►
From YouTube: 🖧 IPLD weekly Sync 🙌🏽 2021-03-29
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
I've
linked
the
prs
in
draft
prs
in
the
meeting
nodes
yeah
still
ongoing.
I
still
hope
to
get
at
least
the
unix
fs
part
done
this
quarter,
which
is
not
that
many
days
left
we'll
see,
and
but
what
is
interesting.
There
were
also
changes,
or
I
propose
changes
on
js
multi
formats
and
it's
interesting
because
so
for
people
that
are
not
that
deep
into
javascript
but
or
for
people
into
javascript
is
interesting.
A
So
there's
two
ways
to
load
modules:
there's
the
node
way
and
there
is
the
new
es6
modules
loading
way
and
so
for
the
js
multi-format
stuff
and
the
new
ipodle
stuff,
michael
and
also
I
read
with
him.
It
makes
sense
to
use
the
new
module
stuff
because,
like
this
is
what
people
will
use
in
the
future,
and
but
we
of
course
need
to
be
compatible
with
the
old
stuff,
because
this
is,
for
example,
what
js
ipfs
is
using
and
then,
of
course
we
had
this
effort
in
js
ipfs
that
added
the
typescript
typing.
A
But
now
you
also
need
to
make
sure
that
typescript
types
work,
but
the
problem
is
you
can't
it's
really
difficult
to
get
them
working
with
the
new
stuff,
as
well
as
with
the
old
stuff,
and
so
we
had
problems
with
those
things,
and
what
we
now
ended
up
with
is
that
in
the
module
system
you
have
something
like
it's
called
default:
export,
which
means
a
single
file,
exports,
a
single
thing
and
not
multiple
things,
and
normally
you
are
happy
to
use
this,
but
in
our
case
it
turned
out
like
it
doesn't
work
well
with
the
typing
and
everything.
A
So
the
easiest
solution
that
goes
allah
proposed
was
just
don't
use
default
exports,
but
just
export
it
by
name.
So
it's
kind
of
like
a
technical
detail
and
like
if
you
then
import
things,
it's
a
bit
diff
different
and
might
not
be
as
nice,
the
inputs,
but
I
think
it's
totally
workable
and
I
also
found
so
when
I
did
the
research
from
blog
posts
about
like
people
just
well.
You
shouldn't
use
default
exports
at
all,
so
there's
also
those
kind
of
people,
so
it's
kind
of
like
it's
kind
of
okay
to
use
so-called
named
exports.
A
This
is
what
we
do
now,
and
the
good
news
now
is
that
it's
really
smooth
to
use
it
with
a
new
model
system
with
the
odd
module
system,
as
well
as
with
with
the
typescript
types,
and
you
don't
get
any
strange
error
messages.
You
don't
understand
so,
and
this
is
about
to
be
merged
with
a
few
discussions
about
the
naming
and
so
on,
but
hopefully
this
js
multiformats
pr
will
be
merged
today
or
tomorrow,
because
I
kind
of
need
it
for
my
work
and
I
think,
moving
forward.
A
We
probably
will
use-
or
I
suggest
that
we
use
for
our
json
modules-
always
named
exports,
because
it's
just
smoother,
so
I'm
not
saying
that
it's
not
possible
with
default
exports,
but
it's
just.
We
would
need
some
way
to
make
it
work
somehow.
It
might
then
break
somewhere
else,
and
this
was
just
the
cleaner
approach.
I
think-
and
it
seems
to
work
so
far
and
also
like
to
make
it
even
more
complicated.
A
If
you
use
some
bundlers,
they
also
mix
up
things,
so
it
then
depends
on
if
using
a
webpage
or
parcel
or
some
other
one
and
so
yeah.
It's
really
complicated
and
I
think
now
we're
at
a
stage
where
it's,
I
think,
kind
of
works
on
all
the
things
that
we
tried
it
on
and
yeah.
The
other
typing
passes,
as
well
as
the
tests.
So
that
looks
promising.
B
Hello,
so
I
have
not
super
many
updates
this
week,
but
a
couple
of
small
ones
that
are
nice:
the
legacy
go
ipld
format.
Repos
have
finally
got
a
status
text
in
them
a
little
section
of
the
readme.
That's
saying
hi
by
the
way,
there's
newer
alternatives
and
pointing
a
link
so
will
started
a
pr
for
that.
B
I
added
a
little
more
diff
on
top
of
it
as
well
that
I
have
an
idea
that
I
want
to
try
out,
and
so,
if
anybody
has
like
feelings
on
this
and
wants
to
comment
later,
please
do
no.
We
didn't
do
it
for
other
repos.
In
response
to
question
the
chat,
we
should,
because
we've
got
a
bunch
of
repos
that
probably
need
some
sort
of
update
to
their
status
text.
B
In
my
wildest
dreams,
maybe
we'll
write
a
little
script
that
goes
through
repos
and
checks.
Who
has
these
bold
text
tags
available
and
then
we
can
build
dashboards
not
going
to
do
that
this
week,
but
it's
an
idea.
Okay,
so
thank
you
will
for
kickstarting
that
there's
an
issue
that's
been
opened
on
the
go.
I
killed
prime
repo
about
the
link
context,
abstraction
if
anybody
is
interested
in
that
and
has
opinions
about
the
friction
of
that
or
has
run
across
it,
especially
in
performance
profiling.
B
It
would
be
interesting
to
get
more
information
on
that.
Otherwise,
maybe
we'll
resume
that
thought
later
and
one
thing
I
want
to
note
from
others
that
are
not
present
in
the
call,
but
I
know
are
doing
awesome
stuff.
This
go
unix,
fs,
node,
repo,
that
I
keep
mentioning
in
passing
over
the
weeks.
B
I
believe
it
is
going
to
be
going
a
little
bit
further.
The
team
that
is
working
on
upgrading
a
bunch
of
ipfs
internals
go
ipfs.
Internals
they've
realized
that
their
road
map
for
upgrading
go
ipfs.
Internals
requires
them
to
have
this,
go
unix
of
s,
node
library
and
be
at
least,
I
think,
able
to
read
the
unix
of
sv1
started
directories.
So
that
feature,
I
think,
is
intended
to
make
it
into
adl
land,
soon
very
exciting,
yeah,
that's
about
it!.
A
Thanks
next
is
daniel.
A
C
Where's
my
tab,
so
I'll
be
brief.
I'm
still
I've
still
been
doing.
Work
on
dac
pb,
so
spec
wise,
we're
still
looking
good.
I
did
mention
on
slack
briefly
that
there
was
one
test
that
was
failing
and
it
turns
out
that
some
people
did
write
their
own,
their
own
pro
buff,
encoders
and
decoders
for
pb
nodes,
and
they
used
a
different
order.
C
So
we
ended
up
agreeing
to
still
have
a
must
for
a
canonical
format,
but
then
be
more
lenient
on
decoding
so
that
we
actually
keep
backwards
compatibility
with
ipfs
data,
so
the
spec
change
is
there
and
I
also
updated
the
code
and
that
setting
on
my
laptop
because
I've
got
another
pr
and
I
don't
want
to
enter
into
conflicts
with
myself
and
we
don't
live
in
a
world
where
I
can
push
a
chain
of
prs.
Unfortunately,
the
other
pr
that's
causing
this
chain
is
a
bunch
of
performance
improvements
in
gokodectpd.
C
C
So
I
I
did
something
a
little
cheeky,
which
is
use
a
package
called
proto
wire
from
the
new
pro
above
library
and
it's
well
optimized,
and
it
has
some
primitives
like
encode
variant
and
consume
variant
and
stuff
like
that.
So
I
was
able
to
remove
a
bunch
of
code,
it's
faster.
So
now
the
overhead
is
practically
nothing.
It's
still
it's
a
little
bit
of
overhead,
because
ipld
prime
still
has
a
little
bit
of
overhead,
and
I
will
talk
about
that
later,
but
it's
practically
non-existent.
C
Now
it
is
a
an
extra
dependency
on
co
on
go
codec
deck
pv.
But
I
reckon
that's
fine,
because
if
anybody
is
importing
goku
deck
pb,
I
think
importing
the
product
of
libraries
is
reasonable
and
if
anybody
thinks
that's
not
the
case,
please
shout
and
lastly,
in
terms
of
ipld
prime,
the
performance
is
fine.
I
I'm
think
I'm
finding
that
the
last
20
of
performance
that
I
can't
gain
back
is
basically
alux.
C
So
you
look
at
the
cpu
profile
and
just
the
top
20
is
just
runtime
functions
and
it's
essentially
just
allocations
that
I
can't
get
rid
of
and
there
it's
kind
of
sad,
because
they're
really
small
allocations
that
maybe
we
could
get
rid
of
one
one,
for
example,
is
each
pb
link
in
the
schema
has
a
maybe
string
for
the
name
and
there's
a
bunch
of
other
optional
fields
and
each
one
of
these,
as
far
as
I
can
tell
when
it
gets
assigned
to
it,
allocates
I'm
thinking
to
myself
integers
and
strings
they're.
C
So
small.
Do
we
do
we
really
need
pointers?
I
don't
know
it
feels
like
we
could
just
stick
them
in
directly
in
the
parent,
and
I
see
eric.
B
C
Fair
enough
so
yeah.
If
I,
if
I
can
turn
that
on,
I
think
I'm
going
to
be
able
to
shave
off
the
next
like
an
extra
few
percent
but
yeah.
Otherwise,
I
wrote
a
benchmark
for
go
merkle,
dag
that
compares
the
old,
encode
and
decode
round
trip
with
the
new
one,
and
we
used
to
be
about
20
slower
overall,
and
that
was
basically
the
new
codec
and
now
we're
basically
within
the
noise
threshold.
So
it's
practically
as
good
as
it
was
before
and
that's
it
for
me.
D
In
particular,
when,
in
when,
when
put
into
I
go
ipfs,
the
integration
tests
using
dan's
optimized
version
of
dagp
are
now
faster
than
our
baseline.
A
Arguably
cool
does
anyone
else
have
any
updates.
E
I
have
a
very
short
update.
I
actually
spent
some
of
you
while
we
were
talking
with
a
few
messages.
I
met
briefly
with
with
volcanized
on
friday,
and
they
had
a
bunch
of
various
questions
about
things.
They
start
with
this
fan
misconception.
What
carpal
actually
is-
and
I
tagged
a
few
of
you
for
like
various
questions
on
what
do
we
do
for
the
bitcoin
idea?
E
Sorry,
the
falcon
adl
sort
of
for
the
actual,
whether
we
do
selectors
as
a
simple
like
list
of
blocks
things
like
that
yeah.
If
we
can
answer
those
like
within
the
week
to
the
nice.
A
All
right
anything.
A
B
I'm
so
excited
about
those
performance
results
that
dan
and
will
have
got
if
you
guys
can
also
snapshot
some
of
your
anything
actually
like
p
profiles,
the
whole
thing,
if
you
want
or
like
just
like
the
png
exports
of
stuff,
I
would
love
to
have
some
like
just
random
snapshot
artifacts
without
floating
around
that
I
can
look
at.
They
don't
have
to
be
like.
D
Be
really
helpful,
I
mean
getting
the
p
prof
out
of
dan's.
Smaller
go
bench
thing
is
pretty
doable
and
I
think
that's
what
dan
has
been
using
as
part
of
his
investigation
for
the
broader
one
that
hannah
posted
the
pings
of
in
the
channel
in
in
our
channel
that
we'll
be
talking
about
more
those
are
running
in
test
run
as
a
larger
integration
test.
I
don't
know
the
way
that
one
magically
invokes
getting
p
props
out
of
test
current.
I
hear
there
is
capacity
for
it,
but
that's
not
an
easier
ask.
B
Yeah,
even
some
like
just
take
a
picture
of
the
thing
at
low
res
and
like
we
can
find
a
place
for
it
to
live.
I
don't
need
particularly
high-res
data.
It
would
just
be
really
nice
to,
for
example,
this
other
issue
that
somebody
opened
asking
oh,
is
link
context,
a
huge
problem
because
it's
doing
struck
copies
like
yadda
yadda,
I'm
like
I
bet
it's
not,
but
I
have
no
data
and
neither
does
the
other
person
opening
the
issue
so
like
the
tiniest
bit
of
low-res
data
would
be
just
like
handy.
C
B
Probably
true
also,
but
synthetic
benchmarks
can
also
be.
A
Data
all
right,
I
I
have
one
thing
that
I
there's
a
public
holiday
next
monday,
so
I
won't
be
here
so
I
just
wanna
like
who
would
run
the
meeting.
I
know
that
rod
has
all
the
access
to
things
to
run
the
meeting
and
stream
it
and
whatnot,
but
will
you
be
there
next
week
or
should
I
give
it
to
likes
like
try
to
get
the
access
to
someone
else
as
well?
Just
for
yeah.
F
A
B
Can
talk
about
this
more
later,
but
in
general,
actually
I've
been
thinking
about
bringing
up.
Do
we
want
to
do
these
every
week,
or
would
this
meeting
be
useful
enough
if
it
was
alternate
weeks.
F
A
A
weekly
meeting-
and
I
so
I
will
be
like
end
of
the
quarters
this
week
so
next
week
I
will
be
full-time
crypto
computer
lab,
so
I
won't
do
any
hyper-related
stuff,
probably
like
some
side
things,
but
not
like
really
like,
like
what
I
really
do.
Is
I
do
half
time
really
like
the
js,
my
stuff,
so
I
won't
do
that.
So
I
guess
it
would
make
sense
to
do
them.
F
F
E
F
I
think
this
time
slot
becomes
more
awkward
for
me
as
well
next
week.
This
goes
to
7am,
but
I
have
to
be
up
for
a
weekly
meeting
at
6am
before
it.
So
I
I
don't
know
what
I'm
going
to
do
yet.
E
For
the
sake
of
the
recording,
let's
decide
this
async
is
there
anything
else
that
you
want
to
discuss
that
people
would
actually
want
to
hear.
A
Yes,
we
can
yeah,
but
do
we
then
basically
say
we
have
definitely
next
week
we
have
a
meeting
so
because
so
the
busy
people
watching
this
now
now
that
whether
they
will
be
meeting
or
not
or
should
we
just
skip
next
week
and
then
go
from
there.
I
think
what
like,
I
think
we
should
have
at
least
a
decision
on
like
let's
skip
next
week.
Voting
for
skip
next
week
with
skip
I'll,
be
away.
A
Okay,
so
we
skip
next
week
and
then
I
will
be
back
so
we
don't
have
to
deal
anything
with
any
things
and
then
we
can
like
discuss
it.
How
we
do
it
or
if
anyone
has
any
comments,
can
chat
with
us,
email,
us,
yeah
and
so
also
like,
please,
if
you
talk
to
any
other
people
that
might
have
tried
in
the
past
just
let
them
know,
perhaps
that
there's
something
going
on
that
we
might
not
do
it
every
week.
Cool,
okay-
and
I
also
put
it
in
the
meeting
notes,
I
guess
cool
all
right.