►
Description
Meeting description here: https://github.com/ipfs/team-mgmt/issues/992#issuecomment-571205170
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
A
Please
add
it
here:
I've
added
UNIX,
fSV
1.5,
which
I've
realized
we
should
have
been
talking
about,
or
at
least
flagging
here
for
a
long
time
now,
and
we
have
not
so
it's
there
now
I
think
Peters
gonna
give
us
a
quick
update
on
the
spec
side
of
things
and
alex
is
out
out
of
town
at
the
moment
until
Wednesday,
so
he
will
be
back
in
yeah.
He
knows
more
about
it
on
the
J
implementation
side
of
things.
A
A
If
you
have,
if
you're
here
and
have
a
update
for
what
you've
been
working
on,
that,
you
would
like
people
to
look
at
asynchronously,
then
please
add
it
to
the
bottom
of
the
document.
There
is
a
space
for
you
and
the
template
for
you
to
follow.
We
won't
go
for
it
in
the
meeting,
but
it's
just
for
everyone
to
to
be
able
to
keep
up
to
date
with
everyone
else's
progress
on
things
and
other
than
that
we
have
a
note-taker.
A
If
you
are,
if
you
are
here
also,
please
to
add
yourself
to
your
attendees
list
on
the
on
the
document,
just
so
that
we
know
who
who
is
here
each
time
and
we
will
get
on
with
going
through
the
initiatives
and
initiatives
other
the
big
chunks
of
work
that
we
we're
working
on
and
that
we
need
to
be
updated.
So,
let's,
let's
start
with
up
Cummings
and
or
shipped
release
things
Stephen.
You
have
written
some
words
here.
Oh.
B
Yeah
so
I'm
proposing
a
release
or
sorry
about
release
for
business,
yeah
I
guess
we
have
some
network
crashers
dinner
causes
people
where
the
problems
and
the
unhappy
obviously
also
the
car
person
does
not
build
on
1.13.
We
thought
what
this
doing
this
a
while
ago
and
decided
not
to
do
it
because
believes
was
right
in
the
corner
and
at
least
is
still
right
around
the
corner.
So
we
think
we
should
probably
do
this.
B
B
So
a
large
part
will
just
be
the
communication
and
well
basically
something
to
do
is
we
need
to
figure
out
which
crashes
you
want
to
actually
take,
so
just
go
through
the
entire,
like
all
the
different
you
know.
First
of
all
like
this
could
generate
infiltrate
jobs
for
doing
real
release,
go
through
all
of
those
changes
and
say:
okay,
which
pieces
to
actually
keep,
which
can
only
be
a
few
things.
But
then
we
try
pick
those
two
circle
at
4.20,
two
branches
of
repos.
B
B
Like
hey,
we
didn't
release,
we
have
to
do
a
short
release
process
which
is
like
make
sure
everything
is
green
and
see
if
CI,
which
is
true,
for
example,
if
we're
going
to
should
so
a
true
and
like
make
sure
this
runs
well
gateways,
which
should
be
fine.
Yeah
really
like
this
is
just
please
it's
not.
There's
not
that
much
to
do,
but
it'll
still
take
a
couple
of
days.
So
just
like
you
know,
make
me
an
open.
B
Minutes,
Eric
five,
others
in
terms
of
that
ones.
That's
the
next
update
so
upgrade
testing
it
breath
release
process.
We
had
the
holidays,
there's
everything
what
I'll
pause,
but
at
this
point
it
works
well
enough
for
local
testing
so
like,
if
you
just
want
to
I,
take
it
for
a
spin,
and
so
one
thing
we
do
right
now
is
like
basic
DHE
tests
like
hey.
How
does
huge
team
work
when.
B
Someone
we
can
seen
something
I
was
like
so
packet
loss
rate
in
that
kind
of
stuff.
It
should
also
be
usable
for
basic.
That's
pop
tests.
I
can't
set
it
variable
agencies
to
kill
excited
about
this
latency.
Well
yeah,
you
can't
say,
like
I,
want
this
latency
this
pH
in
this
tree.
It's
late
to
this
year,
however,
each
node
can
technically
set
like
low
latency.
It's
kind
of
funky
because,
like
you,
can
only
set
egress
latency
about
latency
I'm
so
like.
B
If
you
make
your
node
like
really
hot
I
can
see,
you
won't
actually
victory
about
it.
Only
if
I
drop
out
speed,
but
that's
probably
fine
for
the
list
of
proxy
a
like
check
and
see
like
if
this
wall
works
or
like
you,
have
some
peers
and
set
like
really
high
egress
late
and
seeing
something
lowest
latency
or
especial
something
really
hollow
about
bandwidth.
I
think
you
can
see
like
hey.
Do
we
prefer
peers
that,
like
have
more
bail,
was
available?
Something
like
that.
So
these
are
just
you
know,
run
pretty
easily.
B
The
main
missing
features
are
blocking
internet
connections.
They
setting
these
two
specific
peers
and
running
multiple
versions
of
ipfs,
and
so
with
that,
at
the
same
time,
at
least
from
like
the
local
testing
side
of
things,
from
the
like
large-scale
testing
side
of
things,
I'm,
not
sure
the
correct
statuses.
A
I
vote
that
you
do
release
the
patch
release
and
I'm
I'm
kind
of
more
interested
in.
Why
why
we
just
didn't
do
it
before
and
how
we
can
keep
the
cadence
well
going
so
he's
going
forwards
instead
of
like
waiting
for
this
for
sure
it's
like
said,
0.5
has
been
like
in
the
pipeline
for
a
long
time
and
everyone
to
understand
that
it's
been
delayed
by
getting
test
ground
I'm
so
su,
but
like
it
would
be
cool
if
we
could
just
keep
keep
the
bug
fixes
going
in
the
meantime
yeah.
So
this.
B
Thing
is
like
we
aren't
doing
feature
releases
to
everybody
who
tries
like
taking
the
specific
patches.
We
want
to
include
and
like
putting
out
some
branches
and
getting
it
all
ready
and
the
motivation
previously
was
hey,
doesn't
build
a
1.13
there's,
not
insanely
big
motivation,
because
lucky
were
so
easy
one
to,
and
it
was
silver
leaf
right
there
and
people
just
use
the
real
binaries
or
like
craps
things,
and
it
goes
working
fine,
mostly
as
a
motivation.
B
Now
is
this
crashing
book,
which
is
a
lot
so
that's
why
this
one's
changed
but
like
we
have
other
bugs,
but
they
weren't,
like
critical
things,
would
actually
trigger
our
control
ease,
as
competitors
usually
happens
for
this
water
pipe
list.
I
would
like
you
have
something
actually
an
issue
and
people
actually
upgrades
now,
where
you
can't
just
wait
for
another
release,
because
it
does
take
time.
D
B
No,
this
is
actually
really
annoying.
We
have
no
idea
what
was
causing
it.
We
found
exactly
where
the
bug
is
for
a
second,
beyond
where
the
bug
was
being
triggered.
We
have
a
patch
now
it
appears
to
fix
the
bug,
because
we
just
take
a
bunch
of
locks.
We
still
don't
know
why
we
have
to
take
those
locks,
he
doesn't
make
any
sense.
B
We've
spent
a
long
time,
debugging
it
and
trying
to
figure
out
how
what
actually
happened
here.
Why
are
these
things
like
running
same
time?
I
shouldn't
basically
code,
looks
totally
fine.
It
looked
kind
of
funky,
but
totally
fine,
so
we
just
have
locks
like
you
dealt
with
it
yeah,
so
the
number
of
like
crashes
now
is
much
lower.
We're
only
getting
a
few
people
recording
them
for
a
while.
I
was
like
we're
really
high
I,
don't
know
what
someone
was
testing.
I
bet.
B
So
well,
there
were
two,
but
which
path
is
that
be
able
to
play?
Oh,
you
put
a
new
version
of
gbic,
okay
yeah.
That
means
the
bugs
still
there
somewhere.
The
bug
is
still
there,
but
it's
very
much
attenuated
I
in
the
gateways
in
like
most
of
us
notes,
are
still
running
Tier
one
for
twenty
two
and
we
have
not
been
getting
as
many
complaints
about
people
are
still
saying:
yeah
crashes
every
few
days,
which
is
not
not
good
now.
This
is
all
like
high
loaded
nodes.
B
A
Think
get
the
bad
fats
or
and
and
go
yeah
cool
okay,
in
which
case
the
only
news
I
have
on
Jeff
s
is
that
last
year
we
didn't
manage
to
get
out
a
pre-release,
but
the
the
work
that
I
wanted
to
get
done
has
been
done.
It
is
literally
just
a
case
of
going
through
to
test
now
and
fixing
them
up
and
I
had
very
little
time
over
the
holidays.
A
A
E
The
superior
Ingo
ipfs
repo
when
Steven
starts,
is
working
on
adding
subdomain
Gateway
support,
like
native
support
out
of
the
box
to
go
activist,
and
we
sort
of
like
try
to
figure
it
out
the
more
the
most
flexible,
the
most
useful
way
to
represent
gateway
configuration
in
like
go
idea.
First,
config,
because
we
got
not
only
the
path
space
which
way
when
you
go
to
X
some
domain
and
then
/a
k,
FS,
/
ready.
You
also
got
now.
We
are
adding
subdomain,
so
we
got
CID
the
host
header
and
we
always
had
the
DNS
link.
E
So
the
at
least
three
types
of
gateways
that
we
sort
of
need
to
both
support.
What
also
gives
people
X
ability
to
be
able
to
maybe
just
run
one
type,
maybe
just
run
DNS
link
for
that
one
specific
domain
their
own,
and
they
don't
want
to
allow
arbitrary
stuff
to
be
returned
by
their
gateway.
If
someone
points
some
other
domain
at
their
own
IP
things
like
that,
so
it's
I
think
we
are.
E
So
I
think
if
we
get
like
more
eyeballs
on
it
and
there
are
no
red
flags,
I
think
we'll
just
move
forward
with
the
implementation.
I
think
I
will
probably
take
this
over
from
Steven
and
do
go
and
then
probably
do
the
same
thing
in
jail
so
making
sure
we
got
the
same
of
config
and
implementation.
Both.
E
A
What
are
the
remaining
pieces
of
base
first,
two
and
like
subdomain
gateway
the
way
we're
missing
now.
So
the
like
this,
this
config
option
is
so
this.
Isn't
this
isn't
like
it's
not
a
dependency
of
of
getting
it
done,
but
it
was,
it
would
just
be
good
to
have
it
or
is
it
something
that
is
oh,
it's.
E
Actually
it's
actually,
when
you
try
to
implement
it,
you
really,
you
could
quickly
realize
you
got
like
edge
cases
when
you
got
local
gateway
and
you
don't
want
to
have
both
a
sub-domain
gateway
and
path
based
gateway
on
the
same
host
name,
because
that
you've
got
basically
like
the
origin.
Separation
is
useless,
so
you
need
to
have
like
either.
E
E
So
so
it's
like
the
configurations
wanting
and
now
implementing
native
support
for,
basically
what
is
in
coming
in
the
house
header
to
the
existing
gateway
functionality,
we
got
like
gateway
in
go
ipfs
and
in
Jessica.
First
go
already
looks
at
the
house
header
for
DNS
linked
websites.
Now
it
needs
to
be
like
expert
extended
to
support
those
CDs
in
subdomain
I,
think
that
will
be
more
or
less
I
think
we
should
have
that.
E
It
should
be
fairly
easy
to
add
support
for
like
proxy
mode
for
on
a
gateway
port,
because
that's
like
tiny
implementation,
detail
more
or
less
and
that's
who,
but
but
the
most
important
part
is
like
having
support
on
the
local
gateway
and
that
will
unlock
a
lot
of
stuff
like
on
the
operating
systems
which
supports
local
subdomains.
We
we
could
like
enable
redirect
to
more
secure
host
names
on
localhost.
E
F
A
A
B
B
So
it's
class
I
have
not
done
manifesting
on
a
hard
drive,
but
I
seem
faster
because
they've
hired
agency
actually
have
another
update
that
somewhat
oh,
never
led
to
this.
It's
related
to
CID,
v1
and
I.
Go
has
switched
to
providing
multinationals
instead
of
see
IDs
now
and
the
DHT
you
switch
to
accepting
arbitrary
byte
strings,
provide
our
keys
instead
of
trying
to
validate
them.
G
B
Because
we've
we've
employers
audition,
we
are
actually
using
that
the
own
people
putting
our
CD
zeros,
which
still
fell
our
site
very
multi
hashes,
which
are
catv
zeros,
which
Atal
validate.
We
just
moved
it
because,
like
there
was
actually
no
reason
for
it.
The
only
thing
we
really
validate
was
like
it's
not
too
long,
which
is
what
Bella
English.
A
C
No
too
much
of
an
update
so
basically
I've
been
working
on,
recording
the
concept
changes
I'd
be
making
to
JavaScript
version
Pittsburgh.
So
it's
more
of
a
question.
I
guess
like
I
guess,
with
Steven
I'm,
trying
to
side
whether
it's
worth
taking
another
swing
at
the
test,
ground,
stuff
or
I
should
wait
a
little
bit
until
until
it
stabilizes.
So
it's
sort
of
a
bit
more
polished,
I.
Think.
B
It
should
be
ready
for
testing
it's
what
especially
right
like
setting.
Ladies
and
bandwidths
and
I
see
what
happens
there.
It's
not
good
even
for
testing
multiple
versions
if
it's
wrong,
but
it
should
give
you
a
decent
like,
like
so
anything
HS
plan
it
like,
since
of
us,
actually
already
have
this
plan,
but
like
improving
that
specially
set
late
in
season
like
see
what
happens,
it
doesn't
pick.
The
right
note,
I
think
is
worth
it.
C
I,
like
one
other
thing,
I
I,
was
kind
of
running
into
before
is
the
way
that
I've
written
my
test
plan.
It's
kind
of
like
a
generic
test
plan
that
you
apply
parameters
to,
but
then
like
they
didn't,
seem
to
be
anything
within
test
craft
for
orchestrating
running
a
whole
bunch
of
these
with
different
parameters.
It's
like
I
wanted
tests,
I
see
what.
C
B
A
B
A
B
The
problem
is
like
you
need
to
spin
up
the
test
cluster
from
outside,
and
the
teratoma
can
tear
down
the
in
theory.
You
could
do
this
within
a
single
task
feel
like
you,
can
have
a
test
with
a
bunch
of
stages
and,
like
everyone,
syncs
in
round
one
States
and
syncs
around
next
days
in
the
syncs
around
the
next
stage.
B
And
then,
in
the
actual
test
case,
you'd
have
a
for
loop.
The
very
first
thing
would
be
like
you
know:
do
the
basic
set
up
start
everything
and
then,
like
internally,
you
just
like
spin
up
off
your
nose,
shut
them
down,
spin
it?
Well,
we
know
it's
shutting
down
spit
of
all
you
guys
and
show
them
down.
It's
not
perfect
because,
like
you
would
want
to
make
sure
that
you
don't
have
like
cross-contamination
between
the
test
rounds,
but
it
would
actually
work.
It's.
C
B
F
Yes,
so
we
are
going
to
work
towards
an
RC
this
week,
we'll
probably
publish
another
pre-release
tonight
with
all
of
the
latest
fixes
that
we
have
and
then
try
to
get
Narsee
out
the
door
this
week,
pending
some
finalized
integration
with
JSI
BFS
to
make
sure
we've
worked
out
any
of
the
major
bugs
and
then
we'll
just
be
working
on
finishing
up
the
examples
and
then
getting
all
of
our
Interop
tests
finalized.
Let's
go
and
is
think
it
nice.
A
Cool
yeah
things
are
looking
good,
so
far,
I'm
so
enjoy
your
test
that
I'm,
currently
just
chugging
free
to
test,
like
I,
said
before
I've
done
base
I've,
basically
done
all
in
the
work
that
I
wanted
to
do
on
it,
and
it's
just
getting
tests
passing
and
I'm
finding
bits.
I
need
to
change
bits
that
I
didn't
change,
that
I
should
have
changed
and
stuff
for
that.
A
G
G
G
The
actual
method
that
we're
storing
right
now
we'll
concentrate
on
m
time
and
permission
bid
/s
that
structure
on
the
permission-based.
For
now,
we
are
only
storing
the
bottom
12
bits,
which
is
the
9
permission,
bits
plus
sticky,
SGI
dns,
you
Eddie,
and
on
the
M
time
we
are
almost
finished
figuring
out
how
to
do
it
exactly
not
what
to
do
with
unspecified
and
time
whether
to
create
a
0
or
not.
G
A
Thank
you,
yeah
I,
from
the
gfs
side
of
things,
the
PRS
that
Alex
has
submitted
I
think
he's
done
with
the
work.
It's
just
spectators
and
things
around
that
like
little
final
bits,
I
think
that
need
to
be
sorted
out
and
yeah.
Then
it
exists
yeah
yeah
he
has.
He
already
has
a
bunch
of
tests
and
so
yeah
we
can
always
do
it
more.
A
B
G
A
Okay,
it's
nice
to
have
a
spec
finalized
and
well
nearly
finalized
and
a
pilot
implementation
in
in
the
pipelines
anyway.
Okay,
so,
in
which
case
we
have
finished
the
bit
about
initiatives,
I'm
sorry,
we
are
running
over
time
at
the
moment.
What's
up
next
design
review
proposals,
so
it
Lytle.
Is
this
you
just
asking
people
to
comment
on
the
subdomain.