►
From YouTube: 📦Package Managers WG Weekly Sync April 9, 2019
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
Hello,
welcome
to
the
package
managers
weekly
sink
where
we
play
the
game
over
what
we
did
last
week,
what
they're
locked
on
and
what
we're
going
to
do.
Next,
you
can
brain
I
am
Yost,
that's
it!
That's
the
intro
cool
yeahbsolutely
link
to
the
quick
parted
in
the
ring.
You
could
put
your
name
on
the
attendees
by
Stephen
and
yeah.
Put
your
updates
down
and
we'll
take
notes.
That
kind
of
thing
so
can
I
get
a
look.
Take
companies.
A
Thank
you.
Ollie
I
will
go
first,
so
yeah.
Last
week
there
was
sort
of
fun
with
NPM
that
happens,
and
there
was
a
compilation
on
Twitter
where
phaidor
said.
Why?
Don't
we
to
a
distributed
version
of
NPM
and
some
people
talked
about
some
stuff
and
then
I
kind
of
waited
in
because
they
were
saying?
Oh
yeah,
we
tried
I,
suppose
very
slow.
I
said
well
we're
working
on
that
forgotten.
The
conversation
with
cat.
A
A
Know
as
a
PR
open
against
for
code,
which
should
I
don't
even
know
if
I'm
pronouncing
this
thing
properly,
pronates
Portuguese,
so
I,
don't
know
yeah,
so
I
keep
calling
it
but
good
day
and
explain
that
later,
but
yeah
anyway,
so
that
pay
I
went
in
I
was
cool.
Hopefully
they
might
merge
it.
And
then
we
were
over
my
PFS
and
I
can
testicle
in
NPM,
which
can
be
a
pretty
big
win,
yeah
cool.
A
C
There's
ongoing
just
can
you
meet
there's
ongoing
discussions
about
what
a
npm
on
ipfs
actually
look
like
if
there's
nothing
else
on
the
agenda,
for
this
meeting
be
good
to
get
people's
thoughts
on
that
one
of
the
one
of
the
most
boring
questions,
but
also
fundamental
to
how
we
start
this
seems
to
be.
Should
it
be
a
specific
that
so
that
the
the
backstory
is
a
desktop
app
seems
like
a
neat
way
of
tricking
people
into
running
along
and
running
a
demon
in
the
background,
so.
A
C
Right
now,
if
you
run
npm
ipfs
ampion,
you
get
a
shortened
over
to
chase
ipfs
in
process
daemon
that
fetches,
so
it
basically
encourages
you
to
behave
like
a
leech.
When
you
probably
came
to
this
project,
hoping
to
not
do
that,
if
you're
a
kind
of
the
gamete
Arian
co-host
o
chances
are
you
were
hoping
to
do
some
co-hosting.
So
then
the
thought
was
a
desktop
app
similar
to
where
ipfs
desktop
has
ended
up
running
in
your
menu
bar
or
task
tray.
C
That
just
kind
of
gives
you
a
quick
way
to
see
like
basic
stats
on
NPM
play.
This
is
something
that
you
don't
get
from
NPM
today,
so
just
knowing
from
my
usage,
which
modules
I
depend
on
regularly
and
that
could
also
surface
like
interesting
things
in
your
subtree
and
then
also
thing
able
to
demonstrate
back
like
you've
downloaded
these
modules
a
million
times
and
you've
rehearsed
like
you've,
given
back-end
copies
of
the,
for
example,
as
a
way
of
encouraging
the
behavior
we
were
hoping
to
see.
Eden
is
that
the
Halfhand.
D
Yeah
I.
Yes,
there
is
a
conversation,
remember
hearing
like
a
number
of
months
ago.
Basically,
this
is
that's
sort
of
the
idea
people
are
gonna,
think
of
which
is
yeah.
Let's
looks
bundle
in
a
demon
having
a
desktop
app.
That
does
our
thing
and
then
you
end
up
with
like
10
demons
running
on
your
machine
that
are
all
just
running,
occupies
nodes.
Yes,.
C
D
C
C
B
C
E
C
Yeah,
so
about
what
Aidan
has
touched
on
is
the
first
like
thing
up
like
an
NPM
when
I
open
this
app.
So
my
gut
feeling
was
my
specific
app
would
be
more
compelling
to
the
large
audience
of
NPM
users
and
I
had
some
pushback
on
that
from
various
people.
They
were
like.
Why
not
bolt
it
into
IP
this
desktop,
because
we've
already
got
some
users
on
this
desktop
and
don't
you
don't
want
to
have
multiple
demons
running
and
my
feeling
was
that
that
wasn't
the
right
criteria
to
judge
the
decision
on
Stephen
so.
C
Indeed,
so
this
was,
this
was
kind
of
one's
original
dream
for
the
web
UI
and
the
desktop
app
was
that
it
would
also
become
a
kind
of
like
an
app
store
is
one
way
of
describing
it.
Another
is
like
a
shell
yeah.
It's
like
you
know,
we
I,
guess
that's
what
provides
the
services
that
ipfs
apps
could
use.
It's
certainly
a
plausible
direction.
My
gut
feeling
was,
as
we
haven't
done,
that
yet
and
we'd
like
to
achieve
the
immediate
goal
of
more
users
on
NPM
on
a
quest.
B
So
you
know
newbie,
so
forgive
me
if
I'm
making
sort
of
silly
assumptions
but
I
mean
you're
talking
about
sort
of
scrapping
your
way
into
being
a
FS
user
in
many
different
ways,
and
that
might
just
be
like.
Oh,
you
know,
I've
used
this.
B
The
whole
package,
manager'
others
like
oh
I've,
been
using
these
I've
been
using
ifs
and
NPM,
and
this
is
great,
and
so
let
me
look
at
using
this-
maybe
not
even
sort
of
an
app
store
metaphor,
the
metaphor
of
just
like
a
bunch
of
building
blocks,
so
that
no
matter
what
your
starting
point
is
that
you
come
in
on.
You
can
then
add
more
things
to
it
to
be
more
more
full
function.
Function
I,
give
this
world
II.
B
So
the
idea
being
that
are
facing
you
certain
Lego
bricks
and
you
add,
more
as
you
need
to
become
more
sophisticated
as
a
user
as
you
go
along,
but
but
in
doing
so
you
never
have
to
relearn
the
interface
for
the
first
thing
that
you
learned
how
to
do
so.
He
came
in
just
using
this
as
an
alternative
to
regular
NPM
that
experience,
even
as
you
become
like
I
guess,
power
user
would
be
exactly
the
same
thing.
The
long-term
make
sense.
B
C
Sure,
but
it's
interesting
because
every
time
you
have
this
conversation,
it's
like
lots
of
the
GUI
team
developers
were
initially
like.
No,
no
make
it
part
of
desktop
and
it's
like
and
their
fear
was
like
multiple
demons,
and
that
was
the
first
moment.
Aiden
said
demons,
so
I
think
it'd
be
worth
unpacking
that
feeling
a
bit
further.
B
C
D
I
just
wanted
to
clarify
it.
I
think
the
desktop
like
having
an
installed
thing
makes
sense
because
from
like
a
user
experience
thing
it's
like
this.
Is
it
this
feels
like
a
different
thing,
but
from
a
technical
side
we
should
make
sure
there's
a
way
to
not
have
not
have
to
run
another
demon
right
in
the
same
way
that
if
you
run
multiple
things
that
I'll
use
fuse,
you
don't
need
to
have
like
30
fuse
installations.
C
So
that's
the
kind
of
model
that
I
was
imagining,
but
then
there's
this
kind
of
broader
question
of
like
actually
should
be
totally
legit
for,
like
10
apps,
on
your
system
to
spawn
their
own
IP
those
demons
with
their
own
like
cartooned
configs,
like
maybe
you
want
a
specific
set
of
bootstrap
nodes
for
this
specific
act
like
just
caps.
Specific
swarm
seems
like
a
reasonable.
A
reasonable
thing
like
I
would
want
that.
D
C
F
C
D
I
I
had
a
question
which
is
I,
saw
the
the
amazing
video
of
attempting
to
run
IPs
things
and
having
a
wait
around
for
really
long
time
as
much
appreciated
there.
I
guess
I
want
to
know
what,
how
much
faster
it
needs
to
be
and
like
what
the
use
cases
are
in
this
in
the
sense
of
the
lower
down
you
go
in
like
the
manifest
tree
where
the
IPS
references
are
it's
like
each
package,
then
you
have
to
be
much
faster.
D
A
Think
we
wouldn't
necessarily
have
any
control
over
the
way
entries
go,
because
people
will
just
publish
their
modules
to
say
with
the
dependencies
of
published
over
IP
owners
and
would
just
be
a
thing.
And
then,
when
you
install
your
dependencies,
some
differences
in
the
tree
might
come
via
IP
enesta.
Some
might
come
for
HTTP,
so
really
as
to
be
as
fast
as
possible.
I
mean.
A
So
they're
the
PR
but
I
submitted
to
Paco
T
had
two
ways
of
putting
down
that.
So
you
either
user
CID
can
I
give
us
the
idea,
in
which
case
you're,
basically
linking
to
a
toggle.
So
you
don't
get
any
version
resolution
of
that
and
then
the
IPS
name
I
was
using
links
to
a
package
manifest
or
apartment
which
contains
all
the
available
versions,
which
then
have
C
IDs
or
HTTP
links
to
the
tables,
and
so
so
the
user
would
do
either
and
then
all
the
dependencies
would
do
either.
F
So
since
I
got
called
out
once
already
and
I'm
guessing,
you
guys
are
kind
of
recruiting
for
things
that
could
hypothetically,
maybe
be
okay,
our
material
or
something
one
of
the
things
that
I
would
really
love
to
turn
into
an
ask
and
kind
of
recruit.
A
bunch
of
other
people's
thoughts
about
is
what
we
could
hypothetically
get
in
the
long
run
out
of
package
manager
designs,
which
would
be
composable
in
a
way.
That's
not
right
now.
F
E
We
had
a
lot
of
discussions
are
instant
cookers
of
G
X
or
basically,
we
talked
about
was
using
NPM
cargill,
whatever
to
extract
the
set
of
dependencies
and
you've
dumped
them
to
lock
file,
and
it's
locked
file
describes
like
how
to
actually
pull
these
pieces
together
so
like
that,
it
means
that
you
aren't
entire,
like
you
are
digging
too
deep
into
the
Refn
system.
Fertilized
feed
still
have
3p
disability
and
it
also
looks
like
cross
back
to
Twitter
very
like
if
you
look
at.
E
F
Lot
of
different
ways:
there's
some
things:
I
can
think
packages
and
then
there's
like,
if
you
try
to
think
about
doing
this
ecosystemic
clean.
Well,
it's
going
to
allow
me
to
like
keep
building
taller
pyramid
on
top
of
smaller
blocks
and
individual
packages.
Then
I
think
how
will
you
come
down
somewhere?
Our.
E
Goal
was
mostly
to
not
kill
the
pyramid,
go
to
the
monolithic
rheticus
or
the
top-down
route.
We're
like
I,
have
a
package
and
then
like
I,
decided
okay.
This
is
now
I
want
to
declare
like
this
package
is
using
a
dresser
with
me
and
they
have
locked
everything
plates.
This
means
I,
don't
have
like
I,
don't
take
evidence
in
trying
to
get
into
the
system
I,
just
like
figure
out
whatever
which
burns
everything
I
should
use,
lock
that
down
and
then
I
have
like
plugins,
with
every
single
different,
like
package
manager
that
Windows
uses.
C
A
A
C
D
Is
that
I
can
grab
some
questions
I'm,
trying
to
how
deep
do
you
think
the
like
I
guess,
the
time
and
depth
of
of
a
set
of
package
dependencies?
Do
we
want
to
be
able
to
resolve
over
IP
NS,
just
sort
of
trying
to
target
like
how
fast
it
needs
to
be
Steven,
and
some
other
folks
are
interested
in
like
how
do
we
maybe
have
like
a
semi
centralized
thing
to
help?
D
B
D
Questions
to
answer
the
thing
is
as
long
as
there's
only
one
maintain
as
long
as
there's
only
one
place
to
get
the
data
from
it
has
to
be
slower
because
you're
adding
a
level
of
indirection
right
instead
of
talking
directly
to
the
guy.
With
the
data
you
talk
to
someone
who
tells
you
to
talk
to
someone
else
with
the
data,
so
it
the
latency
has
to
be
longer,
and
the
question
is
like
what's
acceptable,
given
that
in
the
near
term,
I
don't
see
like
80
people
running.
A
Danny
we're
gonna,
you
know
we
stand
ups
and
infrastructure,
the
house
because
of
things
you
know
like
club
from
that
kind
of
thing.
You
come
from
clubs
like
like
creating
ipfs
kind
of
nodes
on
there
as
part
of
their
CDN,
but
anything
is
always
gonna,
be
the
case
that
you
know.
There's
only
a
few
people
in
the
world
had
the
thing
that
you're
looking
for
so
I'm.
D
Not
always,
but
in
the
short
term,
like
you
know,
by
q2,
that's
I
think
that's
fair,
so
the
question:
how
fast
does
it
need
to
be
like,
like,
for
instance,
I
would
be
I
would
be
okay,
saying
you
know,
let's
bring
our
penis
to
like
you
know
a
pretty
reasonable
time
of
just
whatever
it
takes
to
resolve
a
peer
ID
on
the
network.
D
C
D
C
D
E
D
D
C
F
So,
for
what
it's
worth,
I
can
bring
up
an
example
from
a
package
manager
community.
That's
quite
recent.
The
geeks
project,
while
being
quite
fans
of
the
potential
decentralization,
have
also
been
in
a
lot
of
talks
recently
with
them.
I
forget
the
exact
title
of
the
group
software
heritage
or
something
like
this
is
a
very
archival
oriented
project
and
they
are
centralized
and
I
have
a
latency
that
gets
done
and
the
geeks
people
are
happy
to
work
with
that.
So
I
think
that's
a
trade
off.
That's
certainly
not
like
to
consider.
D
A
D
D
If
we,
if
it's
something
we
we
need,
then
either
means
we
need
to
do
DNS
things
or
I
need
to
put
time
into
helping
look
the
appeal
with
rendezvous.
What
we
need
to
put
more
pressure
on
them,
or
we
have
to
say
well,
it's
okay,
we'll
do
it
next
quarter
as
long
as
it's
done
by
then
that's
okay
and,
like
that's
the
intermediate
space,
where
I
don't
really
know
what
we
need.
A
C
Feels
like
the
fewer
unknowns
with
the
dns
version
of
this
story
and
that
it
has
a
centralization
trade-off,
but
it
it
can
be
a
crutch
that
we
use
to
pushes
forwards,
I.
Think
being
massively
like
numerous
orders
of
magnitude
slower,
it's
going
to
really
rest
for
a
long
time.
I,
don't
know
if
the
specific
working
or
to
get
the
DNS
trick
working,
but
it
seems
to
be
the
like
this.
Can
this
can
boost
us
through
the
next
couple
years?
Great.
D
Perhaps
like
the
rendezvous
is
needed
for
other
service
discoveries
anyway,
it
doesn't
support
other
types
of
like
it.
Wouldn't
support
multi
writer
I
can
us,
but
the
rendezvous
would
it's
yeah
and
and
in
any
event,
even
like
the
basic
changes
we're
gonna
make
are
gonna
also
improve
by
one
order
of
magnitude.
What
IPS
is
doing
question
is:
do
we
need
to
this
quarter
or
is
like
one
sufficient
for
now
right
like
if
you
brought
brought
it
down
from,
like
you
know,
over
up
from
like
a
minute
or
two
to
like
five
to
ten
seconds?
A
C
D
D
It
ends
up
being
so
it
ends
up
being
technical
debt
because
there's
a
discussion
that
no
one
has
been
ready
to
have
yet,
which
is,
if
you
have
two
disjoint
networks
where
you
can
store
IP
NS
records,
then
if
you
go
into
a
web
browser
and
you
type
an
IPS
colon,
slash,
slash
on
on
name:
where
does
it
take
you
to
which
of
the
two
disjoint
networks?
Do
you
look
for
your
information?
You
look
at
both
what
if
one
of
them
has
a
different
record
than
the
other
one
which
one
do
you
show.
D
Right
now,
pub/sub
is
not
disjoint
from
the
DHT
because
it
falls
back
on
the
DHT.
So
we
haven't
had
to
ask
that
question
as
soon
as
we
implement
either
pub/sub
only,
which
is
what
I'm
working
on
or
DNS,
which
I
think
Hugo's
working
on.
Well,
that's
a
problem.
We're
gonna
come
across
and
the
more
of
these
things
we
add
the
more
times
we
run
into
that
problem.
D
D
Ya
know
I'm
that
we
need.
We
need
this.
This
needs
to
get
done
right,
I
guess.
The
question
is
like
if,
if
rendevouz
gonna
give
us
the
same
thing,
that
dns
is
gonna
do,
but
it's
gonna
take
us
an
extra
two
quarters
to
get
there.
Do
we
need
an
extra
network
that
we
need?
We're
gonna
need
to
worry
about
longer
terms,
I,
don't
know,
maybe
we're.
Okay,
we
tell
people
it's
experimental
and
if
we
decide
to
kill
it
off
in
a
year
from
now
than
like,
that's
fine
too
right,
I,
don't
know,
that's
not
really.
F
F
Package
management
always
has
the
ability,
as
a
use
case,
to
it's
it's
a
question
that
is
in
scope
for
how
to
pursue
a
package.
Management
is
actually
how
to
distribute,
updates
and
so
like
within
the
gamut
of
options
that
we
can
choose
to
engage
with
in
package
management.
Discussions
is
not
even
distributing
updates
right,
like
we
can
just
distribute
static
indices
and
make
any
sort
of
like
updating
system
something
else.
F
We
could
make
a
definition
of
a
package
manager
that
just
distributes
the
update
route
patch
over
HTTP
and
it's
totally
centralized
and
just
walk
out
right,
whether
that's
whether
or
not
that's
a
good
idea
is
like
a
totally
different
question,
but
it's
it's
within
the
gamut
I
think
we
could
do
so.
I,
don't
know.
That's
that's
just
my
opinion,
but
I'd
say
like
to
do
the
best
you
can
in
like
this
application
will
do
what
I
can
with
what
can
be
done.
D
My
understanding
is
correct
me
if
I'm
wrong,
please
the
that
NPM
over
IP
FS.
Does
you
basically
just
go
all
the
way
up
to
the
root
levels,
like
the
only
new
IP
NS,
for
sort
of
this
global
manifest,
and
then
you
figure
it
out
using
C
IDs
from
there.
It's
like
that's,
like
a
very
low
request,
an
IPS
right,
because.
D
A
A
A
G
Reality
is
because
they
they
have
they
race
them
both
against
each
other,
the
the
DNS
one
always
wins,
because
McIntosh
Matias
and
the
two
servers
and
there's
not
that
much
traffic
on
the
Deaf
network,
it's
sort
of
small
well,
so
the
DA,
the
DNS
always
wins.
So
as
a
consequence,
there's
DHT
was
just
completely
broken
for
like
six
months
and
nobody
noticed
so
I'm
sort
of
wondering.
G
If
we
do
the
racing
thing
it
might
actually
make
sense,
did
like
do
DNS
but
build
like
a
one-second
delay
into
it
and
give
the
DHT
a
chance
to
actually
like
it.
We
can
make
the
DHT
faster
than
DNA
like
rendezvous
or
something
like
that,
like
a
truly
distributed
thing
resolved
faster
than
earth,
intentionally
hobbled
DNS.
That
might
be
the
thing
to
do.
Then.
It
gives
us
a
far
to
try
this
like.
If
we
can
make
this
the
distributed
resolution
protocols
resolve
faster
than
then
one
second,
then
they
win
system
idea.
G
But
nobody
knows
that
there's
this
T
DNS
servers
they're,
like
the
users,
like
you
mean,
speak
your
browser.
You
don't
know
that,
but
it
they've
gone
down
because
it
was
just
like
Matthias
running
them
and
they've
gone
down
for
like
eight
months
for
eight
hours
and
basically
nobody.
Nothing
was
working
on
that,
like
the
whole
system
was
down
and
nobody
really
knew
everybody
just
assumed
that
their
local
software
was
broken
but
was
actually
does
actually
for
a
decentralized
network.
G
D
Also
from
a
semi
practical
perspective,
we
so
they
both
have
like
the
sort
of
goal
of
being
able
to
work
like
offline
and
stuff,
and
also
that
when
we
give
demos,
the
Internet
is
always
terrible
and
we're
basically
offline
anyway.
So
it
is
always
a
little
bit
embarrassing
to
have
like
a
demo
like
our
stuff
works
offline
and
the
Internet's
not
working
in
as
a
result
of
demo.
It's
not
working
because
then
because
then
it's
obvious.
Something
has
gone
wrong,
so
I
think
yeah.
C
G
C
White
venice,
but
it
was.
You
will
tolerate
that
in
a
zero
order,
centralized
Network
environment,
but
you
won't
tolerate
it
when
every
other
website
isn't
raised
and
super
fast
and
the
CDN
and
optimize
so
I
see
the
battle
that
we
need
to
make
sure
that
I
could.
This
is
performin
as
much
as
possible.
We're
a
centralized
option
is
available
and
not
unreasonable
to
lean
on
and
make
sure
that
we
don't
break
things
for
offline,
that
it
does
work.