►
From YouTube: ⒿⓈ⚡️ js-ipfs Core Dev Team Weekly 2018-05-21
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
Yesterday,
yesterday
week,
I
spent
a
lot
of
my
time
fixing
up
tests,
trying
to
figure
out
where
the
flakiness
is
and
fix
that
found
some
kind
of
bugs
that
were
really
hard
to
find.
But
when
I
found
them,
it
became
quite
obvious
that
why
why
there
were
two
flaky
tests,
things
like
callbacks
not
being
called
back
really
hard
to
track
down
and
things
like
intervals
being
like
firing
callbacks
multiple
times,
because
the
interval
wasn't
cleared.
A
So
that's
that's
all
good
fun,
so
I'm
still
working
on
getting
the
0:29
release
out
the
door,
I
was
working
on
those
tests
fix
the
maps.
What
else
I
did
do
a
couple
of
reviews
for
so
I
reviewed.
Something
first
course
on
the
which
is
just
documenting.
Api
path
was
originally
to
fix.
Up.
Someone's
someone's
had
opened
an
issue
about
they.
B
So
I
spent
all
of
last
week,
looking
at
the
note
10
issues
that
is
I'm
running
some
final
tests
today,
with
all
of
the
fix
Alan
that
you
added
it's
looking
good,
the
biggest
pain
point
is
ipfs,
being
a
dependency
of
all
of
its
or
a
lot
of
its
dependence.
Making
testing
difficult,
so
I've
got
that
going
now.
A
A
C
E
Last
week
was
mostly
through
NPR's
and
beginning
to
some
issues,
with
diversity,
CTL
evaluating
some
of
the
nap
implementations
available
for
now
the
whole
bunch
and
specifically,
there's
a
lot
honestly
and
starting
the
implementation
of
the
live
to
be
map.
We
saw
one
of
those
in
available
implementations
work
and
please
extend
in
one
of
those
for
TCP.
They
all
support
DDP,
mostly
no
tcp,
no
punching
and
doing
some
performances
on
the
public
stuff.
There's
a
PR.
If
you
guys
have
time,
please
take
a
look
leave
your
comments.
A
Cool,
thank
you.
So
next
we
have
Vasco,
who
is
not
here.
He's
told
me
that
he
is
at
the
Lisbon
Hat
week
and
he's
in
a
meeting
of
sorts,
so
I'm,
just
gonna
read
out
what
he's
put
here
so
service
worker,
a
gateway.
Okay,
our
he's
working
on
the
cobbled
service
worker
gateway
repo
into
several
verticals
display
files
of
several
types
on
service
working,
getting
away.
Okay
and
then
he's
working
on
issues
always
done
sorry
issue:
five,
eight
one
I
profess
behind
a
proxy.
Yes,
that's!
A
What
else
they
blocked
not
blocked
on
anything?
He
is
at
the
Lisbon
hack,
wake
he's
working
on
the
service
where
he
gateway
okay,
and
he
wants
to
divide
into
modules
several
modules,
refactor
and
half
those
much
was
released.
Her
a
releasing
have
first
version
deployed
to
Jess
that
ipfs
cool,
okay,
rad.
A
Cool
so
done
finding
a
way
to
prove
how
we
handle
errors
in
Jess
IVFs.
Yes,
I'd
also
like
to
have
a
look
at
that
too.
So
I
will
endeavour
to
have
a
look
at
that
issue.
This
week
working
progress,
the
pr2,
a
remove
boot
from
constructor
and
stop
using
event
emitters.
Okay,
that's
work
in
progress
blocked,
it's
not
applicable.
A
F
I
have
been
working
on
the
lipid
appear
and
the
Roger
I
fixed
some
things
about
the
RPC,
because
it
was
only
registering
to
the
only
peer
to
first
peer.
It
had
discovered
that
had
to
run
was
for
the
code
enables,
but
victor
is
also
working
on
that.
So
he
told
me
that
he
is
going
to
work
that
and
yeah
now.
I
probably
will
just
wait
until
he's
finished
with
my
before
doing
something
about
my
PR
and
I.
Do
not
have
anything
planned
for
next
week,
so
I'm
open
for
suggestions.
A
D
Hello,
I
am
also
the
Lisbon
hack
week
is
going
on
behind
me.
I
didn't
see
that
earlier
last
week,
I
was
working
on
em
FS
still
so
I've
got
it
all
integrated
with
J,
so
FS
was
quite
cool,
started
running
the
interface
tests
against
it
and
most
of
it
exploded
because
it
turns
out
there's
a
lot
of
discrepancies
between
go
and
jeaious
when
it
comes
to
first
implementation.
The
most
fun
one
I
found
was
that
if
you
ipfs
add
an
empty
file,
you
get
different
hashes
for
going
for
jeaious,
which
was
a
lot
of
fun.
D
So
I
submitted
a
PR
that
should
fix
that
and
I'm
kind
of
just
going
through
and
making
sure
that
these
things
in
it
the
same
hashes,
because
although
the
tests
depend
on
the
hashes
and
it
sends
out
a
lot
of
tests
for
other
interesting
stuff
like
getting
back
empty
strings
from
go,
I
think
as
a
side
effect
of
having
a
type
system
and
another
kind
of
boring
stuff,
yeah.
So
I'm
going
to
be
that
I
think
just
gonna
be
like
a
long-term
ongoing
project,
just
making
sure
it
was
in
timeline.
D
Yeah
whoo,
so
I
am
also
going
through
the
ducts
now
for
the
files
API,
because
we've
got
a
lot
of
crazy
stuff
like
passing
the
Rays
in
as
arguments
like
a
raise,
the
file
I
feel
like
source
and
destination
pars,
and
instead
of
like
two
discrete
arguments
and
and
that
kind
of
starts
at
you're,
trying
to
make
the
API
a
little
bit
better.
You
know
get
a
better
developer
experience
out
of
it
and
yeah.
Hopefully,.
A
G
Question
yeah
for
Alex
I
would
love
to
talk
with
you
after
this
meeting,
maybe
in
slack
or
something
or
I
can
hop
on
IRC
about
how
you're
testing
those
things
I'm
running
into
a
lot
of
similar
things,
just
not
even
just
even
last
week,
finally
figuring
out
really
how
to
test
and
go
and
then
in
Jas
that
doesn't
seem
to
be
written
down
anywhere.
So,
if
I,
if
we
could
talk
and
maybe
get.
D
G
G
A
G
Well,
in
my
trying
to
get
the
bits
law
stuff
for
JavaScript
up
to
the
ghost
spec
I
discovered
a
couple
minor
things
in
the
testing
of
I,
think
it
was
IPS
CTL
submitted.
Poor
requests
for
those
I've
got
some
really
good
feedback.
Those
are
either
awaiting
review
or
awaiting
release
and
others
outside
of
that.
I'm
I've
got
my
three
up
to
two
or
three
different
feature
additions
into
the
ASI
PFS
for,
but
swap
stuff
I'm
working
on
a
ledger
for
peer
which
have
a
pull
request
into
bits.
G
Well,
for
that,
and
at
this
point
seeing
what
I'm
seeing
about
all
the
tests
and
stuff
I've
got
an
open
question
in
a
comet
in
core
about
how
to
approach
this
between
juggling
the
three
three
repos
kind
of
developing
a
script
and
once
we're
a
little
happy
with
that,
I'm
just
gonna
take
all
of
these
and
make
them
one
because
they're
not
there's
not
a
lot
going
on
there.
It's
really
just
adding
a
surface
area
to
JSI
PFS
and
that
way,
I
don't
have
to
juggle
all
these
PR
as
well.
A
Okay,
cool!
Thank
you,
sir.
Next
we
have
someone
called
named
dragon
questions
cool
okay.
So
what
have
we
got
so
Jenkins
started
using
yarn
over
NPM
in
November
2017.
Is
there
a
reason
for
the
switch
I
couldn't
find
a
clear
one
as
we
are
not
using
packaged
locks
and
several
UJS
ipfs
dependencies
also
use
it
as
a
dev
dependency?
It's
causing
a
lot
of
issues,
locking
down
versions
which
is
aggravated
by
using
different
package
management.
A
E
Beep
thankful
runner,
that's
another
matter
yeah,
so
so
yeah.
That
was
the
reason
I
think
when
yard
had
package
looking
and
when
we
started
implementing
the
Jenkins
stuff,
Victor
would
be
able
to
to
tell
everyone
about
this.
This
here
was
one
that
did
work,
but
this
is
what
I
understand
from
my
conversations
with
him
and
I
had
with
MV
I'm
getting
even
half
package
looking,
but
then
it
was
looted,
but
that
was
the
main
reason
source
understands.
E
E
Sorry,
can
you
guys
give
me
steel,
yeah,
okay,
so
yeah,
so
yarn
was
used
there
use
because
of
the
parallelism,
because
we
were,
we
don't
have
the
Jenkins
runners
isolated.
They
run
on
the
same
box
that
I
can
potentially
be
running
congrats
it
several
runners
that
it
that's
it
so
so
yeah
yarn
has
better
parallelism.
It
doesn't
mess
up
when
you
run
to
yarn
insults
at
the
same
time
as
Austin
smpm
does
I,
guess
it.
E
It
might
have
improve
now,
and
we
also
have
NPM
package
slots
now,
which
I
don't
think
we
had
at
the
time
we
started
looking
at
this,
so
it
might
be,
might
be
something
to
look
into,
and
maybe
we
can't
stand
our
eyes
on
one,
because
I
think
most
people
are
using
NPM,
perhaps
and
I
also
look
into
the
package,
locking
features
that
NPM
Oh
first
right
now
so,
but
that's
what
I'm?
Ok.
E
E
With
surgeon,
the
I'm
missing
the
word:
what's
the
word
for
the
version
spike
anyways,
the
we've
been
running
into
some
issues
and
the
reason
we
have
not
yet
stay
on
the
rice
and
using
package
long
since,
because
we
have
different
locking
mechanisms
for
different
package
manager
managers,
so
they're
not
compatible.
So
if
people
are
using
yarn
they
we
can
just
assume
that
they
will
be
picking
up
the
package
slug
for
MDM
and
vice
versa.
So
that
was
a
good
point
that
he
brought
up.
E
B
E
B
E
A
Yes,
I've
had
some
experience
with
setting
them
to
this
to
a
specific
version,
and,
whilst
it
kind
of
works
on
that,
it's
only
on
that
level
that
it
will
happen
because
dependencies
of
the
thing
you're
depending
on
will
still
have.
You
know
that
the
version
ranges
and
so
you'll
still
get
dependencies
that
are
deeper
in
the
tree.
Then
your
kind
of
direct
dependencies
that
that
have
different
versions
and
they'll
keep
changing.
So,
in
my
experience
you
you
kind
of.
D
Right,
I
also
think
we've
got
all
these
or
so
many
little
modules,
and
then
we
have
to
you
know,
release
like
we
have
to
update
the
package
for
facing
in
all
of
them
when
they
wanna
bomb.
What
the
libraries
changes.
This
version
as
well
just
comes
a
massive
over
there
like
I
think
we
should
go
for
our
stuff.
At
least
we
should
just
release
version.
One
of
everything
and
use
carries.
E
E
If
we
can
do
something
about
it,
like
maybe
Linda
or
there's
a
wave
setting
those
conventions
on
to
package
package
manager
level
and
avoiding
it
changing
that
that's
one
solution,
another
one
is
like
Alan
suggested,
which
is
just
one
thing
versions,
one
I'm
going
from
there
and
not
using
the
zero
versions
of
the
history.
Yes,.
D
All
right,
David,
yes,
why
we
don't
have
version
one
of
anything
and
he
said
there
basically
is
because
we
weren't
feature
complete.
It
sounds
like
some
arbitrary
line
in
the
sand
that
we're
never
gonna
cross,
because
the
go
team
is
so
much
far
ahead
of
the
Juris
team.
They're
gonna
still
be
adding
features
and
there's
a
there's
a
web
page
somewhere.
That
has
a
list
of
all
the
outstanding
things
that
haven't
been
implemented
yet
and
now
this
is
just
going
to
grow
forever
and
we're
probably
never
gonna.
D
A
C
E
D
A
Does
that
sound,
okay,
cool?
Okay?
Did
we
have
any
other
questions?
Let
me
have
a
quick
click
using
you
on
that
yeah
yeah,
yeah
transfer
using
package
locks
yeah,
not
using
package,
oh
okay,
and
that's
it
for
the
questions.
Unless
there
are
questions
that
are
currently
coming
in
any
any
other
agenda
items
by
for
anyone
else
before
we
end
the
cool.