►
Description
Meeting description here: https://github.com/ipfs/team-mgmt/issues/992#issuecomment-654271558
For more information on IPFS
- visit the project website: https://ipfs.io
- or follow IPFS on Twitter: https://twitter.com/IPFS
Sign up to get IPFS news, including releases, ecosystem updates, and community announcements in your inbox, each Tuesday: http://eepurl.com/gL2Pi5
A
Hello:
everyone
welcome
to
the
ipfs
core
implementations
weekly
sync
for
Monday,
the
6th
of
July
2020
I
am
making
brain
I'll,
be
your
host.
We
are
going
to
go
through
our
high
priority
initiatives
and
other
initiatives
and
then
a
Q&A
and
parking
law
etc.
It's
gonna
be
amazing.
We're
gonna
have
a
lovely
time
so
to
start
off
upcoming
and
ship
releases.
A
Ship
anything,
okay,
we're
about
to
ship
Jase
ipfs
there.
The
next
update
has
the
migration
to
multi
hash,
storing
blocked
by
multi
hash.
So
what
this
is
is
changing
the
key
format
for
the
data
stores,
fundamentally
so
that
everything
is
stored
by
the
multinational
block
rather
than
the
floor
CID.
So
a
fewer
lookups
in
centric
said
sure
it's
gonna
be
awesome.
It's
also
going
to
be
significant
because
it's
the
first
time
we've
really
done
a
database
migration.
A
B
C
E
All
right
yep
so
for
the
API
spec
on
the
repo,
some
new
issues
opened
and
I
started.
I
did
a
little
bit
of
triaging
and
created
the
milestone
for
the
MVP
that
we
want
that's
required
for
web
UI
integration,
and
that
includes
that
soft
scope.
We
agreed
on
for
go
and
JSI
DFS
to
implement
client
side
of
that
API.
So,
at
this
point,
I
think
it
would
be
useful,
I've
added
a
section
to
the
design
review
to
make
it
make
it
like
a
call
this
week
and
gather
some.
E
Feedback
from
Goa
like
if
a
suggest
activist
teams
basically
go
over
API
spec
and
see
if
it
fits
our
like
minimum
assumptions
before
we
sort
of
like
plan
to
do
that
on
in
what
UI?
Now
there's
like
the
execution
of
IP,
I
Scrolls
move
to
the
coop
confess
and
Jessa
profess,
and
that
means
no
like
it
will
be
useful
if
teams
from
JCP
Fesenko
IDF
has
provided
feedback
for
we
finalize
this
spec
sorry
I'm
running
out
of
coffee.
E
A
E
So
yeah
the
low-hanging
like
two
low-hanging
fruits
to
resolve.
While
you
are
reading
this
spec
one
is
access
control
the
there
is,
there
are
two
choices,
one
is:
do
we
want
a
simple
API
key,
or
do
we
want
to
do
something
more
flexible
or
useful,
directly
proposed
an
interesting
scheme
where
we
reuse
PKI,
which
is
already
present
on
each
node?
Well,
you
use
pide
key
for
effectively
signing
remote,
pinning
requests
and
then,
instead
of
having
a
vendor
specific
keys,
you
would
simply
in
add
the
pinning
service
you
would
like
whitelist
final.
A
D
E
Exactly
so
so
that's
a
coupon
part
of
of
it
and
the
second
one
is,
do
want
to
operate
on
the
level
of
actual
genes
that
we
want
to
ping.
Or
do
we
want
to
apply
feedback
or
ideas
from
the
pinning
summit
where
we
had
this
idea
that
we
would
have
a
pin
object,
representation
and
the
actual
CI
descends
to
the
pinning
service
would
not
be
the
pin
we
want
to
pin,
but
the
object
describing
the
pin
it
see
id
we
want
to
pin
with
additional
metadata
and
I
feel
that's
a
good
starting
point.
E
C
No
I'll
take
a
look
at
that
today.
I
use
block
I
had
to
do
some
heckuva
stuff,
but
I
will
review
those
today
and
then
yeah
I
think
we
should
just
set
up
sometime
this
week
for
the
three
or
four
of
us
to
just
get
together
and
and
push
that
forward.
C
Yeah
so
from
the
J
sort
of
things,
there
is
an
issue
right
now
with
how
J
s
is
importing
and
exporting
keys
for
IP
and
s,
and
that
is
causing
problems
with
Interop
with
go
for
switching
over
to
IDI
to
five
five
one:
nine
keys
by
default.
So
I'm
going
to
look
at
fixing
that
today,
our
this
week
for
Liberty,
B,
keychain
and
Liberty
to
be
crypto
so
that
we
can
get
that
deployed
out
because
the
EFS
work
will
be
kind
of
blocked
on
that.
D
D
Do
we
need
to
canonical
like
do
need
to
canonicalize
things,
so
people
go
to
the
same
website,
whether
they
put
it
in
is
whether
they
enter
in
the
sub
domain
is
base
32
or
base
36
or
like
how
we
want
some
of
the
shenanigans
around
that
we
need
to
figure
out
the
otherwise
the
PR
to
plan
the
elliptic
curve
keys
by
default
is
pretty
good,
including
the
key
rotation.
However,
there
are
some
IPTV
issues
that
we're
causing
shortness
test
failures
and
IPTV
doesn't
really
want
to
upgrade
right
now.
E
B
Yes,
so,
regarding
Dorinda
for
last
week,
I
had
some
time
to
start
the
second
iteration,
based
on
initial
feedback
that
Jacob
provided
for
getting
to
a
next
state
of
review.
I
basically
need
to
work
on
four
new
things
which,
basically,
since
basically
I'm
starting
for
creating
the
creating
artifact
ring,
in
fact,
literally
writing
a
better
word
within
the
people.
But
the
end
goal
is
after
we
have
the
complete
code
just
to
switch
it
up
and
put
it
in
Jessup.
B
The
garbage
collecting
in
the
cookies
and
also
integrate
the
sign
pair
records
and
that's
kind
of
the
current
status
that
I
am
and
I
went
to
continue
working
on
this
this
week
and
also
regarding
the
next
point,
the
sign
here
records
in
gossip
7.1,
the
sign
clear
records.
Work
is
ready
for
interview,
so
I
think
it
should
be
not
far
away
from
getting
ready
to
be
merged
and
in
the
gossips
of
wonder,
twin
stuff.
B
I've
been
doing
some
progress,
which
came
in
basically
but
last
week
called
spam
protection
PRS
merged,
and
we
are
now
working
on
getting
the
direct
peer
connections,
part
of
the
spectrum,
and
then
we
also
will
only
need
the
peer
exchange,
which
is
spending
on
the
sign,
clear
records.
But
we
can
work
on
branches
for
now
and
when
we
have
that
we
basically
need
12,
a
big
effort
of
our
own
testing
into
end
to
I
profess
and
interrupt
and
all
that
stuff,
but
yeah
that
will
be
maybe
for
the
next
week
and
yeah.
A
A
A
So
a
few
outstanding
questions
on
there
ji
PFS
call
request.
We
surround
things
like
like:
do
we
need
to
define
direct
and
recursive?
If
we
can
just
define
depth
that
kind
of
thing
we
can
remove
some
redundancy
from
from
how
we
store
pins,
which
would
be
quite
nice
and
the
addition
of
metadata
the
news,
con
heavily
influenced
by
clusters,
use
case
and
then
just
the
ability
to
add
arbitrary
metadata
to
a
pin
the
answer.
That's
not
something
to
be
doing
this
week!
A
F
F
I'll
just
go
ahead
yeah,
so
I
was
waiting
mostly
on
reviews.
Last
week,
I
got
some
so
I'll
be
trying
to
address
those
this
week,
not
much
that
had
happened
on
that
I've
been
working
on
improving
web
file.
Add
sync
I
also
try
to
schedule
a
design
review
meeting
clusters,
but
nobody
really
signed
up
for
that.
So
maybe
you
could
I
partially.
It's
probably
because
I
didn't
really
explain
what
it
was
about.
F
One
is
would
be
to
implement
our
own
form
that
as
things
that
could
handle
that,
but
also
avoid
buffering
but
better
yet
would
be
to
not
use
that
and
actually
be
able
to
use
form
data
which
in
that
case
would
have
to
change
HTTP
API
styling
by
in
a
backwards
compatible
way.
So
the
meeting
for
the
design
review
is
attempt
to
kind
of
discuss
which
path
we
can
go
with
so
I
think
somebody
from
go
team
would
be
also
helpful
to
have
on
that
call,
so
that
was
kind
of
topic
or
on
topic
or
topic.
F
I
did
some
implementation
on
the
web
UI.
That
proves
the
concept,
but
it
doesn't
use
this
JSI
PFS
h-2b
client
right
now.
It
just
uses
the
HTTP
itself,
but
you
prove
that
it's
possible.
So
some
time
I
spent
trying
to
figure
out
results
are
still
not
great
I
think
it's
went
from
42
seconds
to
14
seconds
on
to
gig
updates
uploads
on
a
local
host,
which
is
much
more
than
if
you
compare
to
curl,
which
does
it
in
three
seconds.
I
think
it
kind
of
been
a
rabbit
hole.
F
Straight
away,
I
think
I
measured.
It
was
not
that
much
different
from
curl
III.
It's
in
a
pull
request.
I,
remember
the
number
so
yeah,
but
it
was
not
reason
much
different
from
adding
this
curl,
but
I
think
the
difference.
Also,
you
can
see
difference
much
more
depending
on
the
file
size
on
a
smaller
sizes.
You
don't
really
see
that
much
difference,
but,
as
you
increase
the
file
size
difference
becomes
more
clear.
A
B
A
A
Questions
we
have
a
question
from
Ruben.
Is
there
really
a
use
case
for
dense
two
or
three
I
mean?
What's
the
purpose,
then
an
answer
I
think
we
need
more
of
a
thing
like
shot
at
little
X
from
top
or
bottom.
This
would
allow
to
store
large
files
on
the
cluster
and
have
each
node
just
store
a
stripe.
A
Have
a
look
at
scattering
sharding
here:
okay,
my
understanding
of
the
depth
was
that
you
want
to
say,
like
here's,
a
dag
I
take
a
bit
of
that
dag
and
then
just
replicate
a
section
of
it
rather
than
replicate
the
other
thing.
So
if
you
pin
a
CID,
that's
not
the
root
of
this
bit
and
then
say
replicate
like
three
levels:
deep
from
there.
That
was
my
understanding.
A
D
I
had
one
which
is
I,
don't
know
if
this
is
gonna
be
useful
at
all
for
the
pinning
API
stuff,
but
one
of
the
things
that
was
sort
of
nice
about
the
really
slow
store
all
of
your
pins
in
a
tag
instead
of
a
database
is
that
you
can
easily
transport
them.
Is
that
something
we're
still
interested
in
using?
And
if
so,
then
do
we
want
a
way
to
be
able
to
like
recreate
a
dag
thing
in
J
s,
given
that
we're
no
longer
using
it
as
the
base
representation
cuz,
it's
slow.
A
There's
a
comment
from
Stephen
somewhere
that
just
said
that
that
was
always
the
idea
that
you
could
you
could.
You
know,
transfer
this
bag
of
pins
around
and
then
no
one
actually
did
it.
So
as
far
as
I
know,
there
was
no
actual
need
for
it,
but
then
listening
to
Lloyd
will
talk
about
sending
a
CID.
Then
there's
a
tree
of
things.
Maybe
there
is
but
I
mean.
F
A
F
F
E
Was
always
sort
of
always
like
the
ideal
scope?
If
we
had
unlimited
time
and
the
resources
I,
don't
think
will
be
it's
well,
it's
not
realistic
to
think
that
we'll
land
thread
like
thread
based
or
thread
backed
pinning
service,
so
this
initial
version
would
probably
be
very
simple,
but
it
should
be
created
in
a
way
that
does
not
block
adding
something
like
that.