►
A
A
A
The
first
thing
is
upcoming
and
ship
releases,
so
I
can
talk
about
js,
ipfs
53
shipped
at
the
end
of
last
week,
which
is
super
exciting.
It
has
the
experimental
grpc
web
server
in
it.
Web
server
web
rpc
up
there
grpc
web
over
web
sockets
server,
which
means
that
we
have
now
true
full
duplex,
bi-directional
streaming
for
rgfs
ad,
which
is
exciting
and
we'll
roll
that
out
to
the
other
streaming
methods.
As
soon
as
we
can
do
check
the
ipfs
blog
for
the
skinny
on
that,
though,
the
post
hasn't
been
published.
A
B
Yeah
so
excuse
me.
Last
week
we
had
sort
of
like
a
pull
request:
merge
fest
on
the
go
ipfs
site
in
preparation
of
in
preparation
for
0.8
or
release
candidates
too,
and
a
bunch
of
pull
requests
related
to
pinning
services
a
bunch
of
play,
requests
related
to
remote
command,
pin
remote
commands
landed.
B
B
A
C
Yeah
so
last
week
finished
the
pr
with
the
benchmark
setup
and
also
collected
a
bunch
of
metrics,
with
basically
different
network
scenarios
and
scales.
The
results
seem
reasonable
and,
depending
on
the
scale
of
the
network,
we
will
need
to
deploy
probably
a
cluster
of
renewable
servers,
as
we
already
anticipated.
C
So
I
also
created
the
pr
with
an
example
of
using
a
rendezvous
in
jslp2p
and
now,
while
I
will
wait
on
reviews
for
that,
I
will
start
working
on
the
connection
manager.
Overall,
I
need
to
double
check
if
the
previous
plan
proposed
this
still
makes
sense
and
start
working
on
that
after
yeah
and
probably
I
will
be
spending
some
time
also
working
on
changing
the
jsl
p2p
testing
setup,
because
I
do
it
all
and
also
because
we
currently
can't
run
it
on
note
15
because
of
the
way
we
have.
C
A
C
In
the
event
emitter
I
was
talking,
we
do
go
and
think
what
we
could
do
in
the
meantime,
while
that
gets
eventually
fixed
upstream,
and
so
we
basically
now
have
the
event
emitter
types
locally
in
gsl
b2p
and
we
already
released
it
so
that
the
build
is
not
broken
anymore
and
eventually
we
will
probably
think
about
using
a
metering
instead
of
event
emitter,
because
this
will
also
be
a
problem
with
webpack
five.
So
yeah,
probably
we
will
look
in
moving
to
that.
D
E
D
A
G
F
G
Continuing
to
to
do
some
work
on
on
that
it
looks
like
we
can
have,
we
can
delete
nodes
and
without
causing
any
catastrophic
damage
so
performance.
I
didn't.
We
don't
have
a
lot
any
any
new
development
there
in
terms
of
anything,
that's
going
to
be
put
into
anything
any
kind
of
release
or
pr
anytime
soon.
G
G
The
first
one
that
needs
to
be
completed
is
the
review
of
the
distributions
pr,
because
everything
else
depends
on
the
distributions
being
set
up
in
a
particular
manner
that
that
those
are
and
then
everything
else
will,
if
assuming
that
goes
through,
everything
else
will
be
ready
to
merge
at
this
point
pending
any
any
issues
that
come
up
and
reviewing
all
that
code,
because
they've
been
completely
overhauled.
So
those
are
all
ready.
It's
just
a
matter
of
somebody
having
the
time
to
review
them,
because
they're
a
little
bit
lower
priority
right
now,.
E
Yeah,
so
the
background
story
here
is
threebox
had
a
issue
with
styling
pops
up
subscriptions
in
js
ipfs
http
client,
which
was
due
to
node
fetch
bugs
that
I
submitted
there's
a
patch
now
that
fixes
it,
but
that
also
led
to
the
discussion.
Maybe
api
should
slightly
be
revamped,
so
said,
errors
are
harder
to
miss
or
connections.
Closing
are
harder
to
miss
that
I
think
was
approved
for
implementation,
so
I
had
been
working
on
that
which
quickly
during
implementation.
I
realized
that
I'm
basically
reinventing
web
readable
streams.
E
Api
settings
already
there,
including
browsers
to
a
larger
degree,
so
I
pivoted
slightly
to
polyfilling,
readable
streams,
api
e-node
and
using
native
ones
in
browsers,
also
using
extension
of
native
ones,
because
native
ones
in
browsers,
don't
fully
implement
the
stack.
Yet
specifically
s
iteration
is
missing
and
few
other
apis.
A
E
D
Yeah,
so
this
also
came
from
feedback
from
three
box.
They
made
an
issue.
They
were
talking
about.
The
http
client,
that
is
the
memory,
continues
growing
and
shuts
down
the
node,
and
they
need
to
restart
and
all
that.
So
it's
it's,
not
the
http
server,
it's
probably
the
preload
or
bitswap
or
both.
So
I'm
currently
doing
more
research,
I'm
basically
setting
up
a
testbed
where
I
can
better
control
and
trigger
a
bit
swap
between
two
notes.
D
B
What
I
mean
is
that
this
memory
leak
do
we
actually
know
if
it's
like
legitimate,
it's
possible
that
it's
just
they
have
so
many
cued
requests.
It
looks
like
memory
leak
because
we
have
those
limits
on
pre-order
calls
and
on
the
delegated
routing
calls,
and
I
think
I
think
three
box
was
experimenting
with
both,
so
they
may
have
a
bunch
of
like
cute
things
in
memory.
D
D
A
B
B
It
will
be
native
support
for
our
your
eyes,
which
means
you
just
entered
them.
You
don't
need
to
enable
anything.
The
first
time
you
use
the
uri
you
will
have,
you
will
be
able
to
choose.
Do
you
want
to
run
your
own
node
and
load
content
from
ipfs
that
way
or
delegate
resolution
of
ipfs
addresses
to
some
trusted
public
gateway
and
you
can
customize
the
gateway
you're
using.
So
that's
the.