►
From YouTube: IPLD weekly sync 2019-12-02
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
B
B
It
basically
looks
at
like
all
of
the
values
and
then
figures
out
what
types
they
should
be
and
if
they're,
optional
or
not,
and
in
some
cases,
even
figures
out
like
if
they
need
to
be
a
union
there's
the
syntax
in
CSB,
is
for
there's
this
crazy,
sometimes
es
fees
for
doing
like
arrays
of
things,
and
so
that
actually
like
points
to
yeah
anyway,
like
it
figures
on
that
outs
here
and
that
uncovered
some
issues
that
we
have
in
some
of
our
parser.
B
Some
of
our
best
practices
are
actually
assumed
in
our
parsers
right
now,
so
we
need
to
back
some
of
that
stuff
out.
For
instance,
you
actually
can't
use
underscore
in
names
of
things,
and
you
have
to
capitalize
them
and
stuff
like
that,
and
when
you're
working
with
somebody
else's
data-
and
all
you
want
to
do
is
like
point
have
all
the
references
be
the
same.
So
you
point
back
to
the
original
data
without
bunching
it
up
too
much.
B
It
becomes
quite
beneficial
to
just
keep
the
original
names
with
underscores,
rather
than
trying
to
you
know:
camelcase
them
so
yeah,
that's
cool
using
the
notes,
much
stuff!
Writing
some
parsers
for
open,
datasets
and
I
feel
D
right
now.
So
that's
where
a
lot
of
that
stuff
was
coming
from,
and
the
other
thing
is
that
Ron
wrote
this
really
great
browser
testing
channel
called
Paul
Andina,
so
I've
been
slowly
porting
certain
tests
stuff
over
to
that.
B
C
We
started
getting
some
more
external
asks
about
Doc's,
lately
or
maybe
I
just
started,
noticing
more.
Who
knows
and
will
cry
I'm
sure,
you're
going
to
talk
about
this
chance
to
file
coin
stuff
was
related,
but
we
need
more
Doc's
I
started
to
do
some
more
Doc's,
I
haven't
gone
into
and
a
hand
combat
with
our
website,
which
I
think
we
should
do
at
some
point.
But
I
don't
currently
know
where
that
is
I,
don't
know
what
like
activation.
Energy
barriers
will
need
to
be
overcomes
to
deliver
the
first
step
of
value
there.
C
So
this
week,
I
just
filled
out
some
missing
gaps
and
the
go
Docs.
There
are
some
big
ones
like
the
node
interface,
which
is
kind
of
the
most
central
thing
in
the
universe
have
no
Doc's
on
it.
They
had
lots
of
Doc's
than
methods,
but
no
like.
Why
is
this
here
so
started?
Trying
to
fix
things
like
that,
so
there's
some
progress.
C
There
I've
also
been
trying
to
kick
a
little
more
work
in
on
the
code
jam
prototypes
and
trying
to
specifically
make
sure
that
I
can
write
what
needs
to
be
written
for
the
memory
amortizations
stuff
that
I've
been
in
searching
for
Alana
I'm,
pretty
sure
is
above
in
the
go.
Runtime
don't
want
to
talk
about
it.
C
So
far,
so
good
a
couple
of
things
that
turned
out
a
little
more
verbose
and
less
shared
code
than
I
expected
so
still
maybe
iterating
on
a
couple
of
those
things.
But
it's
really
coming
along
and
keeping
lots
and
lots
of
notes
and
reading
it
for
like
why
this
design?
Well
that's
going
to
be
from
archeology
someday.
A
Thank
you
mix.
My
list
is
myself:
I
didn't
do
that
much.
I
poly
related
stuff
I,
mostly
foreign
stuff,
and
there
was
a
meeting
with
oh
yeah,
which
was
related
to
5/9
ILD
and
there,
as
Eric
said,
come
up
some
documentation
issues.
We
should
get
better
yeah
but,
as
I
said,
nothing
that
much
any
related
stuff.
So
far,.
A
E
D
D
So
that's
a
good
addition.
I
think
the
other
two
things
were
some
little
'some
schema
stuff
for
Michael
for
CSVs
Eric,
I
pinged,
you
in
a
a
change
to
the
schema
past
JavaScript's.
He
came
in
Gaza
for
passing
names,
so
we're
allowing
I
wasn't
allowing
lowercase
first
characters
in
names,
but
now
we
are
because
Michael
wants
to
burn
down
the
world
and
the
change
I
added.
D
D
So
like
struct
fields
and
enum
values,
I,
don't
think
we
cover
in
spec
what
they
can,
what
the
character
limitations
are
there,
so
I
just
applied
the
same
rules,
which
is,
must
begin
with
a
character
and
then
can
contain
letters
or
numbers
and
underscores
so
even
that's
in
there
for
now,
and
we
can
tweak
later
if
somebody
comes
along
and
says,
I
need
underscores
is
a
first
character,
but
we'll
see
so
any
other
thing
was
the
thread.
Sorry
who
entreated
Michael.
D
D
Zip
car
file
they're
putting
live
data
sets
into
zip
files
because
the
JavaScript
thing
I
wrote
for
that
was
not
really.
The
original
use
case
was
not
for
large
files.
It
was
for
putting
blocks
in
it
for
mainly
test
fixtures
that
could
be
portable
across
runtimes
and
because
we're
thinking
about
using
them
for
larger
data
sets
I
did
some
work
on
investigating
in
JavaScript,
particularly
how
easy
it
would
be
to
make
it
more
efficient.
D
There
are
some
trade-offs
in
doing
so.
I
I
did
some
of
that
work
and
it
seems
to
be
okay.
I
could
make
a
streaming
interface
without
too
much
pain.
I've
got
I've
got
used
to
completely
separate
libraries
for
read
and
write.
Apparently
we
can't
have
all
the
good
things
in
one
place,
but
I
think
there's
some
trade-offs
in
browser
compatibility,
but
browser
testing.
Again
it's
one
of
those
things
I
wasn't
doing
beforehand.
D
So
I
need
to
add
that
as
a
first-class
thing,
because
I
would
like
to
have
test
fixtures
that
I
can
include
in
the
browser
as
well
when
I'm
testing
stuff.
So
you
know,
test
zip
file.
Full
of
blocks
would
be
nice
to
be
able
to
put
that
in
a
browser
environment
and
extract
the
blocks
and
do
the
tests
on
stuff
in
there,
so
that
was
one
of
the
intentions
of
those
those
files.
So
anyway,
that's
that's
pretty
much
the
main
things
that
I'm
working
on.
A
Thank
you.
Does
anyone
else
from
the
other
folks
have
any
updates
or
want
to
say
something?
A
F
A
Okay,
so
I
have
one
more
item
which
came
up
today.
Actually
it's
more
of
a
question
for
what,
probably
because
we
currently
have
hemp
spec
for
IP
LD
and
there
used
to
be
an
alternative
hemp
spec
that
is
used
in
psycho
and
I.
Wonder
if
the
IP
LD
one
is
a
superset
of
the
far
econ
one
or
if
there's
any
differences
or
like.
If
really
god
are
you
aware
of
any
income
in
incompatibilities
or
should
it
be
comparable.
D
They're
really
close.
Basically,
there
was
just
some
disagreements
about
first
was
naming
so
far,
calling
being
a
being
obsessive
about
names
being
like
one
character
and
mine
was
was
a
little
bit
more
verbose
like
my
daughter
away
array
was
called
data,
whereas
I
think
in
that
one
it's
called
D
or
something
like
that
and
the
the
other
thing
was.
D
They
didn't
want
to
include
any
hampt
parameters
in
the
route
block
because
they
know
the
context
and
they
know
what
the
parameter
is
going
to
be
and
I
disagreed
that
you
need
to
include
that
and
we
never.
We
never
had
a
way
of
reconciling
that,
although
there
was
some
proposals
for
how
to
reconcile
that,
but
they
are
they're
really
close
and
if
somebody
wanted
to,
if
it
was
important
to
reconcile
them,
we
probably
could
but
yeah.
D
But
but
you
could
read
the
ham
speck,
that
I
wrote
and
then
look
at
the
one
that
file
coins
using
and
it
would
be
pre
pretty
much
the
same,
I
think
they're,
even
using
the
bits
bit
sequencing
thing
that
I
wrote
for
pulling
out
indexes
from
sequences
of
bits.
So
it's
it's.
It's
very
similar.
Pretty
much
the
same
algorithmically
just
some
differences
in
the
way
that
they
are
encoding
because
far
coin
likes
to
be
yeah,
obsessive
about
certain
things
that
don't
think
it
quite
that's
important.
B
D
Well,
if
they're
not
aliases,
those
of
us
that
was
the
MIS
name
or
a
misnomer
about
that
property,
it
wasn't
an
alias,
it
was
a
rename,
so
we
would
have
to
adopt
a
name,
so
we
would
have
to
adopt
their
name
their
their
names
for
properties.
I
can't
I'm,
not
sure
exactly
what
they
weren't
I
think
I.
Think
in
like
in
the
key
value
array
they
had
like
uppercase
and
lowercase,
or
something
really
strange
too.
So
there
was
some
weird
discrepancies
that
I
just
couldn't
get
behind,
but.
B
The
best
thing
about
the
rename
feature
ends.
The
schema,
though,
is
that
like
as
far
as
you're
concerned
when
you're,
using
like
the
types
that
get
generated
from
the
schema,
you
can
still
use
data
as
a
keyword.
It
just
isn't
getting
encoded
of
D
so
yeah,
it
would
be.
You
know,
we'd
get
all
the
benefits
anyway
and
I
mean
I'm
fine
I'm
having
their
default
either
be
folks,
I
mean.
C
B
D
D
C
D
B
D
Then
and
then
the
leaves
using
buckets
yeah
and
if
you're
using
buckets
of
the
leaves,
then
you
multiply
that
by
the
size
of
the
buckets,
and
so
you
end
up
with
these
really
small
intermediates
and
then
these
large
end
ones.
The
file
coin
case,
though,
is
like
they're
all
putting
it
on
chain,
so
it
probably
doesn't
matter
that
much
the
size
of
the
blocks,
I
think
so.
D
You're
not
you're,
not
just
pulling
an
individual
block
and
having
to
ship
that
around
the
network,
you're
doing
I
mean
in
a
chain,
so
I
think
the
the
the
the
requirements
for
size
for
file
coin
and
for
any
blockchain
based
application.
I
think
it
different
to
the
way
we
think
about
them
when
we're
talking
about
distributed
applications
where
we're
you
know
sharing
around
these,
like,
if,
like
the
NPM
or
package
manager
case
where
we
want
to
have
a
HAMP
to
address
all
the
packages,
you
don't
to
have
these
tiny
blocks.
D
Just
to
you
know,
you
don't
want
to
navigate
down
eight
different
blocks,
just
swap
back
and
forth.
Just
to
find
a
package
it'd
be
great
to
do
three,
you
know.
So
you
have
a
higher
branching
factor,
it
looks
slightly
larger
blocks
and
so
you,
the
back
and
forth,
is
more
efficient,
but
when
you're
doing
octane
you've
got
it
all,
then
it's
a
bit
a
little
bit
more
different
use
case.
Also.
B
D
The
theory
and
that's
what
Falcone's
going
with
because
they
mutating
the
what
does
it
be
actors
with
map
and
it's
going
to
be
mutating
a
lot.
The
idea
is
because
they're
mutating
a
lot
they
when
you
have
a
higher
branching
factor.
You
are
it's
theoretically,
mutating
less
data,
total
and
total,
but
more
blocks.
So
instead,
if
you
have
a
branching
factor
of
25
to
50
56-
and
you
want
to
change
one
thing
at
a
root,
then
you
have
to
change
larger
blocks.
So
you
copy
an
entire
block
and
just
change
one
element
of
it.
D
D
A
lot
of
these,
and
even
what
even
the
way
I
talk
about
it,
it's
based
on
guesswork
and
we
don't
have
data
for
what
the
what
the
trade-offs
are
here.
I
started
work
on
that
I
started.
Doing
some
research
work
on
large
data
set.
What
what?
What
is?
What
are
the
trade-offs
when
you
have
larger
branching
factor
for
mutation
and
versus
smaller
and
then
and
then
the
trade-offs
for
okay
well
block
size,
because
in
some
cases,
block
size
matters,
because
you
want
to
do
network
transfer,
and
maybe
you
don't
have
the
latency
problem.
D
So
you
can
do
lots
of
small
transfers
or
maybe
you've
got
high
latency.
So
bigger
blocks
is
better
and
then
how
many
blocked
reversals,
because
you
bigger
blood,
branching
factor
the
fewer
traversals
to
get
to
what
you
need,
and
so
it
would
be
nice
to
have
data
on
all
of
those
things
so
that
we
could
say
hey
if
you're
gonna
use.
This
thing
then
consider
what
your
application
is
and
then
and
what
your
environment
is
and
then
choose
your
parameters
based
on
these
things
that
we
know
and
have
measured
about
even.
B
F
We're
gonna
yeah.
Actually
this
the
problem
statements,
Arabs
put:
it
is
more
or
less
my
work
plan
for
the
next
Siemens
I,
quite
literally.
Well,
it's
not
the
entire
three
months,
but
pretty
much
most
of
January
supposed
to
be
taken
by
figuring
out
what
is
acceptable
for
from
performance
standpoint
and
I'm
supposed
can
like
see
what
is
available.
F
There
is
in
those
test
beds
and
if
Nazca
is
available
to
my
own,
but
figure
out
like
what
is
the
cost
of
a
lot
of
smaller
versus
larger
chains,
because
kind
of
a
number
of
parameters
in
the
study
I
have
been
working
on
on
the
side.
You
know
before
a
joint.
Each
parameter
is
extraordinarily
well
over
a
very
large
amount
of
outcomes
right
and
actually
for
for
next
meeting.
Right.
F
D
D
F
I
actually
don't
know
how
it
works
with
with
work
plants,
whether
they're
kind
of
public
or
not,
but
if
they
are
and
I
I,
don't
know
how
much
of
it
I
can
take
in
place
to
you
guys,
yeah,
but
but
it
basically,
it
basically
comes
from
the
standpoint
of
like
talk
about
streams.
Specifically,
like
you
have
a
stream
of
data.
What
can
you
do
with
it,
given
what
we
know
and
given
what
kind
of
streams
of
that
we're
likely
to
encounter?
So
it
doesn't
really
specifically
talk
about.
B
B
The
key
value
store
also
has
to
store
the
keys
and
the
intermediary
blocks
have
the
keys
Valley
at
some
point
here,
like
the
whole
equation:
kind
of
inverts
and
you're,
not
saving
like
any
suspected
and
and
in
the
Hampton
in
the
hands
like
you,
you
you're,
pretty
guaranteed
you're,
just
never
gonna
happen
to
do
application,
and
this
week
immediate
notes.
Anyway.
A
I'd
like
to
come
because
the
meeting
is
almost
over
I
like
to
come
back
to
the
original
question
from
out
and
the
hem
specs
between
if
I
own
and
I,
Polly
and
I
think
it
would
make
sense
to
kind
of
like
being
as
close
as
possible.
So
would
it
make
sense
that,
for
example,
if
they
use
those,
we
could
use
those
renames
that
they
use
or
similar
ones
and
they
change,
but
basically
having
those
three
names
should
be
simple
and
in
regards
to
having
the
parameters
in
the
root
blog
I
think
it
totally
makes
sense.
D
Michael
had
an
idea
that
I
think
was
worth
exploring,
which
was
to
push
the
parameters
up
a
container,
and
so
the
the
root
block
was
a
container
underneath
where
you
store
the
parameters
and
whether
that's
a
separate
block
or
within
the
same
block.
But
it's
a
thing
that
you
path
to
so
you've
got
the
parameters
and
then
you
pass
through
the
root
block
and
then
the
RHIB
block,
just
is
it
just
a
and
like
every
other
intermediate
block
is
just
a
link,
an
array
of
of
links.
D
That's
one
way
to
solve
it
and
then,
when
fact,
when
file
coin
uses
that
they
just
start
there
and
when
we
want
to
use
it
any
bid
parameters,
we
start
one
level
up.
I
might
just
you
know
what
they
could
have
been
more
reconciliation
work
I
just
found
it
really
hard
to
engage
with
the
file
coin.
Folks,
because
it's
so
busy
and
they
they
move
on
so
quickly
and
and
they've
got
their
own
needs.
And
it's
like
it's
like
a
steamroller
that
just
goes,
and
so
we
have
the
same
thing
with
schemas
and
their
specs.
D
A
All
right
is
there
anything
else
we
should
discuss
I.