►
From YouTube: 📦Package Managers WG Weekly Sync February 12, 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
A
Okay,
so
MP
on
our
professor
has
been
very
unstable
in
the
last
week,
because
there's
a
memory
leak
I've
been
trying
to
track
it
down.
I
think
it's
in
a
refactor
of
the
right
file
atomic
module.
Ironically,
the
commit
message
was
to
say:
get
better
at
cleaning
up
file.
Descriptors
I
think
that
she
got
worse
cleaning
up
file
descriptors,
but
we're
about
to
move
off
that
library,
because-
and
that
was
an
outcome
of
some
of
the
way
the
nephron
did.
A
B
B
So
I've
started
noting
some
of
those
things
down
and
then
I'm
going
to
take
a
knife
to
that
spreadsheet
and
kind
of
change.
It
all
up
a
little
bit.
So
we
can
then
get
a
good
idea
of
which
are
the
biggest
or
like
the
the
best
package
manager,
kind
of
types
or
groups
to
aim
at
for
integration
and
then
start
filling
out.
B
Some
of
the
the
different
details
that
either
ipfs
specific
or
just
some
of
those
kind
of
like
approximate
number
of
packages
are
the
the
kind
of
protocols
that
they
have
around
publishing
and
unpublishing
and
things
that
might
be
relevant
and
maintain.
Is
that
we
either
know
or
know
of
as
the
people
to
reach
out
to
for
these
different
kind
of
things?
Just
so,
we
have
some
kind
of
like
canonical
source
of
of
different
package
measures
that
we
can
keep
track
of.
B
D
D
The
the
thing
I
wanted
to
add
to
ask
is
like
we,
you
are
now
reading
a
spreadsheet
with
the
different
characteristics
of
different
package
managers
like
registry
swatch
clients,
one
of
the
things
that
will
be
available
to
have
on
the
function
to
define
like
what
is
the
priority.
It's
like
to
go
after
each
package
manager
is
actually
also
the
relationship.
So
there
are
multiple
package
managers.
That's
already
about
communication
open
with
us.
They
already
did
some
experiments
themselves
and
or
that
are
just
like.
Overall
super
super
excited
about
of
us.
D
B
D
With
the
package
managers
ecosystem
as
a
whole
and
from
saying
hey,
no
I'm
on
average
I
get
reminders,
just
move
like
10
terabytes
data
from
machine
to
machine
a
day
or
on
average,
like
a
user,
consumes
like
like
one
gig
of
packages
a
day,
and
so
they
need
to
like
done
all
those
pretty
fast
until
we
need
to
be
able
to
extreme
it
from
multiple
locations
as
fast
as
possible
or
or
or
yeah
like
like
what.
What
is
the
discovery
process
to
finding
these
packages?
D
It's
kind
of
like
having
these
may
be,
like
kind
of,
like
all
pinch
marks
of
questions
like
easy
as
possible.
Today,
all
right
be
fast
like
if
a
package,
when
there
was
a
woman,
this
is
possible
so
that,
like
that
seeds,
the
thinking
for
the
working
groups
to
start
seeing
like
if
the
things
that
they
are
like
hacking
are
like,
especially
for
the
GFS,
and
go
at
the
pest
to
understand
like
if
they
are
putting
their
right
energy
into
the
right
directions
to
support
Isis
case
or
not.
B
Should
definitely
try
NPM
on
ipfs
see
if
it
works
for
them?
For
me,
the
servers
have
been
falling
over
every
time
over
the
past
couple
days
when
I've
tried
to
use
it,
and
it's
then
like,
hopefully,
that
any
frustration
people
will
feel
will
could
be
poured
back
into
that
Eva's
feedback
or
code
or
infrastructure
help
to
really
kick
the
tires
on
that
and
get
a
feel
for
like.
D
A
Yeah
I'm
guessing
the
any
updates,
really
gonna
be
from
me
and
Angie
today,
so
that's
cool
and
yeah.
So
we
have
these
like
this
idea
that
we
need
to
get
these
capabilities
written
down,
that
we
can
use
to
drive
the
next
quarters
worth
of
chaos
for
the
other
teams.
So
that's
gonna,
be
our
focus.
I!
Think
for
the
next
two
weeks,
pretty
such
gonna
be
out
phone.
D
D
Then
then
there
there
is
also
like
this
other.
Like
creative,
interesting
wall
posts,
it
can
be
written.
I
know
that
Alex
is
working
on
a
1i
just
describing
what
is
the
structure
that
he
developed
to
like
deploy
and
family
papers,
which
is
going
to
be
very
interesting
for
multiple
groups,
not
just
on
breakage
miners.
D
It
was
like
for
multiple
groups
that
they
have
large
other
sets,
but
it's
also
interesting
to
like
just
the
process
of
like
downloading
all
the
registries
and
doing
ipfs
ad
and
seeing
if
there
is
any
deduping
that
actually
reduces
the
data
set
right
like
if
there's
like
deduplication
happening
without
like
any
magic
extra
magic.
That
then
we
could
say
Oh
like
we
could
grab
all
the
packages
in
the
world
and
I
could
reduce
them
by
20%
or
30%
I'm,
not
sure
if
this
is
possible,
but
I
I
do
remember
when
we
were
like
this
2015.
D
We
were
at
this
conference
panel
mesh
like
basically
the
the
mesh
routing
protocols
conference,
and
it
was
in
the
middle
of
the
woods
and
like
people
were
having
trouble
like
I'm
molding,
the
open,
wrt
different
flavors
for
their
different
routers,
and
we
pulled
it
all
together
at
my
PFS
and
we
managed
to
save
40%
of
the
space
that
would
take
to
store
all
of
those
images
like
all
of
those
destroyed
images,
and
that
was
using
Robyn
fingerprinting
if
I'm
not
mistaken.
Co,
for
example,
then,
like
we
have
like
this
different
values
to
tune.
D
Oh,
are
we
using
just
fixed
size,
chunking
that
we're
using
Robyn
fingerprinting?
Are
we
creating
an
interesting
data
structure
to
store
these
packages
that,
like
will
make
even
like
six,
faster
or
just
like
the
whole
storage
faster
and
until
I
started
like
outlining
these
things
that
we
could
publish
to
get
people
to
rally
to
understand?
What
are
the
benchmarks
to
get
people
excited
about
the
possibilities
like
the
the
extra
features.
D
Tears
are
like
hey,
IQ
or
confined
in
or
dissipated,
but
I
was
like
what
these.
Actually
people
like
running
easy
for
structure
bike
administers,
can
gain
from
using
a
PSS
as
a
transfer
protocol
will
be
very
interesting
like
as
like
broadcast
material,
or
this
like
this
is
my
claim,
but
he's
like
one
thing:
I'm
excited
about
yeah
right,
yeah,.
A
It's
definitely
worth
paying
some
of
these
experiments
and
in
a
birthing
in
the
group,
there's
a
link
to
an
issue
said
Michael:
did
some
experimentation
using
rubbin
encoding
and
having
uncompressed
tables
the
idea
being
that,
like
between
releases,
not
a
lot,
would
change
so
in
theory,
you'd
be
able
to
deep
a
lot
like
that
and
it
was.
It
actually
ended
up
being
bigger
than
just
using
straight
tables
and
using
symmetric
chunks.
So
it's
definitely
worth
proving
these
things
up.
B
I
guess
on
a
similar
line,
something
that
just
popped
into
my
head
from
maybe
like
thinking
about
what
we
could
potentially
write
about
is
maybe
giving
them
a
way
to
work
out
like
how
much
an
individual
developers
laptop
is
downloading
or
storing,
say
in
node
modules
like
how
many
times
do.
I
have
requests
on
my
hard
disk,
like
across
every
different
previous
project
that
I've
stored
and
how
many
times
that
get
downloaded
like
and
I
don't
know
if
ipfs
would
necessarily
directly
change
like
the
amount
of
total
use
across
the
disk
baby.
B
Interesting
as
then
like,
oh
well,
maybe
we
should
be
similar
thing
into
packaged
item
from
the
ipfs
directory,
rather
than
like
copying
everywhere
or
other
similar
kind
of
things.
It's
like
leading
down
those
contractions
I
know,
there's
all
kinds
of
like
his
ways
that
are
terrible
and
potentially
those
things
get
better
or
potentially
they
don't
yeah.
D
I
love
it
like
perhaps
like
just
writing
a
simple
script
where
people
we
write
people
to
a.
Can.
You
run
this
on
your
like
how
projects
folder
and
like
we
will
just
like
see
the
package
essence
of
like
of
your
tiny
projects.
You
see
how
many
times
you
are
requested,
like
downloading
the
same
package
yeah
it's
the
same
thing
for
like
even
the
IP
person
p.m.
to
like
when
you
do
install.
D
If
you
would
like
get
metrics
about
how
many
times
you
save
the
round
clip
to
the
NPM
registry
by
just
like
seeing
a
backyard,
you
have
a
module
like
will
you
network
or
that
you
don't
know
the
the
module
not
from
like
even
the
registry
DOJ's,
every
test
that
I
do,
but
actually
the
move
from
like
a
local
area
network
node
will
be
interesting,
metrics,
saying
I,
hey
Ike!
Here's
all
of
these,
like
pen
with
games
which
are
very
expensive,
which
you
are
paying
Amazon
for
them.
D
Any
like
getting
into
the
habit
of
I
getting
out
a
cadence,
so
like
this
blog
post,
like
show
like
the
amount
of
benefits
that
we
have
available
and
again
I.
This
is
like
before,
like
we
even
dive
super
deep
into
data
structures
and
to
make
these
like
super
optimized
right.
This
is
just
like,
like
at
the
front
like
what
we
have
already
available,
that
what
we
can
offer.
B
A
A
D
Yeah,
so
so
again,
I
think
like
really
a
great
outfit
for
like
tissues
for
the
next
following
weeks
is
really
like.
Okay,
what
can
we
avoid
be
fast
images
if
the
team
started
thinking
about
to
make
like
to
plan
for
next
for
about
what
I
should
be
trying
to
optimize
and
also
like
what
they
should
prioritize?
The
most
to
get
done
is
quarter
so
that
they
are
better
set
for
next
quarter.
D
Then
there
is
perhaps
I
can
sew
composition
12
with
the
web
browser
team
because
they
want
to
do
this
jazz
CDN
over
IP
FS,
so
welding,
all
the
web
apps
in
the
world
through
ipfs
itself,
using
things
like
web
package,
or
even
just
like
just
by
putting
the
assets
at
ipfs
and
like
serving
it
to
a
service
worker.
So
like
kind
of
like
putting
a
jetpack
onto
the
script
tag,
if
we
can
call
that
a
package
manager
and
I
mean
I,
guess
like
people
would
be
like
eight
minutes
left
and
and
I
I,
don't
see.
D
D
A
I
concrete
and
because
I
can't
create
private
repos,
otherwise
I
would
have
done
it
already
is
basically
so
we
have
these,
were
there?
Are
these
like
Google,
Docs
and
they're
great,
but
it's
really
hard
to
have
like
nuanced
threaded
conversations
in
a
Google,
Doc
and
they're,
also
terrible
for
discoverability,
and
you
know
now
that
we've
got
like
some
stuff
to
do
it
be
cool
to,
like
you
know,
have
a
waffle
board
and
wood
out.
So
you
know,
what's
going
on
kinda
hard
to
do
that
with
Google
Docs
yeah.
D
Oh,
we
could
totally
like
replace
some
piece
of
infrastructure
they
offer
PFS
can
ever
direct
impacting
them,
and
so
another
can
bike
that
can
be
like
very
productive
because
they
can
use
it
as
email
all-seeing
eye
defense
group
is
kind
of
like
contributing
to
the
whole
ecosystem.
That's
great
or
oh,
my
god,
there
is
a
potential
like
alternative
to
what
we
have
been
building
for
many
years.
We
should
be
careful
and
so
like
when
we
go
full
public.
D
We
have
to
be
very
careful
to
always
talk
about,
like
the
benefits
and
like
was
it
contribute
to
the
ecosystems
and
I
like
what
are
the
feature
sets
rather
than
like
strategic,
like
personal
relationships
that
we
have
with
people,
and
we
have
to
be
very
sure
to
make
that
distinction.
Otherwise
it
will
get
very
very
fast
like
what
do
you
guys
think
I
think.
A
D
D
B
So
I
was
poking
around
a
little
bit
to
see
if
I
could
turn
Alex's
code
into
something
that
would
work
for
something
else.
So
the
idea
of,
like
you,
run
a
local
copy
of
a
register
of
like
a
registries,
kind
of
proxy
I
guess
where
the
proxy
might
be
like.
Oh
I,
already
got
this
or
I
will
go
to
the
wider
ipfs
swarm
I.
B
Guess,
that's
pretty
cool
I'm,
still
learning
all
the
terminology,
and
if
I
can't
find
it
there,
then
I
will
go
now
actually
downloaded
from
the
registry
and
store
that
in
ipfs
and
then
use
it,
which
is,
should
work
for
a
lot
of
registry.
A
lot
of
package
managers
that
have
the
ability
to
say,
like
oh
use,
an
alternative
registry
that
I
wonder
if
you
could
simply
do
it
as
like
the
URL
of
the
tar
ball
as.
A
B
Like
the
key,
rather
than
the
package
name
and
then
like
the
so
I
think
the
name
of
the
registry
or
that
like
how
it's
mounted
on
ipfs
for
NPM
is
NPM
registry.
Could
you
just
take
the
URL
and
be
like
there?
You
go.
The
only
thing
I
can
see
happening.
There
is,
if
you
would
just
try
and
pipe
every
URL,
that
a
package
manager
client
asked
for
so
ipfs
stuff.
That
would
expire
like
whenever
there
was
a
new
version.
B
So
maybe
it's
a
case
of
white
listing
a
certain
set
of
regular
expressions
to
be
like
these
are
known
links
to
tar
balls
for
registry.
So
if
it
ends
in
gem,
then
for
ruby
gems.
That's
absolutely
fine,
like
we'll
assume
that
it
is
something
that
can
be
stored,
but
to
it
in
it
like
a
way
that
doesn't
care
what
package
manager
is
using
as
a
as
a
starting
point,
rather
than
to
start
to
fill
some
of
the
because
I
figure.
The
content
is
the
same.
Once
it's
on
ipfs.
C
B
Even
if
the
index
that
pointed
to
it
was
just
dumb
like
this
URL
says
the
content
of
that
URL
is
over
there,
then,
as
we
work
out,
we
particular
communities
the
best
way
to
map
their
particular
register
is
metadata
to
point
to
those
see
IDs.
The
content
is
already
there.
It's
just
that
there
are
multiple
different
ways
to
point
to
that
CID
and
then
a
lot
of
that
stuff
is
already
kind
of
available
and
as
registry
clients
get
either
like
plugins
or
there
are
Forks
or
rappers
that
actually
add
direct
IP
ever
support.
B
You
also
kind
of
then
can
say
like
oh
well,
we
can
you.
Multiple
people
can
run
these
online
in
a
kind
of
artifactory
style
like
it's.
It's
not
directly
local,
like
to
your
laptop,
but
it's
running
all
the
time
for
your
team,
or
whatever
I
mean
you
could
you
could
probably
put
them
on
Heroku,
it's
just
that
the
poor
little
Dino
would
be
like
I,
never
gonna
store
anything
for
very
long
like
a
little
digital
ocean
droplet.
B
It
would
be
interesting
to
see
if
how
quickly,
rather
than
just
looking
at
individual
users
having
their
own
ipfs
nodes,
could
you
have
lots
of
little
like
partial
mirrors
of
just
the
packages
that
say,
one
company
or
one
group
of
users
cares
about
where
the
code
is
basically
the
same
ever
for
everyone,
it's
just
that
not
everyone
has
all
of
the
packages
and
they're.
Basically,
then
keeping
connected
oh
I'm
running
over.
D
Yeah,
we
need
to
jump
the
other
car
but
I'm
trying
to
answer
that
pretty
quickly.
I,
don't
think
I'm.
A
fast
right
now
supports
that
option
to
em.
Fs
is
just
this
virtual
directory
tree
that
we
support
inside
of
EFS,
so
that
gives
you
actually
stuck
at
normal,
unique
suppose
uniques
like
filesystem,
but
like.
D
You
can
ever
you
can
I
fetch
the
whole
tree
just
like
that.
You
have
a
way
to
observe
what
is
going
on,
but
then
you
just
pin
your
cue
just
like
fetch
soup
soup
parts
of
their
that
graph
like
which
are
the
packages
that
you
are
interested
rather
than
like
having
to
copy
the
whole
thing,
but
I,
don't
think
that's
there
at
all
or
spikes
outs,
maybe,
but
like
Alex
it
for
him,
because
he
has
this.
B
A
B
Earlier
I
ran
into
there
like
pops
up
could
be
used
to
like
avoid
having
to
copy
the
whole
list
of
the
registry
down.
Even
if
you
don't
want
it
all.