►
From YouTube: Package Managers Weekly 2019-08-27
Description
No description was provided for this meeting.
If this is YOUR meeting, an easy way to fix this is to add a description to your video, wherever mtngs.io found it (probably YouTube).
A
A
A
A
Should
probably
go
through
the
you
know
things
we
want
to
do
for
the
week
and
then
leave
time
for
anyone
to
chat
about
my
bike
like
Michael
welcome,
so
we
can
do
that
in
this
view,
it's
a
little
hard
to
tell
if
it's
actually
recording
or
not.
Can
anyone
see?
If
has
it
got
a
little
button?
Okay,
great!
Thank
you.
So
it
is
the
Tuesday
August
22nd,
our
27
package
managers
weekly,
call
also
sprint
planning.
So
we'll
kick
things
off.
Let's
see,
can
someone
have
a
we'll
do
that
thing?
A
D
Basically,
gonna
be
continuing
from
sort
of
the
continuing
stuff
from
last
week
on
building
the
discovery
stuff
into
pub/sub.
Hopefully,
I'll
have
the
how
to
do
discovery.
Things
working
I
have
a
job
for
that
out
today
and
then
tomorrow
can
start
on.
Can
you
know
work
on
alright?
Let's
polish
up
the
pup
stuff
part
of
things.
That's.
E
A
D
A
D
E
Hey
y'all,
so
basically
I've
just
been
lately.
I've
been
thinking
about
how
to
take
the
work
that
I've
been
doing
and
put
it
into
a
like
reviewable
PR.
So
what
that
means
for
at
least
today
isn't
going
through,
like
the
ipfs
config
stuff
and
just
adding
experimental
Flags
and
basically
just
hooking
up
the
stuff.
E
Today
they
can
fix
that's
pretty
obvious
and
then,
after
that,
I
kind
of
have
to
think
of
which,
which
of
the
things
that
have
been
discussed
in
some
of
the
other
issues
like
the
epic
there
or
even
in
some
of
the
research
stuff.
Basically,
what
is
the
most
pressing
issue
to
get
us
towards
something
we
can
have?
That
will
be
like
our
thinkable.
In
my
opinion,
I
guess:
that's
probably
going
to
be
metadata,
some
some
method
of
having
metadata
storage
on
the
node
yeah
and
managing
it.
E
So
that's
something
I
have
to
look
into.
It's
probably
going
to
be
something
specific
to
this,
rather
than
like
a
UNIX,
FS,
v2
or
v
1.5
kind
of
thing.
Just
because
of
the
nature
of
how
this
is
going
to
have
to
work
yeah,
that's
a
lengthy
update
for
me.
Let's
see
anyone
have
any
questions
on
that
before
we.
You
know
just
a.
E
So
I'm,
mostly
just
waiting
on
that
once
it's
there,
we
already
have
some
kind
of
I,
don't
know
what
you
want
to
call
it.
It's
an
unfinished
thing
that
does
it
works
with
the
prototypes?
That's
that's
there,
but
since
it's
not
standard,
it's
not
specs
compliant
or
anything,
it
would
just
probably
be
a
matter
of
adapting
that
you
fit
this
this
back
and
write
test
for
it.
E
It's
a
little
hard
to
do
that
now,
since
we
only
have
that
one
experimental
name
server
that
doesn't
the
values
and
keys
are
backwards
on
that
one,
for
whatever
reason
so
yeah
once
we
standardize
it
I
think
it's
just
fixing
that
up
and
it
shouldn't
take
too
much
time.
It
depends
on
how
involved
we're
going
to
be
with
the
spec
work
and
all
that.
A
A
F
So
last
week,
myself
and
Stephen
works
on
this
new
this
proposal
and
so
I
kind
of
put
together
an
issue.
It's
like
a
draft
version
of
it
and
brief
and
which
elicited
comment
so
Hanna.
Maybe
a
few
comments
there
and,
in
the
meantime,
I'm
implementing
a
concept
I
want
to
try
to
get
to
a
point
where
I
can
benchmarking
to
make
sure.
F
That's
actually
going
to
you
know,
have
the
improvements
that
we're
expecting
and
so
hopefully,
I'll
be
able
to
get
a
benchmark
on
by
the
end
of
the
week
and
in
the
meantime,
like
Panama
has
been
putting
together
a
really
really
nice
test.
Suite.
That's
gonna
be
demoed
later
tonight.
So
once
that's
once
that's
can
I'm
ready
to
go.
We
can.
We
can
actually
do
some
benchmarks
in
like
with
real
machines
and
AWS.
G
Sure
so
I've
been
trying
to
courier
this
async,
it's
very
like
adding
a
single
iterators
to
Joseph
your
best
PR
over
the
mine,
and
so
is
a
big
big
PR,
which
I
broke
up
into
a
bunch
of
small
one,
so
the
small
ones
will
be
merged
ivory
based
the
basic
is
rates
and
adding
agent
integrators
lon
and
like
the
tests
along
with
Joseph,
has
passed,
but
the
interface
test
phone
passed
because
it
needs
a
new
version.
Http
client,
which
then
depends
on
some
other
stuff
that
needs
to
be
merged
but
update
the
right.
G
So
it's
the
the
joke
that,
like
the
ipfs,
the
CTL
module,
which
is
what
how
the
tests
been
up
and
teardown
demon,
so
it
uses
to
run
the
tests
work,
so
that
has
been
ported
to
acing
iterators
and
there's
a
PR
outstanding
on
jeaious
ipfs
to
pull
that
in.
But
it's
failing
on
Windows,
because
Windows,
basically
there's
like
a
bunch
of
outstanding
work
that
needs
to
be
done
to
pull
over
the
line
and
I
feel
quite
bad
about
just
dumping.
All
that
Alan's
laugh
because
he's
only
human
and
only
has
two
hands
to
type
quick.
G
C
C
A
It
took
me
a
second
to
parse
what
you
actually
said:
that's
why
there
was
a
long
pause,
Thank
You
Molly
for
filling
in
so
here
I'll
give
my
quick
thing,
which
is
just
that
same
as
last
week,
making
progress
on
all
of
this.
We
always
have
a
bunch
of
small
like
get
backs
or
whatever,
from
retro
and
I'm,
going
through
all
of
our
old
retro
Docs
to
make
sure
those
battement
and
I
have,
and
so
for
this
week.
There's,
let's
see
I,
think,
there's
an
issue
for
this.
A
Okay,
we've
got
the
package
manager,
okay,
our
thing
which
is
so
close
and
Molly
I
saw
your
note
about
scoring
so
we'll
get
that
this
week
and
maybe
there's
no
humanity
to
make
one
but
updating
the
the
readme
and
other
documentation
around
what
we
do
to
include
the
just.
What
we've
been
up
to
this
this
quarter
and
make
that
a
bit
clearer,
because
the
the
we've
there's
been
an
open
action
from
the
beginning
of
the
quarter
to
turn
you
know
that
is
the
repo
is
no
longer
just
kind
of
research.
A
We,
though,
we've
got
some
other
things
going
on.
So,
let's
make
sure
that's
updated
and
people
can
be
clear
about
what
we're
doing
yep
so
I've
got
something
started
here
on
my
computer,
but
I'll
put
in
a
PR
on
that,
hopefully
soon
when
my
lymph
nodes
aren't
going
crazy,
anyways.
Okay,
any
questions
for
me
or
things
that
excuse
me
that
you
would
like
me
to
move
up
on
my
a
to-do
list.
A
A
C
Sorry
I
got
lost
there
for
a
second
and
iPad
windows.
Yeah
should
I
share
my
screen
with
the
doc.
For
a
second
sure,
it's
not
Michael.
G
C
C
Yeah
I
just
turned
on:
oh
there
we
go
yeah
that
works
cookin
so
this
year.
The
long
short
of
it
is
that
there
there's
a
pretty
big
migration
right
now.
That's
gonna
have
to
happen
in
the
JavaScript
ecosystem,
so
there's
about
a
million
modules
and
then
p.m.
with
just
a
lot,
and
there
is
now
a
module
standard
towards
browser
and
people
have
been
like
quote-unquote
using
this
module
standard
for
a
long
time
through
compilers,
and
so
it
gets
a
little
complicated
to
sort
of
keeping
your
head
like.
C
This
is
actually
like
a
rather
new
thing
in
terms
of
people
using
it
natively
in
the
browser
and
writing
modules
that
work
natively
with
out
a
compiled
tool
chain,
but
that's
always
been.
The
intention
of
the
standard
is
to
you
know,
enable
browsers
to
have
a
native
module
system
that
doesn't
require
extra,
tooling,
very
little
things
if
you
try
to
use
it
today.
C
You
you
end
up
in
this
huge
box
where
you
actually
can't
use
anything
from
him,
including
all
the
stuff
that
is
quote
like
in
the
new
module
standard,
because
all
of
it
assumes
that
you're
running
it
through
a
compiled
tool
chain.
It
actually
breaks
to
try
to
use
it
natively
in
the
browser,
which
is
quite
interesting.
This
kind
of
provides
an
interesting
opportunity
in
that
you
kind
of
need
a
new
registry
and
a
new
sort
of
package
management
flow
for
this
new
style
of
module.
C
For
one
thing,
if
you
do
it
right,
a
package
registry
could
open
up
every
module
ever
created
without
a
sort
of
install
step
without
like
a
sort
of
package
management
step
you
could
just
like
rely
on
it.
Essentially,
it
could
just
be
like
available
as
the
moment
that
it's
published,
which
is
a
very
different
flow
than
any
of
the
JavaScript
package
managers,
have
right
now
it
could
be
him
and
because
there's
such
a
small
selection
but
I
mean
I,
think
selective
public
zero
right
now
of
stuff
in
TM
that
actually
works
natively
in
this
environment.
C
It's
not
a
stretch
to
build
a
new
registry
for
it
because
it
has
such
a
different
flow
and
because
you
know
you're,
actually
there
is
kind
of
a
compatibility
break.
That's
gonna
happen
here.
I
had
done
some
experiments
with
this
thing
called
bundle,
sync,
which
the
idea
was
that,
if
we're
abandoned
like
using
some
ipfs
technologies,
but
not
all
of
my
PFS,
we
could
Ravin
encode
the
chunks
and
then
move
around
like
this
list
of
parts
for
bundles
rather
than
loading
javascript
bundles
in
the
browser.
C
The
amazing
thing
about
this
new
module
standard
is
that,
like
nobody,
is
compressing
and
minifying
these
modules,
like
they
just
load
up
a
tab,
a
browser,
so
we
can
rabbit,
include
them
really
effectively,
and
it
shows
a
lot
of
the
new
location
benefits
that
we
have
the
problems
with
this
module
standard.
Are
that,
like
there
are
all
these
glaring
performance
issues
that
the
they
advocates
of
the
module
system
and
the
browser
vendors
are
saying?
Look
like
people
can
solve
this.
C
The
transport
layer
they
keep
just
saying,
like
people,
will
figure
out
how
to
solve
this
at
the
transport
layer
effectively.
What
they
think
is
that,
like
some
very
fancy
server
side
thing
will
start
using
HTTP
2
and
push
in
order
to
sort
of
prefetch.
All
of
these
modules
nobody's
actually
built
this
yet
like,
like
I've,
seen
demos
that
sort
of
show
that
it
doesn't
work
that
well,
but
I'm
like
nobody's
actually
like
doing
doing
well.
C
What
we
could
do
is
something
different
in
between,
so
we
can
solve
the
performance
issues
in
a
similar
way,
but
do
it
actually
in
the
client
in
a
serviceworker,
where
we
have
a
local
effectively
like
an
IP,
FS
native
store
and
I'll,
get
into
sort
of
like
how
this
separates
from
from
all
of
my
theaters
in
a
second
but
like
in
NTFS
datastore
in
the
serviceworker.
That
would
be
able
to
pull
down
all
these
modules
and
do
all
the
rapid
dipping
and
do
all
the
prefetching.
C
So
it
could
solve
these
performance
issues
and
you
would
get
really
efficient
deduplication
and
you
could
get
stuff
that
we,
we
don't
have
it
all
on
the
web.
Right
now,
like
background
updates
right.
So
when
you
update
your
application,
we
could
get
all
those
assets
in
the
background,
while
the
application
is
running
a
little
service
workers
running
and
by
the
time
you
load
the
new
application.
The
code
is
just
there,
so
this
can
solve
like
a
lot
of
like
kind
of
common
performance
issues
I'm
using
a
lot
of
our
technology.
C
So
it's
it's
a
very
interesting
opportunity
for
us.
If
we
want
to
kind
of
go
after
it,
it's
a
little
bit
greenfield.
We
know
that
the
space
is
going
to
expand
really
dramatically,
and
you
know,
even
if
we
didn't
get
like
some
giant
market
share,
I,
think
that
we
would
end
up
exposing
a
lot
of
developers
to
the
primitives
that
we've
been
building
but
yeah.
That
was
sort
of
the
this
idea
that
I
landed
on
and
tried
to
write
up
here.
A
Since
you
can't
see
me
because
one
of
the
work
I,
what
am
I
trying
to
say
this
quarter,
we've
been
poking
on
some
of
the
core
ipfs
performance
issues
that
are
preventing
other.
You
know
that's
from
engaging
with
other
package
manager.
Work
is,
is,
do
you
know
if
taking
this
round
is
essentially
different
or
if
it
seems
to
me,
we'd
have
the
same
performance
issues
and
we
need
to
solve
those
anyways
I'm,
not
sure
what
I'd
yeah
have
you
along
different
paths?
C
Would
end
up
changing
more
next
quarter
than
this
quarter,
so
it's
sort
of
it
needs
unit
52
for
some
complicated
issues
that
I
can
give
into
with
codex
in
a
little
bit.
It
really
needs
you
necessity
to
to
be
performant
and
I'm
working
on
new
accessory
to
any
way
and
like
plan
to
have
a
JavaScript
version
by
the
end
of
the
quarter
and
that's
kind
of
like
the
main
prerequisites
of
us
working
as
far
as
other
performance
work.
C
According
any
performance,
working,
JSI,
PFS
would
certainly
port,
but
none
of
this
is
using
the
go
infrastructure
like
go
idea.
That's
at
any
point,
so
none
of
that
would
directly
translate
over,
but
it's
not
like
that
would
be
wasted,
work
at
all
and
then
I
think
some
of
the
issues
that
other
package
managers
are
dealing
with
may
not
be
the
same
problems
that
we're
dealing
with
here.
D
So
just
I'm
not
100%
sure
I
I
grasp
all
it's
gone
on
here.
So
just
like
clarifying
a
little
bit
the
idea,
basically
just
to
have
your
your
browser
like
fetchin,
basically
just
cache
updates
for
you.
Our
cache
like
blocks
for
you,
based
on
their
hashes,
that,
like
the
TLDR
for
like
instead
of
yeah,
no.
C
Well,
so
what
happens?
Is
that
the
so
in
in
imports,
basically
like
a
module
import
in
the
browser
turns
into
an
HTTP
fetch?
Basically
like
an
HD,
get
call.
We
can
intercept
that
in
the
serviceworker
and
then
turn
that
into
like.
You
know
a
call
in
to
effectively
MF
s
like
a
needle
file
instance
locally,
like
in
a
local
cache.
So
it's
using
like
effectively
ipfs
should
just
say
like
for
this
Earl
space.
C
All
these
files
are
going
to
come
out
of
idea,
fess
and
since
we
know
that
all
of
these
like
because
we
own
this
publishing
point,
we
know
that
they're
all
being
rather
than
chunked
and
and
all
of
that
we
know
that.
There's
going
to
be
actually
like
very
efficient
chunking
by
hash
of
all
these
files
that
accept,
and
because
we
own
the
publishing
point,
we
can
do
stuff
like
when
a
module
gets
published,
look
at
the
import
graph
and
then,
if
somebody
is
doing
a
cold
start,
we
can
ask.
D
Okay,
yeah
I,
guess
I'm
just
trying
to
like
think
a
little
bit
about
whether
the
people
are
trying
to
solve
this,
not
and
I
confess,
I,
kill
deal
and
they're.
Just
gonna
rely
on
like
okay.
What
they're
called
basically
like
the
the
signed
signed
assets
to
just
have
cash
signed
assets
and
get
like
a
good
chunk
of
the
benefits?
That's
not
so
much
more
work.
D
I
mean
if
the
if
the
idea
is
just
like
I
have
to
fetch
things:
I'll
fetch
them
over
HTTP
from
somewhere,
then
I'll
have
I'll,
have
I'll,
have
hashed
things.
So
I
know
that
they're
the
same
across
various
different
websites
whenever
I
need
a
module,
then,
even
without
like
IP
LD
and
multi
formats,
you,
you
can
still
get
like
right.
There's
like
a
standard
forgetting,
like
forgetting
like
hashed
objects,.
C
C
If,
if
we
have
the
service
order
there-
and
we
can
already
get
like
you
know,
an
entire
sort
asset
tree
and
and
even
within
that
asset
tree,
we
had
video
application.
I,
wouldn't
I,
don't
see
why
we
would
end
up
using
my
packaging
and
and,
like
I
said,
the
standard
is
actually
stalled
because
Mozilla
is
objecting
to
it
like
on
principle,
so
I
don't
actually
I,
don't
think
that
that's
going
to
go
anywhere.
The
Dietrich
has
been
a
little
bit
more
involved
in
that
I.
Think
and
Jim
tech
I
think
has
been
involved.
C
Okay,
so
Michelle's
question:
what's
the
big
selling
point
to
package
manager
maintainer
and
if
a
consumer,
so
the
biggest
benefit
to
a
consumer,
is
that
they
don't
have
to
use
a
quote
unquote
package
manager
like
there?
There
is
no
pull
this
down
and
run
it
into
my
compiled
chain
step.
It's
literally
like
I,
I,
just
start
using
modules
and
I
can
use
anything
in
the
entire
industry.
So
it's
like
actually
a
really
nice
much
better
workflow
from
web
developer
perspective
on
their
end.
C
As
long
as
you
can
solve
these
performance
problems
that
are
there
with
the
SM,
then
it's
a
huge
benefit
to
them
and
a
much
easier
sort
of
package.
Consumption
experience
on
the
package
for
I
think
package
manager
maintainer.
That
would
be
us
in
this
case.
I
think
patent
package,
maintainer
czar,
certainly
an
audience
though
like
package
maintainer,
is
that
people
publishing
packages
and
I
mean
the
benefit
to
them.
Literally,
is
that
you
know
it's
very
easy
for
anybody
to
now
use
their
package.
C
I
think
like
early
in
a
new
ecosystem
like
this,
the
incentives
are
a
little
bit
different
than
they
are
in.
Like
a
long
lending
system,
I
think
that,
like
if
you
have
something
new,
that's
using
a
lot
of
like
newer
technologies,
a
lot
of
the
people
that
show
up
are
actually
the
people
that
want
that
are
like
interested
in
green
field
spaces
and
want
to
publish
the
first
packages
in
that
space
like
in,
like
I,
think
that
you
know
the
first
people
that,
like
you,
know,
do
their
web
framework.
C
All
in
this
way,
style,
you
know,
will
get
you
know
some
cool
points
or
whatever,
and
that
attracts
a
lot
of
people.
We
saw
this
in
the
early
node.js
days
too,
because
we
also
had
like
no
packages
and
everybody
got
to
show
up
and
write
all
the
first
stuff
and
then
the
tracks,
like
a
certain
type
of
person,
but
I,
think
like
over
time.
The
benefit
really
is
that,
like
this
is
a
way
for
me
to
publish
my
work
you
to
a
global
ecosystem
that
is
immediately
accessible
to
every
browser
user
in
the
world.
C
Yeah
I
really
didn't
figure
out
that
this
would
be
a
thing
until
I
started,
trying
to
actually
use
yes
I'm
in
the
browser
natively
and
quickly
realized
that
I,
just
I
couldn't
use
anything
from
NPM
that
you
sort
of
are
on
your
own,
like
yeah
you're
running
all
of
your
own
dependencies
from
scratch.
Now,
because
there
is
no
package
manager
overlay,
that's
only
doing
like
new
style
packages,
some
resource
integrity.
B
I
kind
of
I
have
a
spidey
sense
that
that's
going
to
be
actually
a
lot
more
work
than
than
the
couple
bullets
it
seems
within
this
talk
and
that,
like
the
gateways,
will
actually
end
up
investing
like
if
we,
if
we
proceed
this,
pursue
this
forward,
and
it
even
has
like
modest
uptake
continuing
to
to
support
that
and
make
it
a
good
experience
for
people
and
not
burn
them,
might
might
be
a
significant
chunk
of
effort.
So
that
should
be
like
I
guess,
weighed
against
the
other.
The
other
side
of
things.
C
So
I
did
some
of
this
already
on
consular
workers
when
I
was
doing
the
bundles
thang
experiment,
because
I
was
thinking
like
it
would
be
nice
that
this
was
just
already
in
a
CD
end
and
even
the
sort
of
smart
end
of
it
was
also
like
had
an
edge
cash
and
it's
just
a
lot
less
infrastructure
to
run,
because
you
we're
not
putting
servers,
we're
not
managing
all
that
stuff.
It's
just
sort
of
like
automated
and
it's
certainly
doable
too
like
we
could
do
all
of
this
with
cloud
letter
orders.
C
Definitely
you
you
have
to
kind
of
pare
down
the
number
of
codecs
and
hashes
that
you're
gonna
use,
but
since
we
own
the
publishing
side
of
it,
we
can
make
sure
that,
like
we're
not
just
taking
arbitrary
like
blocks
and,
however
that
doesn't
mean
that
there's
nothing
to
ever
manage
and
that
there's
not
like
a
potential
cost
Center
so
and,
for
instance,
like
I'm,
using
the
club's,
our
key
value
store,
which
is
incredibly
fast
and
really
nice.
And
you
know
it
also
has
a
2
Meg
block
limit.
C
So
it's
like
really
really
fits
nicely
inside
of
our
stack
actually,
but
one
of
the
problems
with
isn't
like
I,
don't
know
what
the
cost
structure
of
that
would
look
like
over
time
and
if
we
would
have
to
actually
like
move
to
a
different
storage
structure
and
then
just
use
cloud
for
workers
as
sort
of
the
compute
layer
as
a
computer.
It's
it's
pretty
cheap.
C
C
So
we
would
want
to
look
at
opportunities
to
reduce
that
cost,
like
maybe
partnering
with
with
Leffler
and
saying,
like
hey
when
you
kind
of
co-sponsor
this
with
us
and
host
the
infrastructure
like
something
like
that,
I
think
could
really
help
us,
but
yeah
I'm
much
less
worried
about
the
infrastructure
maintenance
side
of
it,
because
we
can
build
it
this
way,
but
I
am
worried
about
the
infrastructure
cost
still
so
so.
Yeah
Andrew,
if
so,
enter
gonna,
come
into
my
pika.
So
pikas
awesome,
I,
love
them
and
a
lot
of
this
I
don't
I.
C
Don't
think
that
I
would
have
gotten
to
this
place.
Sort
of
mentally
I
hadn't
dug
into
the
pika
stuff
before,
but
pika
is
sort
of
punting
on
on
both
the
registry
and
the
transport.
So
pika
is
trying
to
find
modules
an
NPM
that
they
can
sort
of
make
work
with
how
to
compile
step,
which
is
like
a
pretty
minimal
set,
and
then
they
they're
not
doing
anything
with
the
transport
layer.
C
So
all
of
the
performance
problems
that
I
talked
about
earlier,
with,
like
every
individual
file,
basically
being
a
new
round
trip,
that's
all
still
an
issue
with
pica
and
they
they're
I.
Don't
think
that
they're
even
really
working
on
that
I
think
they're
actually
looking
at
like
how
can
they
improve
some
of
the
other
parts?
The
experience
so
I
think
that
if
we
did
something
interesting
enough
that
they
would
end
up
by
they're
like
using
our
stuff
or
merging
with
us
or
like
I
didn't
go
there
one
end
up
being
a
lot
of
collaboration.
C
Yeah
I
mean
I,
I,
didn't
I,
didn't,
say
the
thing
about
cloud.
Flair
sort
of
offhand
either
I
think
that,
like
there,
there
may
actually
be
a
real
opportunity
there
for
us
to
partner
with
them
they're
they're
really
looking
for
like
they
would
love
it.
If
all
of
new
JavaScript
developers
were
pulling
for
their
infrastructure,
they
would
definitely
love
that
there
they
already
the
node.js
foundation,
uses
them
as
a
CDN
for
all
of
for
all
of
no
de
estar
balls,
and
they
provided
that
for
free
for
years.
C
A
C
Well,
in
terms
of
resourcing
this
wouldn't
pull
anybody
from
go,
even
if
we
went
all
in
on
it,
because
literally
all
of
the
work
is
in
JavaScript,
including
the
the
worker
stuff,
but
I
mean
you
could
still
see
it
having
a
huge
impact
there
and
it's
not
like.
We
don't
have
the
same
sort
of
needs
and
wants
of
growth
more
generically
on
the
side,
the
other
side.
C
So
it
definitely
still
applies,
like
I
said,
like
the
sort
of
the
most
important
parts
of
if
I'm
gonna
have
to
build
this
corner
anyway,
and
so
there'll
be
sort
of
meaningful
progress
in
that
direction.
So
we
have
quite
a
while
to
just
sort
of
like
think
about
it
and
stew
on
it
before
we
need
to
think
about
prioritizing
or
resourcing
it
or
you
know
little,
you
know
taking
anything
else
off
of
the
plate
or
off
of
our
priority
list.
C
So
you
know
I'll
continue
to
kind
of
plug
away
at
it
and
and
maybe
get
something
demo
a
bowl.
And
then,
at
that
point
you
know
if
it's
something
that
we
feel
like
investing
and
further
than
we
could
maybe
change
some
resourcing
of
it,
but
I.
Don't
think
that
this
needs
to
pull
away
from
anything
that
we're
currently
working
on
right
now
and
I.
Don't
think
that
it
really
warrants
it
at
this
stage,
got
it
thanks.
I.
F
Just
want
to
say
it's
a
really
cool
idea.
You
know
it's!
It's
really
nice
to
hear
your
like,
exploring
something,
that's
kind
of
going
in
a
very,
very
ambitious
direction
and
I
like
the
idea
of
doing
something
much
more
efficiently
than
what's
out
there
at
the
moment,
so
I.
Definitely
it's
worth
with
kind
of
pursuing
somewhere
at
least
to
see
to
have
my
abilities
in
practice.
C
There's
like
this
import
map
standard
that
could
make
some
of
the
the
pre-loading
happen
like
immediately
before
even
the
serviceworker
is
installed
and
stuff
like
that.
So
things
are
gonna,
keep
progressing
like
in
the
web
platform
in
a
direction
that
like
aligns
and
even
like
improved
this
idea.
So
but
at
the
same
time,
like
I,
don't
see
anybody
really
working
on
this
problem
all
that
diligently.
So
it's
not
like
we.
We
need
to
change
everything
we're
doing
and
going
on
it
like.
We
can
keep
messing
with
it.
Yeah.
F
I
guess
my
next
question
there
would
be.
Do
you
think
it's
something
that
we
could
get
people
excited
about
enough
that
they'll
work
on
it?
For
us
I
mean.
C
It
needs
to
be
working
and
then
the
work
it
needs
like
yeah,
I,
think
that
people
can
actually
use
and
then
I
think
yeah.
A
lot
of
people
I
think
would
help
out
like
there
are
parts
of
it
that
I'm
not
as
worried
about
doing
a
ton
of
development
on,
like
you
know,
making
the
the
command
line.
You
know
have
a
bunch
of
cool
features
like
people
will
fill
that
part
in
I'm.
Not
bringing
about
that
I'm.
C
Yeah
so
there's
some
stuff
about
tank
and
entropic,
so
I've
actually
I've
talked
to
the
Isaac
quite
a
bit
about
tank,
because
we
still
hang
out
of
a
few
months.
So
one
of
the
things
that
is
going
to
weigh
and
pay
it
down
in
this
transition
is
that
the
moment
that
you
use
any
of
the
existing
sort
of
million
modules
from
NPM
from
one
of
these
new
style
modules,
you
now
can't
use
the
module
natively
because
part
of
its
like
Deb
tree
now
has
this
thing
in
it
that
actually
needs
a
compile
step.
C
So
it's
it's
more
of
an
ecosystem
problem.
It's
actually
it's
very
similar
to
the
issues
that,
like
you
know,
Python
has
had
for
years
trying
to
adopt
like
new
concurrency
patterns,
which
is
that
they
have
this
huge
ecosystem
of
modules
that
use
blocking
I/o
and
then,
when
you
try
to
use
non-blocking
I/o
you
you're
like
now
in
this
corner,
you
go
system
at
the
moment
you
use
one
of
these.
C
You
like
break
your
entire
sort
of
service,
and
so
like
they're
gonna,
have
a
very
similar
problem,
getting
like
the
existing
and
kami
ecosystem
to
function.
Well,
with
like
a
native
system
yeah
in
tropic,
it's
sort
of
built
on
content,
disability
I'd,
like
I,
mean
we
can
get
more
into
entropic
at
some
point.
But
I'm
not
gonna,
put
a
lot
of
faith
in
that
project,
for
a
variety
of
reasons
like
I.
C
Don't
think
that
that's
really
gonna
go
anywhere
like
you
know
more
power
to
them
if
they,
if
they
do
I
mean
they're,
not
have
a
feel
about
the
maintainer
they're,
not
the
people
that
I
would
choose
to
do
that,
like
they're,
not
great
at
like
building
a
community
or
working
with
others
supporting
an
ecosystem.
C
But
yeah
I
don't
know.
Yeah
I
mean
I've
worked
at
the
mill
of
stuff
they're,
very
smart
engineers,
but
well
I
I,
don't
like
it's
very
hard
to
see
heartbreak
in
the
npm
ecosystem
for
any
reason
other
than
like
npm,
literally
failing
like,
like
literally
just
going
offline
or
collapsing
entirely.
C
C
I
mean
we
saw
this
a
little
bit
with
yarn
right,
which
was
like
you
know,
like
they
had
the
benefit
of
like
the
largest
framework
in
like
blunders
days
from
work
in
the
world,
like
you
know,
pushing
their
stuff
and
and
really
hyping
them
and
they're
still,
like
you
know,
not
a
huge
factor
in
terms
of
the
overall
ecosystem.
I.
C
Think
that,
like
like
the
the
shift
here
like
if
you're
gonna,
think
about
a
new
package
management
or
JavaScript,
there
has
to
be
something
more
fundamental.
That's
gonna
push
people
in
that
direction
and
I
think
that
that
is
this.
Just
native
module
experience
that
is
now
enabled
in
the
browser
like
we've,
been
using
modules
in
the
browser
exceeds
compile
chains
for
like
10
years.
So
people
haven't
really
noticed
how
different
it
is
and
how
much
the
platform
has
caught
up
when
you're
doing
things
natively
yeah.
C
Oh
Dino
yeah,
Ryan
yeah
I
talked
with
Ryan
a
lot
about
Dino.
Actually
do
you
know
it
has
some
interesting
ideas
about
package
management
and
about
how
sort
of
modules
should
work.
It
could
be
interesting.
C
C
It's
like
good
browser
constraints
are
incredibly
complicated
and
they're
multiplicative
so
like
when
you
layer,
one
on
another
you're
like
it
here
to
continue
to
botch
yourself
in
a
more
and
more
corner,
and
none
of
the
the
existing
package
manager
is
like
in
topic
or
NPM
or
really
anybody
are
trying
to
live
within
the
constraints
at
the
browser.
All
of
them
come
from
some
node
background
like
like
me,
honestly
and
so
you're
used
to
having
this
incredibly
flexible
environment.
C
C
Yeah
yeah
yeah
there's
a
comment
to
that
xhr
and
when
people
realize
that
that
such
I
could
do
things,
that
was
a
really
important
thing,
though,
because
like
people
feel
it
realize
that,
if
things
are,
could
you
think
right
when
Microsoft
stop
dumping
a
browser
when
they
had
done
that
market
share?
So
literally,
like
the
only
new
things
that
happen
in
web
development
for
like
four
or
five
years
with
xhr
and
all
the
kind
of
hacks
on
top
of
it?
And
then
you
know,
Missoula
had
to
take
enough
market
share
away
from
Microsoft.
C
I
Wouldn't
mind
talking
about
experience
with
speaker,
browser
Paul
is
building
bigger,
browsers
sort
of
arrived
at
this.
Yet
try
experimenting
with
yes
modules
a
year
ago,
and
it's
just
for
the
exact
same
reasons
like
because
it's
like,
if
you
don't
remove
the
bundler
from
the
equation,
works
really
well
with
decentralized
use
case
and
everything
gets
cash
everywhere.
It's
actually
pretty
awesome
and
also
an
appoint
another
project
from
the
dot
community
which
actually
works
on
ipfs
from
Ranger
Bob.
I
He
seemed
that
that
2.0
leading
that
and
before
he
got
that
contract
he's
working
on
this
project
when,
in
his
little
projects
it's
called
web
read
and
it
allows
you
to
run
sort
of
unbundled
yes
modules
for
nodejs,
please
the
same
modules
that
you
use
in
the
browser
yeah
you
can
put
them
on
like
FS
and
then
load
them
into
node
or
dad.
So.
I
I
C
I
C
So
one
cool
thing
here:
if
you
look
at
this
syntax
down
down
the
way
for
importing
from
that
URL
and
from
an
IP
FS
URL,
so
you
can
do
that
in
node
because
know
it
has
a
hooks
infrastructure
in
inside
of
its
like
the
new
modules
and
tags.
To
do
that,
you
can't
do
that
in
the
browser
like
they
turn
into
HP
year-olds,
but
if
we
have
a
serviceworker
installed,
what
you
could
do
instead
is
just
have
an
Earl
space
that
is
like
underscore
at
the
FS.
C
C
If,
if
you
have
this
the
service
worker
that
a
lot
of
people
are
using
in
their
apps,
you
can
start
to
just
sort
of
add
other
protocols,
and
you
don't
even
have
to
really
pollute
the
bundle
you
can
have
all
that
get
loaded
and
dynamically.
Once
people
request
stuff
out
of
those
networks,
you
can
do
some
pretty
cool
stuff.
C
Okay,
so
reproducibility
of
the
dependency
graph
locks
and
bundlers-
yes,
yes,
so
yeah
I,
didn't
I
didn't
want
to
get
into
that
because
it
gets
like
really.
It
gets
really
nerdy
and
early
detailed,
but
effectively
so
because
this
is
like
an
IPL,
be
the
registry
of
the
methyl-p
graph,
and
so
the
Tildy
graph
has,
like
you
know,
every
name
of
every
package,
a
link
to
some
metadata
and
a
link
to
in
a
unix
of
sq2
directory.
C
So
if
you
want
to
do
something
like
a
lock,
for
instance,
you
could
really
easily
generate
like
a
hash
of
that
entire
tree
and
then
load
and
then
specifically
load
that
hash.
So
you
lock,
like
creating
sort
of
lock
infrastructure,
is
really
easy
once
we're
on
top
of
sort
of
once
the
industry
is
on
top
of
this
content
address
tree,
there's
also
some
interesting
stuff.
C
You
have
the
same
thing
if
you
just
load
a
bunch
of
script
tags-
and
you
just
did
a
big
push
to
the
infrastructure
you,
you
could
potentially
wind
up
in
a
weird
situation
where
people
get
like
half
of
the
resources
for
the
new
site
and
half
of
the
ones
for
the
old
one,
unless
you
have
like
really
complicated
infrastructure
to
solve
that.
But
in
this
system
the
registry
has
a
single
hash
for
the
root
node,
the
equity
graph.
C
So,
at
the
point
that
the
application
loads,
we
can
actually
lock
the
State
of
the
registry
and
know
that
every
dependency,
like
all
the
dependency
resolutions,
are
getting
the
right
thing.
Even
if
you
didn't
explicitly
lock
it
right
and
then,
once
an
update
happens,
we
can
go
into
a
lot
of
those
updates
and
then
also
go
from
like
load
the
application
again
from
the
same
state.
B
Had
a
hand
on
a
slightly
different
topic,
we're
just
I've,
think
I'm
subscribed
to
ipfs
notes
or
something
and
noticed
some
movement
in
the
last
three
days
or
so
on
a
note
about
IP
FSA
Nix
integration,
which
was
curious.
If
folks
had
taken
a
look
at
it.
Kind
of
toward
the
end
turned
turning
to
more
of
a
thought,
thought
exercise
around
blockchain
layers
and
and
using
that
as
kind
of
a
layer
above
these
sorts
of
things.
B
C
H
H
C
H
C
D
H
Does
provide
that
sort
of
rightness
quicks
does
I,
don't
think
things
as
well.
We're
like
it's
trying
to
be
reproducible,
so
the
build
script.
Tells
you
exactly
how
much
you
get
if
you
find
a
lot,
but
you
could,
in
theory,
check
it
and
the
ideas
that
people
would
check
this
and,
like
you,
get
you'd
ask
multiple
parties
like
okay,
what's
the
hash
of
this,
these
things
than
you
but
yeah
like
honest
in
this
case,
he
should
just
like
point
to
then
just
you
point.
D
H
C
So
we're
I
have
that
issue
open
about
the
deeps
Enix
sp1,
adding
file
metadata.
Well,
it's
adding
file
like
permission
data.
It's
all
structured
right
now,
should
we
add
just
like
a
generic
string
field
that
we
don't
say
what
it
should
be.
C
Like
yeah,
ideally,
would
be,
ideally,
we
could
just
add
an
arbitrary
CID,
but
we
don't
because
everything
is
typed
where
we
don't
want
to
like.
We
don't
want
to
add
a
field
just
for
this
use
case,
and
do
we
want
to
have
something
there
so
that
when
people
do
hacks
they're
not
like
the
worst
hacks
ever
but.
H
H
C
All
right,
I'll
poke
around
this
little
bit
and
see.
If
maybe
we
can
do
something
before
I
land.
Those
and
changes
to
the
COBOL
cool.