►
From YouTube: 🖧 IPLD weekly Sync 🙌🏽 2020-03-02
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
Welcome
everyone:
through
this
week's
IPL
desync
meeting,
it's
March,
the
2nd
2020
and
as
every
week
we
go
over
the
stuff
that
we
have
done
last
week
and
that
we
plan
to
do
and
then
discuss
any
other
items
we
might
have
and
also
what
we
normally
do.
Is
we
start
with
introducing
ourselves
if
there
are
new
people
on
the
call,
if
they
want
to
introduce
themselves,
they
don't
have
to,
and
so
yeah
hi
Chris
I.
B
Guess
so,
I've
been
a
software
developer
and
I've
been
in
medical
imaging
for
about
20
years
so
long
time
and
before
I
found
IPL
D&I
PFS
was
kind
of
thinking
about
how
it
could
change
the
way.
Medical
images
for
I
guess,
stored
kind
of
formats
are
pretty
naive
and
have
those
less
issues
with
it.
I
think
that
kind
of
been
vented
that
oh,
should
be
a
graph
restructure.
B
My
general
feeling,
after
that,
digging
in
those
last
couple
weeks
is
it's
bit
tribal
knowledge
I,
don't
you
know
in
some
kind
of
extracting
things
share
there,
but
it
feels
a
little
bit
slow.
So
anyway,
that's
kind
of
neat
and
hoping
I
can
find
ways
to
help
in
and
in
good
size,
I'm,
actually
trying
to
learn,
I,
building,
trying
hope
out
how
I
can
and
also
figure
out
how
to
get
the
medical
imaging
work.
You
fight
building
ipfs
too,
so
that
it
kind
of
several
concurrent
threads
going.
A
Good
things
yeah
that
sounds
pretty
exciting.
So
could
we
have
your
the
call
yeah?
So
let's
go
over
our
weekly
updates
and
in
the
past
week
I
haven't
been
a
structure
that's
last
week.
So
therefore,
it's
not
very
exciting,
but
still
and
so
I
work
on
the
rust
side
of
things
and
yeah
things
are
shaping
up
so
I
now
so
mottley
hash
is
in
a
good
shape.
Multi
base
is
slowly
getting
there
and
now
I
move
on
to
rust,
CID
and
also
the
other
people.
A
So
it's
kind
of
like
a
group
of
three
or
four
people
are
moving
together
with
me
like
from
grade
to
grade,
and
we
make
all
of
our
changes
at
the
same
time
on
the
same
crates
for
whatever
reason,
but
that's
pretty
cool,
so
yeah
there
is.
There
are
many
pull
requests
so
expect
a
better
version
of
rust
CID
soon,
and
also
this
also
kind
of
news
that
I
am
now
officially
the
person
of
contact
protocol
EPS
for
the
rust
ipfs
people.
A
So
Molly
basically
made
me
the
person
of
contact,
so
she
was
previously
and
then
I
took
it
over
now.
That's
pretty
cool,
and
so
also
like
for
the
ipod
et
might
be
interesting.
So
my
involvement,
so
I
plan
to
be
really
like
people
involved
in
the
ILD
side
of
things
and
for
the
IPS
side
of
things
I'm,
mostly
just
watching
so
I.
A
Don't
actually
also
don't
do
any
any
reviews,
except
they
asked
me
for
review,
but
basically
they
should
mention
those
things
themselves,
but
it's
more
a
matter
of
like
if
they
need
something
or
if
they
think
Oh.
Someone
from
protocol
Labs
should
look
over
it
or
someone
from
the
other
IPS
teams
to
look
over
it.
They
can
just
contact
me
because
I
get
sometimes,
as
Chris
probably
also
found
out
difficult
to
find
out.
A
You
now
need
to
ask
every
contributor
and
to
agree
on
the
license,
change
and
it's
interesting
that,
for
example,
at
Sadie
had
only
40
commits,
but
11
people
contributing
and
so
hmmm
yeah,
but
it's
cool
I'm
getting
there
and
yeah,
and
the
last
thing
I
did
is
kind
of
related
to
also
Phi
quite
a
bit.
Is
that
the
previous
course
you
also
might
have
heard
about
some
calculations
called
comb?
Comb
P
for
there
yeah?
A
It's
just
like
a
yeah
regulation
of
tennis
is
for
five
coin
and
we
want
to
run
them
at
larger
scale
on
Emerson
and
I
felt
like
I
really
need
to
look
into
the
miracle
tree
stuff
and
I
did
and
made
kind
of
screaming
and
now,
instead
of
using
two
times
the
memory
of
the
input
size
in
our
text
like
five
megabytes,
so
ten
megabytes
of
RAM.
So
this
pretty
cool.
It's
also,
basically,
this
the
scalp
type
of
worker
that
I've
worked
on
my
previous
job.
A
C
C
I've
been
working
on
some
coroner
phase,
three
factors
for
a
while,
they're,
all,
basically
in
the
name
of
performance
and
unblocking
some
stuff
that
I
wanted
to
be
able
to
do
in
the
future
around
cogeneration
and
going
real
fast
and
today,
I
finally
have
a
commit
in
which
everything
is
granting,
and
this
is
an
enormous,
terrifying
diff.
It's
many
thousands
of
lines.
It
is
overall,
basically,
the
entire
size
brats
having
a
laugh
yeah,
all
right,
it's
the
entire
size
of
the
project.
C
Again,
basically,
almost
this
is
including
lots
and
lots
of
lessons
learned
about
how
to
use
internal
pointers
in
order
to
make
the
whole
system
be
way
more
GC
friendly
and
generate
less
garbage.
In
memory
of
the
first
place,
all
the
traversal
subsystem
is
refactored
to
use
these
new
interfaces.
Selectors
are
all
we
factor.
C
The
codecs
are
already
factored
their
shared
benchmarks
and
tests
for
all
these
things
are
using
the
new
set
of
assembler
interfaces,
there's
more
of
those
tests,
and
this
was
also
a
great
chance
to
make
them
more
consistent
and
sort
of
stamp
them
out
for
all
the
different
packages
in
the
same
way.
So
that
all
happened,
a
bunch
of
error
type
consistency
got
fixed
in
the
middle
of
this,
because
I
basically
rewrote
all
of
the
notifications.
Cool
literally
every
single
thing
is
faster.
Some
things
are
twice
as
fast.
Some
things
are
four
times
as
fast.
C
Basically,
everything
is
awesome
and
I'm
so
excited
about
that
canoe
and
after
this
I
have
no
idea
what
I'm
doing
in
the
next
week,
because
this
has
been
the
focus
for
such
a
long
time,
I
kind
of
have
to
come
up
for
air,
so
I
think
after
this
I'm
going
to
go
through
the
specs,
repo,
mapper
and
stuff
that
needs
to
finishing
Peters
been
poking
me
about
paths
and
some
specs
around
there.
Maybe
it's
time
to
look
at
that
with
full
attention.
C
D
Most
of
my
time
spent
on
this
internal
golang
for
distributed
systems
course,
so
I've
been
formalizing,
my
go,
laying
knowledge,
which
has
actually
been
more
helpful
than
I
expected.
Actually
now,
I
died
and
I.
What
I
expected,
but
just
more
helpful
in
just
formalizing.
My
thinking
about
certain
things,
particularly
their
point
of
semantics,
was
something
I'd,
never
really
fully
dropped.
What
go
was
trying
to
do
with,
with
its
with
its
version
of
pointers
so
and
it's
good
to
get
feedback
from
a
bunch
of
go
programmers
that
are
focused
on
teaching
people
go.
D
So
you
know
that's
still
ongoing
this
week
that
it
turns
out
that
the
course
work
was
way
too
dense
for
I
I
guess
for
the
amount
of
time
allocated
or
for
the
speed
I'm,
but
there's
a
final
component
to
go.
That's
much
more
focused
on
ipfs
and
live
peer-to-peer.
So
that
should
be
an
interesting
finish
up
aside
from
that
I.
Don't
have
really
anything
interesting
to
report.
E
A
So
I
think
we
clearly
need
some
like
yeah,
some
better
yeah
documentation
about
like
at
least
like
where
we're
going
like.
Even
if
it's
not
like
yeah,
that's
the
new
thing
and
just
use
it,
but
just
like
yeah
because,
like
for
example,
Technic
someone
was
that
you
were
talking
about
this,
the
blog
service
stuff,
which
is
also
one
thing
that
I
wanted
to
get
rid
of
since
I,
so
I
joined,
perhaps
three
years
ago
and
I
looked
at
the
code
because
I
was
like.
Why
is
this
blog
service
in
IPL
de
and.
C
D
Word
and
so
so
Chris
from
for
me,
you
know
I've
been
working
on
this
for
less
than
a
year
now,
but
for
me
the
most
interesting
cop
piece
of
context
that
we're
profiling.
Some
of
your
questions
is
the
history
of
IP
LD
and
how
it
was
extracted
from
IPs
vests
as
a
discrete
component
at
one
point
after
it
was
after
was
realized,
hey
there's
this
thing
in
the
middle
of
ipfs.
That
is
its
own
thing.
D
For
you
know,
little
experimentation
that
we're
doing
we're
doing
little
projects
here
and
there
completely
bit
skeptic
up
'old
from
ipfs,
so
using
it
as
a
way
to
deal
with
distributed
data
without
thinking
too
much
about
that.
How
does
it
get
distributed
bit,
so
you
know
putting
them
into
into
storage
services
that
are
not
distributed
using
in
just
as
ways
to
create
blocks
to
to
store
in
car
files,
those
archive
files,
so
the
JavaScript
side.
D
C
From
a
little
different
angle,
if
you
were
to
look
at
the
go
repos,
and
especially
the
one
that
I
linked
to
the
my
updates
for
the
day,
will
be
a
very
different
take
precisely
because
what
Ron
said
is
true,
and
in
this
repo
I
made
a
very
clean
break
and
essentially
started
over
again
a
little
over
a
year
ago.
So
this
is
God
not
necessarily
more.
The
lessons
learned,
but
the
ones
that
are
are
more
consistently
applied.
Since
is
you
know.
C
B
What
that's
helpful-
and
you
know
I
there's
a
couple
things
I
guess
I
forgot
to
mention-
is
that
there
didn't
mainly
writing
JavaScript
last
I,
guess
six
seven
years,
but
yeah
I've
done
C++
and
C,
sharp
and
jog,
and
all
of
my
past
and
I
know
that
we
have
a
specific
of
an
interest
in
learning.
Rus
they've
read
the
book
probably
several
times,
and
it's
a
lot
to
grok
and
take
in
and
actually
had
a
project
to
and
enough
time
to
focus
on
getting
over
that
curve.
B
But
that
is
an
interest
of
mine,
but
right
now
I
can
you
know
I
can
before
word
JavaScript
quickly
and
in
actually
thing
a
lot
of
code
that,
with
my
project,
I'm
working
on
probably
be
in
that
so
could
certainly
help
out
anything
I,
don't
mind
documenting
going
over
the
specs
I
have
no
problem.
This
part,
that's
one
of
help
and
I
think
one
of
the
things
that
might
be
helpful,
I
guess
requests
is
just
rain,
don't
know
I
mean
if
people
are
willing
to
take
half
hour
out
and
just
stay
with
me
on.
B
The
video
conference
calls
and
want
me
to
pick
your
brain
and
it's
just
I
haven't
I
could
use
that
ramp
up
pretty
quickly.
I
know
the
help
I'm
sure
this
kind
of
like
requests
and
destiny
and
I'll
give
you
a
lot
back
and
pretty
sure
I
mean
also
just
if
a
reference
to
you
so
I
actually
started
a
open
source
project
myself
called
Cornerstone,
Jas
and
so
I'm
actually
created
from
scratch
about
five
or
six
year,
and
it
now
has
one
1,200
github
stars
and
it's
the
most
popular
open-source
project
in
medical
imaging.
B
So
you
know:
I,
don't
Lots
like
grow
that
community
and
build
it
up,
and
so
I'm
very
familiar
and
comfortable
working
with
open
source
projects
and
whatnot,
and
what
I
wanted
to
do,
even
with
my
new
project.
I'm
working
right
now
is
make
it
completely
open
source
for
medical,
imaging
and
healthcare.
So
that's!
That's
the
you
know
the
goal
or
the
direction
so
pretty
comfortable
in
this
environment
and
know
how
important
it
is
to
be
available
to
the
community
and
answer
questions
and
be
patients
and
staff.
So
you'd,
be
that
with
me.
A
Cool,
like
that's,
really
exciting,
so
yeah
so
but
I
think
you
can
certainly
get
those
like
yeah
I,
don't
like
ping
us
all
right,
yeah
so,
and
so
just
also
like
it's
a
background,
so
I'm,
probably
yeah
so
I'm
so
far
from
this
team,
I'm
the
longest
on
JavaScript
IP
LD,
but
like
in
the
last
so
being
this
year.
Basically
I
kind
of
like
moved
on
to
doing
the
rust
side
of
things,
but
before
this
I
did
for
two
years,
I
did
the
JavaScript,
I
purely
stuff
and
I
think.
D
A
D
D
It's
open
source
is
open
sourced,
so
I'll
just
put
this
in
the
chat
Chris
that
that's
really
fresh
and
that's
using
the
newest
IP
LD
things,
and
this
is
pushing
towards
ideas
about
how
do
you,
I
PLD
as
an
independent
thing
so
and
that
actually
might
be
really
relevant
to
what
you're
working
on
and
it's
not
just
the
patterns,
but
maybe
even
as
a
consumer
of
it,
so
have
a
poke
at
that.
But
yeah.
B
Okay,
yeah
I've,
actually
I
appreciate
that
I'm
in
the
u.s.
to
Minnesota,
so
I've
had
a
couple
of
dialogues
with
Michael
earlier
this
week.
I
guess
the
big
thing
is
just
want
to
ask
permission,
because
maybe
sometimes
he
board
heads
down
crazy
busy
like
Volkers,
been
trying
to
get
that
the
actually
no
actually
important
trend
get
the
go.
I
feel
d-prime
a
much
last
couple
of
meetings
that
even
working
hard
on
that.
So
don't
want
to
disturb
you
so
there.
This
is
like
not
a
good
week
to
bug
you.
Let
me
know.
D
A
May
also
think
that
Michael
is
good
for
the
like
for
the
Pacific
javascript-based
vision
of
how
I
probably
should
work
and
I'm
honestly,
more
or
less
just
currently,
they
kind
of
like
the
maintainer
of
Betty,
making
sure
that
things
still
work
with
ipfs,
so
I'm
not
like
yeah
so
just
two
year,
because
the
perspective
or
like
who's
working
on
which
things
so
I'm
they're
doing
the
releases
and
fixing
small
bugs,
but
not
really
like
you're
doing
the
next-generation
stuff.
So
I
I,
yeah
kind
of
Hannibal's
overthrew
rod
and
Michael.
B
B
A
Anything
it's
also
like
it
I
think
it's
true
for
all
of
us,
like
it's
kind
of
like
we're.
Also
using
busy
doing
all
those
thing
open
source
were
so
busy.
Just
ya
know
just
ping
us
and
go
easily.
If
we
don't
have
time,
we
will
just
tell
you
and
yeah,
so
it's
just
like
yeah
feel
free
to
ping
and
okay
I.
E
So
I'm
kind
of
on
this
call
more
or
less
in
the
same
capacity
as
you
are
I,
don't
work
and
I
peel
the
direct
here
I
just
kinda,
listening
to
things
together,
for
you
know
for
ipfs
longer-term
projects
when
you
approach
your
current
problem
that
you
want
to
try
to
use
IPO
do
ways.
I
very
much
would
like
to
encourage.
E
You
too,
like
don't
feel
constrained
by
what
you
read
about
IP
ID
right
now,
so
try
to
have
a
better
idea
of
what
your
best-case
scenario
is
for
your
you
know,
data
model
and
so
on
so
forth,
and
then
see
if
I
feel
the
mattress
and
it
doesn't
match.
Ask
this
group,
because
it's
very
likely
that
it
can
be
can
be
sufficiently
been
to
accommodate
whatever
you
need,
as
opposed
going
type.
Ok.
E
Well,
there
is
this
idea:
they
think
how
can
I
shoehorn,
what
iswhat
I
have
as
ideas
into
this
supposed
aesthetic
thing,
because
everything
is
like
super
evolving
and
also
there
is
a
lot
of
stuff
to
read
so
hearing
kept
very
like
solid
as
from
what
is
it
that
you
want
to
express?
If
you
have
all
the
you
know,
all
the
perfect
libraries
in
the
world
definitely
helps.
When
approaching
you
know
what
is
already
available,
what
can
be
built
up.
B
Yeah
yeah
well
good.
The
good
news
there
is
I,
think
I've,
read
I.
Think
I
was
good
enough.
Understanding
about
he'll,
be
at
least
how
it
works
today
and
I
read,
read
tons
of
the
source
code
and
every
like
information,
page
I
could
find,
and
so
they're
probably
I've
read
at
least
80
to
90%
of
content.
That's
out
there
on
this
and
I
think
it
actually
what
I'm
trying
to
do
what
I
feel
he
has
today
I
think
actually
is
enough.
So
I'm
curious.
B
You
know,
there's
new
things
to
understand
how
the
newer
things
that
yes,
I
think
it
about
the
future,
how
they
can
kind
of
intersect
if
I
don't
feel
like
I'm
blocked.
Oh
I
need
this
feature,
but
it's
the
struggle.
I
pad
is
just
to
make
sure
I
understand
kind
of
the
right
way
of
doing
things.
You
know
I
guess
from
you
guys
is
point
of
view
and
pick
what
else
I'd
say
is
I
just
want
to
get
it.
B
You
know
and
I
think
I
think
I
would
pregame
you're
saying
getting
it,
but
you
think
I
should
mention.
Is
that
I'm
a
pretty
strong
focus
on
performance?
That's
one
of
the
things
that,
in
my
careers,
I
just
make
very
fast
applications
and
medical
imaging.
That's
actually
harder
than
me.
You
may
think,
because
if
it
just
a
large
standard,
weird
and
so
a
lot
of
times,
I'll
look
at
IP
LD
from
a
performance
point
of
view.
You
know
one
person
is
I'm
like
how
fast
is
seaboard
sterilized.
B
That's
a
good
question
and
decent
deserialize
actually
and
also
looking
I
P
FS
I've
been
working
with
colleague
and
we're
like
hey
I,
moved
it
he's
in
California
I'm.
Here
this
is
like
how
fast
we
move.
200
megabytes
of
data
from
my
notes,
you're
known
over
1
gigabit
connections,
and
so
like
we're
just
playing
like
that,
trying
to
understand
the
base
performance
profiles
and
when
I,
look
at
design
architecture.
I
always
have
that
in
mind.
So
anyway,
it
looks
good
as
this
day
just
would
like
to
get
it
document.
B
I
mean
I
actually
didn't
think
about.
Like
my
project,
being
open
source
might
be
a
good
way.
I,
don't
know,
I
guess
gets
into
medical
imaging,
which
is
lucky
the
warm
no,
but
show
an
example.
How
you
take
I?
Guess
a
data
model
way
of
doing
things
and
map
me
and
I
PLD
would
be
probably
beneficial
a
lot
of
people
there's
something
out
there
right
now.
It's
like
there's
a
lot
of
small
examples
and
kind
of
like
conceptual
things,
but
not
like.
B
C
C
One
example
of
modeling
data
that
I
think
went
well,
that's
kind
of
a
recent
one
is
I,
don't
know
if
you've
taken
a
peek
at
the
selector
system,
but
that
was
something
that
we
designed
with
in
IPL
B
as
a
tool
for
IPL
D.
So
you
know
it's
a
bit
met
it
in
that
way.
But
if
you
disregard
that
believe
some
of
the
IP
of
these
schema
systems
as
a
design,
language
and
the
entire
thing
is
gone
engraved.
It
yeah,
that's
good.
C
With
selectors,
the
description
of
what
you
are
going
to
do
with
the
selector
has
some
human,
readable
syntax,
and
it
also
has
a
way
of
being
understood
canonically
as
the
IPL
de
deck,
the
thing
that
we
would
maybe
call
an
AST
when,
though
it's
not
very,
we
haven't,
figured
out
what
to
call
this
thing.
Where
you
have
a
data
model.
C
C
Selectors,
which
you'll
find
some
information
about
those
in
the
specs
repo
I
think
there's
just
a
directory
at
the
top
called
selectors.
No,
this
is
one
of
the
things
that's
unfortunately
only
implemented
in
go
as
of
yet
I.
Don't
think,
there's
a
giant
blue
tation
so
to
bruise
your
eyes
on
my
code
over
there
and
a
lot
of
that
was
written
by
somebody
else
or
a
team
called
Hana.
B
E
E
There
are
couple
things
that
stood
out
where
he
basically
mentions
that
maps
will
map
keys
are
essentially
strings,
which
means
the
Derry
valid
Unicode.
As
far
as
I
understand,
this
is
no
longer
the
case
and
hasn't
been
for
a
while.
So
like
this
is
not
something
probably
needs
to
be.
Can
I
talked
about
on
the
pier
itself,
because
it's
not
really
that
important.
This
iteration,
but
in
general
map
keys
tend
to
be
just
any
bytes,
so
it
can
be
malformed
unicode
and
it
can
be
in
invalid
pass.
D
They
saying
we
need
to
make
sure
that
selector
supports
bytes,
so
that
we
can
reference
the
full
range
of
keys,
regarded
of
how,
regardless
of
how
they're
represented
in
strings,
because
the
data
model
I
build
a
data
model
is
still
still
says
strings
as
the
only
form
of
keys
for
maps.
Why?
This
thing
is
that
so
dated.
C
C
E
Mean
from
my
perspective,
it
doesn't
really
change
anything
right
this
moment,
but
yeah.
We
need
to
have
like
a
final
verdict
on
all
this
and
cause
other
stuff
like
like
a
lot
of
the
I
know.
The
web
Gateway,
for
example,
will
totally
need
to
have
a
definition
of
how
to
get
from
a
URL
pass,
which
doesn't
support
like
it's
much
half
of
the
corner
cases
to
something
that
could
potentially
be
the
entire
gamut
of
any
white
possible
accommodation,
yeah.
D
I'd
see
this
when
we
get
it's
gonna,
be
interesting.
If
you,
if
you
do
some
spec
work
on
this
because
in
JavaScript
it
gets
very
tricky,
but
with
the
way
we're
doing
it
anyway,
I
mean
with
if,
if
we
evolve
till
it
to
a
place,
that's
more
similar
to
IPL
d-prime,
where
you
navigate
through
through
nodes
in
a
via
a.
D
An
external
API
that
you
know
jumps
through
your
data
rather
than
giving
you
porcelain
and
that's
fine,
but
if
we
go
the
way
we're
doing
it
now,
which
is
just
give
you
the
the
porcelain
as
it
is,
it
just
doesn't
work
for
the
JavaScript.
If,
if
we
jump
outside
of
string
keys
like
just
say,
strings
I
know
that
that
that
it's
the
Unicode
that
makes
that
complicated.
But
that's
where
that's
where
we're
gonna
need
a
lot
of
spec
work
and
alone
ago.
She
ation
I,
think
true.
C
For
what
it's
worth,
some
of
the
latest
stocks
in
the
I
appealed,
the
primary
photo
do
also
still
have
a
bunch
of
warnings
on
saying
this
standard
domain
here.
This
is
not
gonna
round-trip
right
now,
and
that
specification
needs
a
little
more
work
in
the
core
specs
we
go
and
we
will
return
to
this
SoftLayer
yeah.
D
C
I
think
it's
historically
been
really
difficult
to
document
this,
because
there's
a
bunch
of
different
applications.
The
IP
OD
is
being
extracted
from
the
ipfs
project.
Things
really
clearly
here.
That's
piece
of
history
yeah,
because
ipfs
mostly
is
an
application
that
regards
files
and
filesystem
paths
and
making
sure
those
work
in
URLs
in
an
unsurprising
way,
making
sure
that
they
can
be
mounted
as
a
POSIX
file
system,
and
all
that
implies
is
restrictions
and
challenges
and
I
feel
D,
isn't
I.
Peel
D
is
a
way
of
working
with
maps
and
lists
and
data
in
general.
C
It
does
not
necessarily
have
applications
to
a
file
system
and
certainly
not
necessarily
a
POSIX
one.
All
the
rules
of
that
implies
that
just
happens
to
be
one
of
the
biggest
historical
users.
So
when
we
deal
with
this
part
of
the
specification
about
keys
and
strings
and
paths,
we
have
to
make
sure
that
we
know
to
acknowledge
both
of
those
things
and
say
when
you're,
using
this
kind
of
application.
Here's
some
semantics
that
you
can
probably
expect
when
you're
using
this
kind
of
application
or
I
appeal
to
you
directly.
C
D
C
D
To
be
honest,
I
think
it's
much
more
to
do
with
the
way
that
IPL
is
implemented
in
JavaScript
than
it
is
about
JavaScript
strings,
because
in
JavaScript
in
the
JavaScript
API
is
now
you
interact
with
IPL
D
as
instantiated
objects.
We
give
you
an
object
that
is
instantiated
from
the
codec
and
and
we
just
can't
do
non
string
keys
because
they
need
to
go
on
objects,
and
so
it's
the
way
that
JavaScript
objects
have
field
names.
D
C
E
B
Well,
what's
the
you
know,
what
can
you
do
with
the
IDL
and
they
you
had
always
kind
of
weird
bindings
and
to
work
around
concepts
in
the
ideal
that
want,
in
the
other
languages
that
you're
working
with
and
I
think
it
young
obvious,
is
doable,
and
so
we
find
that
having
integers
as
keys
in
a
map
is
absolutely
critical.
You
can
create,
you
know,
Java
scripts
kind
of
wrappers,
around
objects
or
whatever
it
is
provided,
provide
a
library
like
navigate
that,
but
one
of
the
things
I
think
actually
likes.
B
The
I
feel
when
you
just
come
as
a
newbie
person
is
I
was
like
JSON
is
a
pretty
common
denominator
and
everything
so
far
has
really
designed
around
that.
So
if
it
works
in
JSON,
you
know
Tucson
works.
Everyone
in
particular
works
in
JavaScript
natively,
which
is
obviously
a
very,
very
important
language
nowadays,
so
deviating
from
that
you're
gonna
run
into
some
of
the
same
problems.