►
From YouTube: 📦Package Managers WG Weekly Sync April 16, 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
B
C
Didn't
update
my
name
it
just
his
name.
Sorry,
yes!
So
the
past
couple
of
weeks,
I've
been
off
with
flu.
I
still
have
a
sinus
infection,
which
is
hurts
here
and
here
and
it's
a
little
bit
debilitating,
but
I
can
actually
hold
a
thought
in
my
brain
for
a
reasonable
amount
of
time
now
so,
like
I
can
start
to
catch
up
again
and
yesterday
then
kind
of
went
back
through
and
looked
at
a
lot
of
different
things
that
have
been
talked
about
and
trying
to
kind
of
get
back
up
to
speed
again
this.
C
After
listening
to
the
call
last
week,
Eric
was
talking
about
like
some
of
the
different
components
of
package
managers
and
how
that
might
be
useful
to
break
apart.
Some
of
the
work
and
I
think
I
have
that
kind
of
encoded
into
this.
This
brainstorm.
It
just
doesn't
kind
of
describe
itself
well
because
you
need
a
fair
amount
of
context
to
understand
what
all
the
different
boxes
mean.
C
B
C
Than
just
one
server
running
that
thing,
because
it
does
not
handle
anything
particularly
well
right
now
other
than
the
ER
sinking,
an
initial
adding
so
I,
don't
think
the
infrastructure
team
has
enough
capacity
to
do
that
at
the
moment.
Looking
at
how
responsive
they've
been
to
other
things,
at
least
on
github,
but
it's
not
like
it's
incredibly
pressing
either
so
I
think
it
is
a
case
of
it
just
needs
more
servers
rather
than
there's
something
like
performance
problem.
That's
that's
breaking
it.
D
Be
worth
chatting
to
Hector
and
the
cluster
T
mollies
nodding.
So
I
guess
yes,
just
just
in
terms
of
if,
if
infrastructure
don't
have
the
bandwidth
to
help
you
right
now,
but
Hector
has
some
experience
of
running
multiple
nodes
that
have
a
ton
of
data
on
and
it
might
be
something
to
factor
into
the
experimentation
adding
cluster
into
the
pile
so
that,
it's
even
easier
to
say
we
are
mirroring
the
same
data
yeah.
B
Exact
same
suggestion
it
was
like
well
sounds
like
either.
We
can
run
this
on,
pin
bot
with
many
many
duplications,
or
we
can
try-
and
you
know,
use
it
more
as
a
test
case
of
like
how
do
we
do
this
is
that
everyone
else
can
can
do
it
as
well.
Not
just
people
who
have
their
own
magical.
Huge
clusters
did.
C
I've
tried
a
number
of
times
to
get
pim
bot
to
to
pin
the
thing
it
never
works,
because
it's
humongous
and
I
guess
is
recursively
trying
to
do
everything
if
you
just
give
it
a
name.
It's
gonna
timeout
before,
if
there's
definitely
some
kind
of
reasonable
timeout
on
it
rather
than
it
just
it
does
eventually
come
back.
Usually
it's
like
two
days
later
and,
like
sorry,
I
didn't
successfully.
Do
your
thing
for
you.
It's.
A
Thing,
that's
also
there's
an
open
issue
on
the
go
if
it's
Reaper
about
being
able
to
shard
pins.
So
if
you
pin
a
bag,
it
will
pin
all
the
nodes
underneath
the
dag
everywhere,
whereas
what
they
want
to
be
able
to
do
is
say
well,
I
need
like
stylist
no,
but
then
they
need
pen
like
two
levels.
Deep
and
I
miss
node
over
here
we'll
take
this
one
and
pin
Taylor
this
IP
person
over
here
will
take
this
note,
just
downloading
pin
to
those
feet,
but
they
never
never
really
landed.
That
pin.
You
know.
E
C
E
A
A
All
right
so
last
week,
I
was
trying
to
look
into
my
PFS
issues.
The
first
my
littell
is
that
the
like
copying
the
update
registry
is
command,
is
really
slow,
so
the
registry
and
what
it
does
is
it
copies
down
the
dag
node
that
is
the
first
root,
or
at
least
the
root
of
the
NPM
registry.
On
the
mirrors,
we
should
be
super
quick
because
I
won
no,
but
it
isn't
because
of
those
way
warden
thinks
it
should
and
I
tries
to
start.
A
They
starts
loading
the
whole
day,
which
is
why
so
that,
because
it's
like
so
I
changed
it
to
not
do
that,
and
now
it's
much
much
faster,
but
I
was
trying
to
I
was
trying
to
do
in
a
nice
way
and
I'm
just
so
sick
of
looking
at
the
MF
s
code.
It's
just
squirrely
squirrely
pool
streams
everywhere,
but
now
that
I
peeled
D
Garibaldi
has
landed.
The
async/await
API
can
refactor
the
whole
things
of
the
async/await.
A
So
that's
what
we've
been
doing
and
I'm
down
to
like
my
last
three
failing
tests
and
there's
so
much
better.
It's
just
so
much
easier
to
read
you
so
much!
You
just
wanders
down,
what's
going
on
so
I'm,
going
to
try
and
learn
that
in
the
next
day,
also
not
a
few
days
and
I,
probably
me
for
the
rest
of
the
week,
any
questions.
C
One
quick
little
other
update
and
I
forgot
to
write
down
eric
worre.
Pork
is
coming
over
to
London
I.
Think
he's
literally
on
a
plane
right
now
and
will
be
there
for
a
few
days,
so
I'm
going
to
go
up
on
Thursday
and
hang
out
with
him
and
maybe
that'll
end
up
colliding
with
some
table.
Flip
people
I'm
not
sure
what
part
of
London
he
needs
to
be
in,
but
I'll
just
come
up
for
the
day
and
then
brainstorm
some
things
and
see
how
that
goes.
You.
C
D
D
I
haven't
performed
the
work
yet
merely
teed
it
up
for
someone
who
might
be
interested
in
performing
the
work
in
as
much
as
agent
of
user
commented
on
the
thread
where
we
were
looking
into
NPM
my
PFS,
the
app
edition
with
a.
Why
not
examine
lower
hanging
fruit
such
as
integrating
with
existing
MPN,
transparent,
proxy
software,
so
something
that
you
could
run
for
a
team
of
developers
and
have
it
casually.
The
modules
for
you
and
Vedek
sure
seems
to
be
gaining
in
popularity
and
is
notably
used
by
some
hip
projects
in
the
JavaScript
world.
D
Did
a
nice
sort
of
team
level
effect
of
if
every
team
starts
to
have
a
cache
of
the
subset
of
NPM
that
they
use
and
that
they,
in
a
pool
of
production
instances
sharing
that
data
you
get
this
kind
of
it
plays
nicely
with
the
sizes
of
data
sets
that
okay.
This
is
good
at
right
now
and
starts
to
demonstrate
the
benefit
if
we
can
have
like
two
disparate
teams,
sharing
their
caches
and
start
to
pull
pull
modules
from
each
other's
caches
rather
than
from
central
registry.
That
would
be
a
good
demonstration.
D
C
It
seemed
like
something
we
should
reach
out
to
them
and
be
like
hey
this
one
might
be
a
good
thing,
Oh,
potentially
an
easy
way
to
have
like
a
lazy
and
I
have
cash.
In
that
you
don't
need
to
keep
everything
on
disk.
If
you
got
old
versions
that
you're
not
like
seeing
any
active
downloads,
then
you
could
almost
like
wipe
them
all
away
and
assume
that
someone
else
has
them
on
their
equivalent
thing
and
let
IVFs
take
care
of
all
of
the
the
actual
like.
Oh
well.
C
If
this
isn't
it
isn't
here,
I'll
go
load,
it
kind
of
things
also
like
I
would
group
it
as
a
slightly
separate
bit
of
work
to
MPN
on
ipfs,
like
the
tar
balls
themselves,
hopefully
have
the
same
CID,
and
so
the
same
work
gets
done
collectively
eventually,
but
trying
to
turn
the
Fleck
poke.
All
of
those
things
together
just
makes
it
a
lot
harder
than
trying
to
kind
of
let
them
do
their
own
implementation
of
ipfs
support
for
whatever
measure
that
is.
E
C
D
B
D
B
There's
something
interesting
here
in
addition
like
we
can
kind
of
face
this.
It's
like
doing
doing
like
research
and
prototyping
and
then
like
also
trying
to
talk
to
more
people
in
the
space
to
see.
If
there's
enthusiasm,
excitement,
that's
what
I
stuff
but
I
think
in
addition
to
a
lot
of
individual
machines,
kind
of
charting
part
of
a
package
manager
registry
that
they're,
using
also
being
able
to
like,
have
how
the
a
machine
that's
kind
of
doing
heavy
lifting
and
caching
is
something
that,
like
people,
might
click
on
the
idea
that
oh
there's
now.
B
Of
course,
we
have
that
like
server,
that's
like
running
that
we
use
for
our
test,
appointments
or
demos
or
whatever,
or
it's
like
just
one.
One
of
the
engineer's
particular
has
a
machine
like
I.
Think
many
workplace
setups
end
up
with
that
sort
of
model.
Where
there's
like
a
thing:
that's
not
kind
of
constantly
coming
on
and
off,
and
on
and
off,
and
on
and
off,
which
is
maybe
a
little
bit
tricky
from
the
like
pure
routing
use
case,
but
having
a
single
kind
of
local
server.
E
Right,
what
do
you
think
of
doing
like
us
like
doing
our
own
sort
of
tailored
subset
of
just
like
all
the
NPM
s
that
are
required
to
build
ipfs
and
like
everything
we
use
in
our
organization
and
the
dependencies
for
those?
So
we
could
you
know,
do
it
offline
demo,
for
example,
like
you'd,
be,
like
you
know,
unplug
the
network
and
then
everybody's
able
to
work
for
a
day
like
it
wouldn't
be
all
of
NPM.
It
would
just
be
enough
of
NPM
that,
but
I
think
it
would
still
be
a
fair
amount
of
work.
C
C
So
you
need
the
index
file
hours
as
well
as
the
the
packages
themselves,
because
NPM
needs
to
be
able
to
query
something
and
I
be
the
NPM
on
my
office
basically
will
store
the
last
time
it
queried
the
thing
onto
its
own
local
MF
s.
For
those
indexes
right.
Alex
is
not
English,
so
yeah
you
can
once
you've
installed
it
once
you
can
unplug
and
you
could
install
again
and
in
theory
you
could
change
it
to
a
version
that
you
already
had
like
an
older
version
and
it
should
be
able
to
assuming
the
old
version.
C
Doesn't
then
declare
some
dependencies
that
you
don't
have
in
your
cache,
because
things
might
be
round
quite
a
lot
that
you
like
that
already
works
without
needing
any
kind
of
extra
configuration
or
like
most
of
these
proxies
work
lazily.
So
you
just
like
plug
your
request,
your
set
of
packages
through
it,
and
then
it
will
cache
it
forever.
Like
you,
don't
need
to
try
and
manually
keep
a
list
of
those
things.
C
C
E
I
guess
my
user
story
would
be,
you
know
a
bunch
of
octopus
developers,
beaten,
Vancouver
or
somewhere.
We
go
off
to
a
cabin
in
the
woods.
With
this
note,
no
internet,
it's
like
we
want
to
be
able
to
just
take
one
hard
drive
that
has
all
the
dependencies
on
it
and
figure
it
out
when
we
get
there.
You
know
you
mean,
like.
C
E
Have
to
eat
some
code,
no
I
I
think
we
could
figure
that
out
so
I
think
because
they
could
just
they
could
just
swarm.
Connector
II,
yeah,
well,
just
custom
bootstrap
list,
yeah,
so
I
think
the
thing
is
like
you
shop
in
the
woods
and
then
there's
like
the
one
dependency.
You
need
to
build
all
the
packages
that
no,
but
that
we
forgot
to
think
before
you
left.
A
D
E
D
I
was
thinking
so
I
had
the
same
thought
when
I
was
investigating
without
chair,
and
then
it
occurs
to
me
that
the
there's
a
bit
of
work
to
do
to
integrate
with
the
Dutch
show,
which
is
worth
us
doing
or
or
encouraging
without
chair
to
do
but
I
think
there's
a
component
of
ipfs
on
NPM.
That
is
the
registry
proxy
Alex.
C
You
also,
then,
trusting
that
you're
trusting
the
NPM
on
a
PFS
registry,
like
you
put
another
party
in
the
middle
of
it.
If
you
had've
adachi,
oh
basically
go
like
I
fetch
something
from
NPM
registry
as
a
proxy
and
then
I
stored
it
both
on
disk
and
on
ipfs
I.
Only
trust
my
own,
like
I
trust
what
I
just
did
and
then,
if
I
need
to
get
that
back
later,
like
I
can
delete
my
table,
but
I
can
always
get
it
back
from
ipfs.
C
C
You
could
definitely
like
I,
just
wonder
if
there,
if,
especially
as
video,
is
aimed
at
companies
saying
like
let's
wedge
another
trusted
entity
in
the
middle
of
it
is
going
to
be
the
like
it.
It
definitely
would
could
work
that
way
like
from
a
technical
point
of
view,
but
from
a
social
point
of
view,
companies
are
already
like
struggling
to
trust
registries
as
it
is
and
will
pay
money
for
like
essays
and
things
from
enterprise
offerings.
C
All
you
are
you
talking.
Are
you
thinking
about
the
data
to
the
to
the
end
users,
client
over
dot
you
to
the
registry
over
that
YouTube
Adachi?
Oh,
like
there's
a
few
different
NPM
on
like
BFS?
Could
talk
to
video
and
potentia
could
be
set
up
to
be
like
I?
Can
I
can
Pierre
was
so
in
the
simplest
case,
I
was
thinking
just.
D
Add
an
IP
FS
storage
mechanism
to
the
Nazi
oku
and
have
them
connects
to
each
other
deliberately.
So
if
a
thousand
companies
spin
up
and
their
own
ambien
proxy,
then
each
one's
subset,
each
one's
cat,
local
cache,
becomes
a
shared
co-hosted,
communal,
cache
and
with
no
not
like
it
would
be
nice
to
integrate.
Npm
I,
listen
to
it,
but
in
the
simplest
case,
like
just
NPM
talking
to
a
registry
proxy
and
then
get
that
working
and
then
figure
out
more
elaborate,
integrations
I'm.
C
Glad
you're
on
the
same
page
in
theory,
then
NP
on
on
NPM
on
ipfs
would
be
able
to
go
like
well.
I
I've
got
the
CID
from
the
the
registry
document
that
we
ran,
but
then,
when
I
went
to
get
that's
the
ID
I
got
it
from
a
nearby
video,
oblivious
to
the
fact
that
that
happened.
So
yeah
you're
talking
about
that
bootstrap,
the
the
network
of
packaged
houses
rather
than
transport,
every
different
hop
should
be
ipfs.
A
D
A
C
B
I
wonder
what
an
IP
FS
Howard
package
manager
can
and
should
look
like
and
then
to
you
know
when,
when
we
start
talking
to
those
humans
and
like
bringing
them
into
the
community,
do
a
good
job
of
like
percolating
their
needs
and
their
asks
to
all
the
other
working
groups,
and
so
a
thought
for
the
first
one
is
that
I
think
it'd
be
really
useful.
If
we
had
like
our
our
vision
or
our
stages
of
visions,
for
what
ipfs
powered
package
maintainer
Ares
can
and
should
look
like
kind
of
written
down
somewhere
where
there's
like.
B
B
They
need
they
need,
like
the
visualization
of
what
of
what
they
can
look
forward
to
to
like
spark
their
imagination.
Maybe
they
come
up
with
something
different,
but
it'd
be
really
useful.
If
there
was
something
we
that
it
said,
these
sorts
of
things,
because
otherwise
you
have
to
explain
them
like
kind
of
one-off,
be.
This
is
why
it
would
be
so
exciting
and
what
it
would
do
and
like
here,
you
have
all
the
same
normal
like
reasonable
questions
about
how
the
heck
would
that
even
work,
you
can
put
people
somewhere
and
then
vice
versa.
C
Right,
that's
that's.
Definitely.
My
plan
for
the
next
few
weeks
is
to
turn
my
brainstorm
into
something
that's
useful
to
someone
who
has
no
contact
with
us
to
be
able
to
go
like
okay,
yes,
I,
understand
that
the
levels
of
the
parts
that
play
into
this
and
then
eventually
connect
that
to
how
people
have
done
it
already
and
can
do
it
in
the
future.