own
creator
and
each
is
two
for
those
people
to
move
quicker.
But
the
idea
is
to
take
most
of
those
ideas
and
get
it
back
into
the
proper
rusts,
not
here
upstream
and
and
it's
quite
complicated
and
I'm
still
looking
into
if
it's
yeah
really
cool,
really
smart
or
just
crazy,
but
I'm
trying
to
figure
this
one
out
and
I'm
close.
A
But
it
has
a
few
cool
ideas
and
certainly
also
like
API
wise.
It's
it's
way
better
and
then
kind
of
all
this
work
was
triggered
by
I
was
looking
into
last
file
implementation
because
they
currently
use
a
fork
of
all
their
own
version
of
iPod
II
and
what
needs
to
be
done
in
order
to
get
them
upstream
on
the
multi-format
stuff,
as
well
as
on
the
upstream
ILD,
and
it's
kind
of
interesting
because
there
it's
like
they
have
a
like
one
of
the
few
you
world
use
cases
that
use
rust.
B
B
B
Spent
some
time
on
just
to
not
for
any
good
reason,
but
just
because
it
was
something
that
I
could
do
while
I
was
relaxing
was
my
JavaScript
C,
boring
corner
and
decoder
just
to
tinker
with
it
and
uhm
was
playing
with
numbers
and
running
into
the
classic
JavaScript
number
problems
and
discovered
that
bork
has
interesting
behaviors
with
numbers
that
we
probably
should
be
documenting.
I've,
put
notes
in
there
in
the
meeting
minutes,
just
basically
anything
above
2
^,
53
javascript
has
difficulty
handling
with.
B
B
So
you
know
they're
like
above
that
value
used
to
use
you
seriously
loose
functionality,
so
you
shouldn't
try
and
represent
numbers
safely
value.
There's
been
various
efforts
to
deal
with
that
in
Bourque,
our
JavaScript
C
bore
in
current
Dakota
for
exe
bore
it
uses
a
dependency
called
big
number
J
s
to
deal
with
these,
and
so
you
can
pass
at
these
big
number
objects
and
it
will
deal
with
them
properly
and
also,
if
you,
if
you
decode
numbers
large
enough
it'll,
give
you
this.
These
big
number
objects.
B
I
think
I
think
that's
right,
maybe
not
actually
anyway,
but
if
you
give
it
a
integer
to
encode
that
is
above
max
a
future,
then
it
will
include
it
as
a
float
in
C
bore
so
with
major
major
number.
Seven,
so
it'll
be
don't
say
it's
a
float,
it'll
kind
of
come
out
as
any
other
language.
As
a
float
now,
arguably
that's
correct
for
JavaScript,
because
I
give
Li,
there
is
no
integer
in
JavaScript
and
above
that
amount,
it's
you
know,
that's
a
float,
so
that's
kind
of
messy.
B
There
are
ways
to
deal
with.
We
have
this
new
bigand
object
in
JavaScript,
it's
white
Mike
working
its
way
through
JavaScript
engines
that
deal
with
some
of
that,
not
all
of
it
but
yeah.
There's!
That's
just
an
interesting
challenge.
Point
of
what
do
you
do
above
max
a
safe
future?
You
do
have
choices,
but
it's
not
clear
that
one
choice
is
better
than
another.
B
So
there's
that
there's
some
just
stuff
in
the
meeting
notes
about
it
that
I
I
put
out-
and
I
I'm
building
in
big
in
support
into
my
thing
early
on,
but
and
assuming
by
the
time
we
makes
its
way
out
of
our
stack
that
all
the
major
engines
will
support
and
final
reminder
in
there
is
that
just
because
it
we
keep
on
stumbling
up
against
this
refund
the
one
for
go
still
does.
Does
the
old
floats
at
sixty
four-bit,
whereas
in
JavaScript
it
goes
for
the
floats
as
smallest
possible,
accurate
representation?
B
So
we
still
have
a
difference
in
the
way
the
two
languages
deal
with
it
and
we
need
to
resolve
that
either
one
way
or
the
other
and
I
did
why
I
was
tempted
to
going
to
go
and
see
what
it
would
be
like
to
make
that
make
it
do
the
small
as
possible,
but
beyond
my
depth
and
also
not
sure
that
that's
the
way
we
want
to
do
it
cuz.
Maybe
we
want
to
do
64-bit
and
just
be
done
with
it,
so
anyway,
choices
choices,
but
some
of
this
stuff
needs
to
be
in
documentation.
B
A
D
Yeah,
okay:
now
what
did
I
do?
Alright,
so
Doc's
did
some
dock
work
got
the
dock
side
out,
there's
a
tutorial:
it's
got
multi-line
code
examples
or
the
ability
to
have
multi-line
code
examples.
We
still
need
to
get
them
on
there
doing
more
of
that
this
week
as
well,
we
landed
the
as
CID
changed
to
multi
formats,
which
had
been
a
little
bit
controversial,
but
after
I
talked
with,
goes
all
and
got
my
head
around
and
again.
I
realized
like
okay.
Yes,
this
is
a
better
way
to
do
this
and.
D
Addition
to
being
a
better
way,
it
actually
solves
another
issue
that
we
have
so
I.
Actually
I
just
explained
this,
because
I
think
people
here
do
need
to
kind
of
think
about
it
a
little
bit,
especially
for
anything
that
we
ever
do
that
touches
JavaScript.
So,
in
order
for
things
to
really
easily
go
between
worker
boundaries,
they
need
to
be
simple
types
and
a
few
other
things
that
will
serialize
other
than
just
Jason
types.
Symbols
don't
serialize
across
that
boundary,
so
we
were
using
symbols
to
do
the
CIE
check
and
that
doesn't
work
well
goes.
D
All
figured
out
is
like
this
very
clever
hack,
which
is
make
this
s
yeah,
put
an
add
the
CID
property
on
the
CID,
there's
a
reference
to
itself.
So
it's
a
circular
reference
in
a
predefined
property.
That
means
that
it
doesn't
serialize,
so
it
doesn't
actually
conflict
with
any
existing
space.
We
don't
have
to
reserve
like
this
property
or
anything
in
the
serialization,
but
it
it
does
mean
that,
like
you
know,
we
can
safely
push
these
across
the
worker
boundary
and
then
what
you
want
to
do
is
instead
of
doing
an
in
CID
check.
D
You
want
to
do
this,
use
this
as
CID
method
that
actually
returns
you
either
the
CID
at
that
reference
or
another
interesting
thing
that
it
can
do
is
that
it
will
return
you
that
CID
bound
to
the
current
CID
class.
So
this
is
a
really
big
issue
with
the
new
multi-format
stuff,
because
if
I'm
in
my
program
and
I
want
to
check,
if
this
is
a
CID,
I
actually
want
it
to
be
a
CID,
it's
bound
to
my
multi-format
stable,
not
somebody
else's
multi-format
stable.
D
So
what
this
will
do
is
actually
like
always
return
me,
a
CID,
then
it's
bound
to
the
right,
multi-format,
stable
or
will
return
false
or
no,
if
it's
not
a
CID.
So
this
means
is
like
across
all
of
our
downs.
We're
gonna
have
to
replace
all
these
in
CID
checks,
but
this
is
like
a
much
better,
safer
method
and
it
works
really
well
with
workers.
So
that's
a
good
change.
Over
the
weekend,
I
updated
a
bunch
of
our
code
X
and
the
block
interface
to
use
this
as
well.
D
D
So
Ron
I,
don't
know
if
you
saw
this
yet,
but
the
idea
is
actually
pretty
simple
you.
Basically
you
do
all
you.
You
basically
say
in
the
header:
there's
another
property
now
next
to
routes
called
footer
and
then
you
can
give
it
like
a
footer
version
or
we
can
do
some
kind
of
capabilities
or
whatever.
But
you
check
the
header,
and
you
say:
oh,
this
has
a
footer
and
then
you
look
and
the
last
block
ever
encoded
is
a
Rob.
It's
a
raw
block
of
nine
bytes
with
sha
to
256.
D
So
that
is
a
guaranteed
fixed
length
in
the
footer
to
get
that
whole
raw
block
address
out
and
that
last
nine
bytes
is
a
VAR
int
and
it's
the
VAR
int
for
the
manifest,
which
is
the
block
right
ahead
of
that.
So
that
allows
you
to
look
at
the
footer
and
get
the
manifest
out
of
the
foot
without
breaking
the
current
file
format
at
all
or
changing
involved.
D
D
It's
really
ideal
for,
like
this
particular
performance
case,
so
yeah,
that's
fun,
to
look
at
I
want
to
give
that
on
people
during
their
so
that
they
can
think
about
it,
because
I
am
now
start
iterating
on
this
footer
and
we've
all
talked
about
it
for
quite
a
while.
Now,
so
that's
in
a
repo
and
then
just
a
minute
ago,
I
sent
a
PR
for
the
FBL
teams
that
we
talked
about
I.
Think
last
week
that
Peter
brought
up.
They
basically
just
allows
you
to
do
inline
bytes
when
you're
in
this
Union.
D
E
E
Please,
which
is
about
it,
uses
the
the
journey
for
how
we
want
UNIX,
invest
v2
to
work
through
something
like
ipfs
to
introduce
all
the
different
features
that
we've
extracted
and
defined
for
that
quest.
So
they
kind
of
took
a
bunch
of
the
stuff
that
we
talked
about
in
our
last
weekly
meeting
about
like,
and
this
is
why
you're
going
to
need
to
like
enumerate
these
ad
O's
they'd,
be
like
squawks.
Do
that
so
I
hope
that's
going
to
be
good!
Well,
it
would
love
reviewing
multiple
reads.
Of
course,
yeah.
A
A
F
So
I
have
been
seen
almost
in
Hong
Kong
file.
Coin
lands.
The
stuff
interesting
for
ideology.
Is
that
I
just
bit
the
bullet
and
actually
landed
the
changes
to
the
data,
see
ADP
CIT
and
CEO
commitments
and
all
this
kind
of
stuff.
So
this
is
now
Hong
chain
is
working
against
uncovered
some
minor,
very
minor
issues
with
the
blockchain,
but
it's
not
fixed
rot
by
the
way
I
believe
JSC
ID
is
not
yet
updated
to
understand
this
two
pieces
I
know
my
idol
reached
out
your
now
to
pay
stats
for
the
CID
decoder
sites.
F
It
doesn't
understand,
then
you
say
this
yet,
and
I
also
try
to
do
an
update
of
go
car
which
is
actually
way
more
complicated
than
it
should
be
because
of
dependency
chain
of
that
not
being
very
healthy
and
that's
progress.
Unfortunately,
and
also
I
had
number
of
conversations
with
Adam
about
I
didn't
see
id
length
and
stuff
like
that,
actually
probably
two
hours
to
mostly
talking
about
this.
We
think
that
we
have
a
better
idea:
how
to
reorganize
the
table
to
make
it
clearer
to
an
implementer.
D
F
My
my
thing
is
actually
the
other
way
because
markdown
is
still
like.
Yes,
it's
it's
easy
to
pass,
but
it's
not
great
for
machines,
so
my
thing
is:
it'll
actually
be
a
JSON
which
spits
out
the
markdown,
the
CSV
file,
so
on
and
so
forth,
but
this
one
will
be
the
machine
possible
thing.
That
is
the
point.
Well.
D
The
markdown
is,
the
markdown
is
easier
for
us
to
take
pr's
on,
and
so
we
have
automation
that
generates
the
rest.
Then
it
would
actually
be
cleaner
to
have
the
reference
in
markdown
and
then
we
can
spit
out
the
CSV
in
JSON
I.
Actually
like
I
see
women.
Let
me
see
I
don't
have
time
to
do
today
might
be
able
to
poke
this
today.
D
B
B
F
A
B
D
Doesn't
render
it
nicely
when
you
have
a
long-ass
description
like
we
have
in
a
bunch
of
dummy
that,
like
that's
what
messes
up
look
at
the
table?
That's
why
the
tables
in
markdown
look
a
little
bit
nicer.
Also
I
will
say
that,
like
your
editor
may
not
like
it,
but
the
diffs
in
github
look
really
nice
when
you're
using
markdown
and
not
as
much
now.
F
D
F
B
D
I
think
I
mark
no
table
is
only
reasonable
if
we
can
keep
the
line.
Links
low,
but
I
think
that
we
can
keep
the
line
links
low
in
markdown,
because
we
can
link
lower
in
the
file
to
have
more
details,
and
we
don't
have
to
put
everything
into
like
one
line.
That's
it's
the
description,
because
the
description
field
is
like
completely
unnecessary
for
the
CSV
in
the
jail
like
nobody,
I.
A
B
Will
it's
only
that
I'm
skeptical?
It's
also
that
some
of
these
things,
sometimes
it's
good
to
be
evolutionary
rather
than
revolutionary,
and
it's
CSV
now,
but
you
basically
saying
you
want
to
throw
that
away
and
go
to
marketing
table
and
then
transform
it
into
Jason,
like
maybe
evolutionary,
would
be
the
best
approach
for
something
that
is
heavily
used.
It's.
G
F
D
F
Anything
but
the
extra
stuff
will
go
into
markdown
and
the
JSON
and
stuff
like
that,
so
that
new
code
can
actually
have
a
spot
to
discover
new
tags.
As
you
know,
as
we
progressed
now
in
the
Senate
progressed
over
there,
like,
for
example,
we
will
at
some
point
have
to
dumb
stuff
like
minimums
which
we
will
have
to
update
going
forward.
So
things
like
that.
F
A
D
I'll
cover
the
dark
stuff,
so
I
I
just
fixed
the
permission
on
the
repo,
so
we
should
all
have
permissions
to
push
in
that.
Let
me
know
if
you
don't
have
push
permissions
in
the
docs
repo.
D
The
idea
right
now
is
like
any
doc
any
doc
that
you
have
an
idea
about
just
create
a
stub
for
it
and
there's
like
a
drives
folder
that
Eric
created,
and
even
if
you
just
have
like
a
title
in
the
description,
that's
fine
like
you
go
and
fill
it
it
later,
but
like
there
should
be
no
barrier
to
just
like
writing
a
doc
and
getting
a
doc
into
the
new
site.
It
can
be
about
anything.
D
It
can
be
any
piece
of
content,
we'll
take
stock
of
all
of
this
in
August
and
try
to
organize
it
all
and
figure
out
like
where
things
are
going
and
what
the
kind
of
user
stories
are.
But
right
now
like
we
just
we
just
need
to
write
content.
So
anybody
that
has
any
extra
time
or
an
idea
for
a
thing
that
they
think
should
exist.
Please
write
it
as
soon
as
possible
and
then
we
will,
you
all
sit
back
up
in
August
to
really
power
through
it.
D
Mean
push
the
master
like:
don't
don't
send
PR,
don't
do
branches,
just
push
the
master,
anything
that
you
want
to
do
and
then
we
and
we're
not
gonna
do
edits
yet,
like
we'll
do
anything
in
a
later
phase.
It's
now,
like
you
know
in
that
first
draft
I'll,
just
get
all
the
words
out
that
we
can
think
of
and.
F
So
I
brought
this
up
so
earlier
today,
I
because
office
hours.
Somebody
said
something
like
you
know,
ideally
is
not
quite
ready
yet
for
devs
and
Chris
can
jump
to
that.
I.
Don't
understand
why
they're
saying
this
is
like
so
radiant
you
should,
you
know,
be
promoting
for
people
to
be
using
it
and
stuff
like
that,
and
my
counter
to
this
was
like
well,
we
don't
really
have
any
type
of
conversions
story
at
all.
D
Mean
it
actually
is
being
used.
A
lot
is
the
thing
it's
just
like
Matt
I
mean
the
docks.
Are
the
big
one
honestly
like?
We
don't
have
a
great
way
to
explain
these
things
to
people,
especially
in
the
context
of
our
new
or
libraries.
So
that's
like
the
big
thing
like
just
getting
the
docks
out
of
the
way,
but
a
lot
of
the
libraries
are
already
in
in
use.
D
C
C
There's
no
representation
from
my
building
team.
That
kind
of
provides,
updates
or
lightning
talks,
or
anything
like
that.
So
I
think
there's
a
lot.
There's,
not
a
clear.
You
know,
idea
of
kind
of
how
fits
in
so
I
think
that
comment
actually
came
from
someone
on
the
ipfs
team.
I
Maru
was
Peter
but
basically
said:
oh
yeah,
we're
not
I
feel
the
one
else
was
using
yet
because
it's
still
in
progress
or
something
and
so
I
think
one
one
thing
in
I
mean
it's
nice
to
hear.
We
probably
all
feel
obviously
is
more.
C
We
could
do,
but
despite
documentation
that
helps
them
kind
of
like
use
it
effectively
the
codes,
not
bad.
It's
not
like
it's.
You
know,
there's
a
lot
that
can
be
used
in
this
being
use.
I.
Think
people
just
need
a
better
understanding
of
how
they
can
put
you
using
what
it
is
and
I
think
you
know.
One
thing
we
can
think
about
is:
can
we
try
to
cue
something
up
for
those
lightning
talks
or
whatever
I'm,
not
sure,
because
we
kind
of
start
more
of
a
pop
for
layers
can
be
kind
of
hard
to?
C
You
have
demos
on
that
sometimes,
but
at
least
share
something
out
there.
Mike
Mike,
good
and
I
know
it's
talking
to
Michael
this
morning.
Just
I
think
there's
a
lot
of
developers
out
there
that
don't
need
don't
really
get.
What
content
dress
storage
is
it's
pretty
easy
to
say:
hey
I'm
gonna
file,
my
file
system.
Let
me
stick
it
into
IB
FS,
but
it's
another
thing
to
think
about
how
do
I
create
a
complete
application?
C
A
C
D
I
think
I
I
can
kind
of
see
where
someone
on
one
of
the
ipfs
teams
would
be
coming
from,
because,
on
the
one
hand
we
have
two
new,
like
next-generation
libraries,
affright
Kilby
and
go,
and
in
JavaScript,
both
of
which
are
being
used
by
some
people,
but
not
by
those
ipfs
implementations,
and
they
did
a
bunch
of
work
to
add
this.
Dag
API
like
before
I
peeled
II,
really
had
these
libraries
and
we've
been
actively
telling
people
like
don't
use
those
dang
API.
Is
there
like
not
great
like?
D
Instead,
you
should
use
like
and
I
filled
the
librarian,
and
then
we
don't
have
proper
documentation
anywhere
on
here's.
How
to
use
our
new
I
filled
the
library
that
we're
encouraging
people
to
use,
but
then
to
put
that
data
into
ipfs.
It's
like
actually
like
a
specific
doc
that
Edina
asked
it's
like.
This
would
really
help
in
a
lot
of
the
back
of
that
stuff.
In
particular,
we
should
really
get
this
written,
so
I
think,
like
my
my
tasks
like
this
week,
is
to
get
the
getting
started
in
JavaScript
like
tutorial
ready.
D
D
The
right
way
and
I
feel
be
into
ipfs,
probably
with
the
block
API
most
likely,
but
you
know,
like
here's,
actually
how
you
get,
because
most
people
that
are
using
agility
right
now
are
taking
data
and
stickyman
ipfs
so
like
we
need
to
bridge
that
thing,
because
we're
saying
don't
use
the
daga
api,
but
they
do
so
yeah
I.
Think,
like
part
part
of
this,
is
it
is
documentation
and
then
also
I.
D
C
One
other
relay
thank
you,
I
placed
a
link
in
the
notes
there,
but
there
is
some
issue,
as
someone
recently
filed
about
kind
of
like
there's
this
big
SJS
ipfs
light
from
textile,
which
is
essentially
kind
of
like
a
network
player
data
layer
which
they
build
the
unite
PFS.
It
doesn't
have
all
the
other
things
that
come
in
with
the
flow
I.
Give
this
install
kind
of
the
idea
here
is
that
there's
a
lot
you
can
do
with
this?
C
It's
not
the
bear
bit
on
them
and
I
think
you
know
I,
don't
know
it's
like
that.
Confess
group
is
focused
on
I
guess
not.
This
is
too
low
level
of
an
issue
for
them.
They're
kinda
like
trying
to
solve
application
issues,
but
this
got
more
like
an
expansion
of
the
infrastructure.
There's
like
a
little
mini,
you
know
blue
mini
infrastructure
set.
You
need
to
do
just
basic
stuff
with
IP
LD,
without
having
to
like
understand
everything.
Oh
I
give
s
and
deals
all
over.
C
D
Actually,
really
nice,
the
I
think
that's
white
stuff,
it's
actually
it's
pretty
nice
we
may
in
our
documentation,
want
to
to
mention
both
but
have
examples
of
working
with
both
when
we
do
the
getting
stuff
and
I'll
give
us,
because
it
is
really
good
and
the
story
on
the
JavaScript
side
at
least,
is
actually
a
lot
better,
because
we
they
are
using
our
new
libraries
in
JSF.
People
is
quite
unlike
the
full
JSF
BFS
right
now.
D
F
Guess,
I
guess
one,
but
one
thing
that
kind
of
dovetails
with
what
Chris
said.
Is
that
a
lot
of
the
folks
who
are
asking?
How
do
I,
you
know,
stop
my
data
and
why
don't
you
properly?
They
come
from
the
ipfs
angle,
which
implies
that
and
I'll
be
able
to
use
with
the
gateway
which
is
there
is
deciding
on
the
case
today.
Right.
D
If
they
want
files,
they
should
use
idea
fast,
like
we
shouldn't,
we
shouldn't
tell
them
to
use
I
feel
D
if
they
want
files
like
that
is
kind
of
like
what
passes
menthol
like.
So
you
know
it.
The
gateway
could
return
them
block
data,
but
I,
don't
think
they're
gonna
get
like
a
ton
of
use
out
of
the
the
Gateway,
if
they're
not
going
files
and
if
they're
doing
files-
and
we
should
point
them
and
I-
think
this.