►
Description
Meeting description here: https://github.com/ipfs/team-mgmt/issues/992
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
B
A
A
All
of
the
tests
for
the
examples
and
all
of
the
repos
that
are
our
early
testers,
have
the
use
ipfs
they're,
all
being
all
of
the
tests
for
those
those
things
are
being
run
by
CI
automatically,
so
I
don't
have
to
do
that
anymore,
and
that
was
previously
a
very
time-consuming
task.
So
it's
super
good
news,
but
yeah
0:40
is
out.
Now
it
has
a
few
things
in
it.
There's
a
new
repo
repaint
migration
tool.
A
So
that
is
that
it's
good
news
with
us
in
there
and
ready
to
go.
We
have
support
for
peer
IDs
encoded
as
base
52
in
IP
NS,
so
you
can
now
ask
IP
NS
to
resolve
a
period.
Basically,
two
encoded
CID
EP
a
peer
ID
and
it
will
be
able
to
resolve
it
for
you,
which
is
super
cool.
This
isn't
the
full
CID
v1
version
with
a
Lib
p2p
key
codec.
This
is
literally
just
an
old
style
peer
ID,
encode
re-encoded,
as
as
a
CID
and
and
encoded
using
base
photo
but
yeah.
A
B
Thank
you.
Sorry,
I'm,
just
desperately
trying
to
finish
up
the
notes
for
okay,
so
we
had
a
testing
infraspecific
test
ground
a
hack
week
last
week
got
a
bunch
of
stuff
done,
I
progress,
but
we're
not
quite
where
we
hope
to
be
so.
The
correct
status
is
that
we
can
now
run
a
hundred
to
two
hundred
baby
nodes
on
AWS
instance.
We
can
spin
it
like
many
of
these
APIs.
B
B
Kind
of
do
this
mixed
networks,
there's
a
good
vest
with
JSF
es
or
different
versions,
different
things
like
this.
We
haven't
quite
started
on
this,
yet
we're
not
entirely
sure
the
right
way
to
do
this.
The
current
thinking
is
maybe
just
like
this.
My
current
thinking
is:
you
should
be
deploying
images
that
have
everything
and
then
like
at
runtime.
We
just
figure
out
which
we
want
to
run
the
problem
years
like
many.
This
is
very
using.
Don't
let
you
like
configure
individual
machines.
Instead,
they
say.
B
C
B
B
B
So
that's
still
working
progress,
I
think
I'm
not
sure
the
end
goal.
This
is
the
next
item
there
kind
of
integration,
so
we
would
like
to
be
able
to
just
make
this
like
you
just
submitted
the
art
you
like
call
out
to
some
bot
and
the
bot
and
runs
this
test
on
Amazon.
We
actually
think
that
we
should
build
run
media
or
midscale
tests.
This
way
without
cost
too
much
money
because
like
if
we
just
spin
up
these
nodes
on
demand,
run
the
test
with
who
could
shut
them
down
like
they
cost
like
like.
B
Really,
our
experience
is
like
it
will
costs
almost
like
a
buck
per
physical
machine
if
we
in
the
each
fiscal
cheek
and
that
we
think
around
hundred
two
hundred
nodes.
Although
we're
testing
with
under
two
hundred
knows,
we
were
getting
a
bunch
of
other
crashes,
so
we're
going
to
have
fifty
about,
like
I,
should
be
affordable.
B
Next
dashboard
Jim's
correctly
working
on
that
tried
to
make
it
smooth
country
infuse
the
results
in
some
nice
using
integrated
way.
We've
been
going
back
and
forth
on
the
right
approach
this,
but
progress
improve
literally
tracing.
This
is
something
it's
kind
of
critical
we've
realized
we're
like
we
can
display
it
like
certain
results,
but
we
can't
like
dig
down
into
like
what
is
actually
doing
so
like
we
can
spend
a
timing
result.
So
you
could
say
like
oh
this,
like
this
change,
improve
some
of
this
change
didn't
reverse
link.
B
Really,
we
really
nice
be
able
to
exact
time.
You
like
this
little
piece
here
is
what
took
a
lot
of
time.
This
is
like
this
event
here
got
stuck,
but
we
don't
currently
have
this
and
I,
don't
think
anyone
was
working
on
this
I
think
I
do
not
enjoy
her
role
was
introduced,
I
think
who's
trying
to
find
someone
to
take.
B
So
that's
the
current
status
of
testing
infra
tied
in
with
this
is
the
good
fest
release
process.
But,
of
course,
that's
blocked
on
this,
so
status
update
is
good.
This
will
probably
not
have
a
release
before
lab
week
or
so
a
nursing
for
that
week,
but
it
will
have
a
release
before
the
end
of
the
year,
regardless
what
happens
and
I'm
hoping
to
find
a
time
at
lab.
Wait
to
get
a
bunch
of
people
together
to
final
push
like
writing
tests
protesting
from
and
actually
using.
C
B
Make
that
happened.
It
needs
actually
usable
by
a
Thrall
that
Deb's
like
not
just
in
laptops
right
now.
You
can
totally
use
drops.
It
would
work
great
but
like
actually
like
usable
on
the
network
so
yeah.
That
is
the
testing
update,
which
my
testing
update,
don't
see.
Anyone
else
here
who
was
just
weak
so.
D
With
respect
to
I
think
you
said
there
that
you
can
spin
up
a
whole
bunch
of
instances
on
one
AWS
machine
is
alright,
rather
than
yes
start
rather
than
using
separate
AWS
machines,
yes,
okay,
and
and
is
that
how
we're
sort
of
like
I'm,
just
wondering
if
we're
gonna
have
any
tests
that
I
are
gonna,
be
like
really
simulating
what
might
happen
in
the
network?
You
have
literally
multiple
machines,
I.
B
So
the
thing
is
with
Amazon
like
unless
we
actually
have
like
real
machines,
we're
not
really
gonna.
Have
that
so,
like
the
system
is
supposed
to
be
flexible
enough,
that
I
mean
ideally
we
could,
if
you
want
to
go
out
and
buy
a
bunch
of
enterprise,
it
just
make
it
happen,
but
even
it
like
basically
comes
out
to
cost
we're
like,
even
if
we
can
ask
like,
even
if
they
use
separately,
this
is
Amazon
Amazon.
So
it's
gonna
give
us
different
VMs,
so
they're
not
really
going
to
different
machines.
B
Do
we
were
plenty
of
some
of
this?
Is
using
TC
and
sending
we
can
see
roles
so,
basically,
all
packets
will
think
you
mister,
wait
and
see
what
habit
was
setting
network
bandwidth
like
drop
rates,
all
that
kind
of
stuff,
so
we
think
it
should
be
as
close
as
we
can
get
it
even
if
we
had
thrown
stuff
machines
so
wouldn't
they
would
so
quality
within
the
same
data
center,
so
changes
yeah.
The
ideal
solution
would
be
to
just
set
up
a
massive
farm,
perhaps
revise
all
across
the
country
like
the
next
episodic.
B
Well,
if
you
just
that,
could
be
that
it
would
be
really
awesome
to
be
able
to
just
like
ship,
raspberry,
pies
to
contribute
and
say
hey.
Could
you
just
set
this
up
on
your
network
connected
to
your
router
and
just
like?
Let
us
like
test
like
this
global
internet
testing
system.
That
would
be
awesome,
but.
B
We
do
that
ourselves
to
start
with.
Yes,
no,
like
I,
think
right
now
we're
gonna
focus
on
making
this
just
work,
initially
things,
like
wit,
bit
worthy
says,
but
yes
being
able
to
just
like
set
that
up
deploy.
We
definitely
need
very
good
doc.
Caching,
we're
gonna
do
this.
We
don't
just
like
destroy
people's
bandwidth
by
like
constantly
on
the
new
images
but
yeah.
If
we
could
just
ship
people
raspberry
PI's
that
we
have
access
to,
that
would
actually.
E
B
B
D
One
other
question
there,
when
you
say:
mix
networks,
IP,
f,
s+,
j,
SI,
p,
FS
versions,
etc.
Does
that
include
like
versions
of
dependent
libraries
like
this
one.
B
Yes,
yeah
basically
want
to
be
able
to
play
multiple
versions
of
nodes
for
like
they
have
different
configurations
at
the
moment
like
every
node
is
the
same
and
like
you
just
special,
is
it
runtime,
but
like
most
languages,
don't
let
you
import
multiple
copies
of
the
same
library
unless
they're
like
different
major
versions,
easily
do
that
yeah.
So
again,
the
current
thinking
here
is
like
oh
basically,
a
few
options.
We
could
either
deploy
multiple
images
or
one
image
with
everything
as
specialist
runtime.
B
Now
there
are
trade-offs
like
the
way,
we're
just
rated
it's
better
technical
integration,
the
way
we're
communicating
this
machine
using
Redis.
So
basically
everyone
writes
to
Redis
and
then
everyone's
subscribe
to
read
us
the
problems
like
if
we
end
up
starting
a
new
process.
That
means
like
both
these
processes
need
to
thread
this
and
need
to
coordinate,
and
then
it
just
gets
slightly
trickier.
It's
not
that
bad
okay,
yeah
but
I've
got
it
yeah.
A
Just
so
just
a
quick
clarification,
the
the
go
ipfs
0.5
release
is
hopefully
to
be
expected
by
the
end
of
the
year,
for
things
go
to
plan,
which
means
that
the
test
ground
should
be
getting
ready
or
ready
enough
to
be
able
to
test
things
to
an
adequate
level
that
we're
okay
to
release
zero
fine.
And
so
what
does
that
mean
for
zero?
For
twenty
three?
Twenty
two,
the.
C
C
B
Like
to
fix,
but
it's
not
like
a
burning
fire
like
actually
fixing
the
DHT
and
cutting
at
least
just
takes
time,
even
if
it's
just
natural
ease
so
yeah,
the
plan
is
just
gonna
for
0.50
and
making
out
he
terms
like
what
would
be
included
in
45.0.
That
really
depends
on
what's
ready
when
we
were
trying
to
shoot
for
that
week.
The
goal
was
the
DHD,
because
that's
what
we
think
that's
the
minimum
has
to
be
in
a
release.
C
A
F
Yeah,
so
I
can
swear
pub/sub
it's.
This
is
mostly
like
bubbling
things
up
and
dealing
with
all
the
conflicts
that
show
up
since
pub
set
this
up.
Unsub
changes
to
basically
landed
now
so
yeah.
One
of
them
is
ready
for
review,
which
is
just
like
removing
all
the
old
crufty
food
strapping
code,
because
now
we
have
new
snazzy
stuff
they're
breaking
API
changes.
No
one
should
care.
There's
been
a
forum
post
on
discussed
outlet,
pita
Pio
saying
this
is
gonna
happen
since
June
no
one
has
said
anything
other
than
thumbs
up
so
should
be
fine.
F
Putting
this
into
go.
Ip
of
us
is
the
next
thing.
I
had
not
had
much
experience
with
char
NIST
tests
and
making
sure
that
I
deal
with
all
of
our
fun
new
dependency
injection
stuff.
So
getting
some
review
to
make
sure
I've
done.
That
right
would
be
great
and
then
I've
got
to
do
some
writing
test
ground
tests
for
IPS
things,
which
also
looks
a
lot
like
DHT
put
and
get
so
yeah
I'm
gonna
have
to
talk
with
the
test,
run
team
about
how
to
do
that,
and
and
what
comes
next
any
questions.
F
F
C
F
Windows,
more
than
is
already
broken
because
Windows
is
using
asynchronous
data
stores.
All
the
time
we
just
didn't
realize
it
so
yeah
I
could
go
and
look
through
everywhere.
That
leaves
a
data
store
in
our
code
base.
But
if
people
who
already
know
the
answer
to
this
could
help
me,
that
would
be
great
2:13.
B
F
B
Away
mostly
yeah,
so
just
do.
The
search
interview
like
I
can
check
out
that
you
can
check
your
work,
make
sure
we
got
the
right
things
like
actually
a
wonderful
way
of
doing
this,
like
you,
create
an
issue.
There's
audio
uses
of
like
the
datastore
and
like
says
doesn't
use
ink
does
not
do
you
think
you
just
check
everything
off
that
it'll
check
the
same
thing,
so
I
think
what
quickly,
oh
sure,
that'd
be
great.
Yeah.
F
And
then
the
last
thing
is
that,
because
switching
to
asynchronous
data
stores
is
either
a
use
badger
instead
of
flat
FS
or
make
your
badger
use
a
different
flag.
We
should
consider
whether
the
upgrade
path
is
telling
people
who
have
badger
already
installed
to
change
a
flag
or
to
fix
our
config
files.
I.
B
Think
we
need
to
fix
the
config
files
as
we
talk
to
it
before
we
don't
want
to
get
into
it
now,
but
for
now,
like
look
badger
is
still
technically
experimental.
There's
no
reason
we
can't
just
tell
people
if
you
are
hurting
from
this
performance
issue
switch.
Otherwise
you
find,
but
most
users
don't
like
it
kind
of
hurts,
but
it
like
honestly.
The
performance
is
not
terrible,
congratulating
Justin
tons
of
data
that
really
care,
and
they
will
listen.
Heaven
able
change
this
so
cool.
A
D
So
yeah
I've
implemented
a
test
ground
plan
for
bid
swap
which
is
multiple
seed,
multiple
each
scenario,
and
so
when
I
say
implemented,
I
mean
that
it
works
on
my
local
machine.
So
we
still
need
to
you
know,
make
sure
it
works
in
the
clouds
and
then
I'm
gonna
start
looking
at
how
we're
gonna
actually
visualize.
D
What's
coming
out
of
that
meaning.
Is
it
passing
the
tests?
I
need
to
add
a
couple
of
extra
things
into
that
test
ground
plan
one
is
going
to
be
being
able
to
stagger
when
we
make
requests.
So
if
there's
multiple
leeches,
then
I
don't
know
stat
at
the
same
time
and
I'm
gonna
be
so
I'm
gonna
be
hearing
here
today,
it's
Maura
and
then
from
Wednesday
to
Friday
I'll
be
in
Nicaragua,
so
I'll
be
taking
that
time
off.
Some
people
know
then
I'll
be
in
Costa
Rica.
G
Yeah
chase
left
EP,
we
finished
the
identify.
Refactor
is
done.
That
also
includes
new
support
for
identified
push
nsj
surrogate
relay
reef
actors
in
progress
should
be
done.
This
week
we
refactored
the
pier
store
and
also
added
a
registrar
that
we're
gonna
be
using
for
connection
management,
pub/sub,
flood
sub
and
gossip.
Such
should
hopefully
be
finished
this
week
because
we
finished
blockers
for
those
psychiatry.
G
Factors
mostly
done
I
need
to
review
it
and
I
think
there
might
be
a
couple
small
blockers
that
new
quirk
mkg
did
a
some
local
benchmarks,
there's
a
link
to
those
and
that
showed
a
pretty
significant
performance
improvement
over
callbacks.
So
hopefully
we
should
get
some
boost
in
performance
for
a
jeaious
through
that,
so
that
is
it
for
lipid
P.
A
H
A
People
on
the
async/await
stuff,
we
will
looking
really
good
for
finishing
off
this
work.
This
quarter
there's
only
a
few
things
that
are
left
that
are
in
flight.
The
p2p
is
almost
done
and
there's
a
bunch
of
things
so
gonna,
hopefully
land
this
week.
Ipfs
is
coming
down
to
just
the
HTTP
client
and
ipfs,
and
largely
ipfs
has
already
been
done
so
yeah.
This
whole
thing
can
finish
at
some
point
soon
and
we
can
start
to
realize
some
of
the
benefits.
So
that's
good
news.
B
C
B
Second
design
of
view
on
unix
best
4.5,
but
we've
started
screamin
on
things.
So
we
should
like
talk
about
this
and
just
like
now
that
we
actually
have
like
the
spec
rosa
there.
We
can
heard
about
last
year
things
that,
hopefully
is
emerge.
The
spec
proposal
live
it's
beating
or
make
up
its
live.
Oh,
let's
see,
but
like
this.
B
F
Yes,
I,
don't
know
this
is
maybe
maybe
neither
of
these
may
be
neither
a
blocker
and
or
naw
request,
or
an
ask
nor
a
question
but
I,
given
that
just
live
data
P
like
pub/sub
stuff
is,
is
so
far
along
now
and
gossip
stuff
is
in
there.
Is
there
any
interest
in
porting
over
the
pub/sub
router
changes
as
well,
so
that
the
IP
nests
over
pub
sub
stuff
works
in
JSON?
Or
should
we
wait
on
that
until
it's
released
in
l5
OH.
F
B
F
B
H
F
This
to
why
they
might
be
sad.
The
reason
the
what's
odd
is
that
they're
gonna
try
and
ask
for
they're
gonna
try
and
like
ask
the
knows
that
they
connect
to
and
be
like
hey.
What's
the
latest
version
in
case
I
missed
an
update
and
the
j/s
ones
will
be
like
I,
don't
understand
what
language
you're
speaking
and
the
go
ones
will
be
like
here's,
the
answer.
So
if
they
happen
to
connect
to
a
Jas
know
if
they
happens
to
connect
to
like
only
Jas.
No,
it's
they'll
be
very
sad.
If
I.
B
C
I
Have
to
almost
immediately
so
I
just
wanted
to
mention
that
we
did
see
some
very
weird
behavior
on
the
gateways
on
Sunday,
where
they've
been
running
reliably
for
weeks
and
then
they
all
started
crashing
on
Sunday
all
6
15
of
the
16
nodes
would
crash
roughly
once
an
hour
all
with
the
same
error.
That's
linked
to
from
the
notes,
so
it'd
be
good
to
get
some
eyes
on
that.
So.
B
I
Know
I
just
did
that:
okay,
cool
good
to
know,
and
just
in
case
anyone
else
in
the
message
already.
The
four
new
bootstrap
nodes
are
now
monitored,
so
we
can
consider
them
production-ready,
so
you
can
also
using
them,
but
your
code
will
be
using
them
anyway,
unless
you're
Jaso
give
s,
in
which
case
we
need
to
figure
out
whether
you're
going
to
start
supporting
dns
adders
soon
or
whether
we
should
get
you
some
DNS
flavored
multi
adders
for
the
new
restrain
nodes,
but
I
can't
stay
any
longer,
so
I'm
going
to
leave
goodbye.
B
D
C
C
Things
like
the
async/await
refactor
things
like
getting
o
top-50
at
the
door
and
really
want
to
encourage
everyone
to
focus
and
think
about
how
they
can
best
empower
their
teammates
to
get
these
really
important
things
across
the
line
and
that
you
know
it
will
be
tempting
to
find
new,
shiny
things
and
start
trying
to
push
the
board.
But
let's
be
really
thoughtful
about
kind
of
how
we
collectively
spend
our
effort
to
help
land
these
p0s
and
can
be
thoughtful.