►
From YouTube: HTM Hackers' Hangout - June 3, 2016
Description
Thinking we'll talk about the forum migration and NuPIC tickets.
A
Welcome
to
the
HTM
hackers
hang
out:
we're
live
on
YouTube
I'm,
Matt,
Taylor,
community
manager
for
the
HTM
community,
and
we've
got
David
Ray
as
usual
and
Richard
Crowder,
calling
from
the
UK
and
Marcus
Lewis
in
the
midst
of
office.
Hi
everybody
thanks
for
joining
us,
hey
guys.
I
have
a
few
topics,
but
not
a
whole
lot,
so
let
me
run
through
them
real,
quick,
first
off
if
first
off
HDM
forum,
so
there's
a
big
for
migration.
A
I
hope
you
guys
made
it
if
you
haven't
made
it
to
the
new
forum,
it's
the
it's
the
place
to
be
so.
It's
a
discourse,
new
meta,
gorg
you'll,
have
to
create
an
account
and
everything
most
everyone.
Well
most
of
the
most
active
members
of
the
community
have
migrated
twenty-nine
percent
of
all
of
our
mailing
list.
Subscribers
across
all
our
mailing
list.
That's
unique.
Building
with
subscribers
have
migrated
to
create
a
forum
account
which
sounds
slow,
but
it's
actually
pretty
good.
A
Considering
there's
a
lot
of
ER
the
years
we've
had
these
mailing
lists,
there's
a
lot
of
people
that
have
joined
and
just
forgotten
or
just
never
interacted
with
the
community
at
all.
Both
the
vast
vast
majority
of
those
people,
so
I
got
a
better
impression
of
you
know
that
we've
got
318
of
those
thousand
plus
plus
people
have
migrated
over
the
forum,
so
I
know
at
least
those
are
active
interested
users
because
they
they
moved
over
to
the
forum.
A
So
we've
got
a
total
of
435
users
right
now
and
an
average
of
about
60
different
people,
users
visiting
the
site
every
day
over
the
past
week,
which
is
not
bad
186
topics
and
825
hosts
so
healthy,
not
too
too
much,
which
is
good,
but
I
think
that's
that's
pretty
healthy
yeah.
I
would
say
it's
about
the
same
level
of
conversation
that
we
were
having
on
the
old
mailing
list.
A
So
the
next
thing
is,
if
you're,
a
new
pic
developer.
This
is
just
a
new
big
thing:
I've
been
slowly
working
through
all
the
issues,
I'm
going
to
review
every
single
issue,
there's
over
500
issues
in
New
pick
a
new
record
I'm
trying
to
do
10
or
20
today,
I
just
can't
stomach
to
take
like
a
week
and
do
it
all
so
I'm
trying
to
do
10
or
20
a
day.
So
you
might
notice
me,
resurrecting
these
ancient
issues
and
asking
people
to
comment
on
them.
A
So
if
you
see
me
at
mention
you-
and
these
please
respond,
if
you
don't
I,
might
end
up
closing
that
ticket
I'm
marking
some
of
them
as
under
review,
the
ones
I
think
needs
some
more
interaction
from
the
original
poster
or
someone
to
vent.
The
idea,
or
some
discussion
needs
to
happen
on
it
before
we
can
like
triage
that
and
prioritize
it
once
I
go
through
them
all.
I'll
try
and
prioritize
them,
so
we'll
actually
have
a
real
backlog
of
valid
issues
and
then
some
that
kind
of
our
priority
one
two
three
four
to
help.
A
People
at
least
get
an
idea
of
what
is
important
to
work
on
and
that
oh
and
the
last
thing
is
Jeff's
new
research
ideas
about
sensorimotor
inference,
I
wanted
to
point
out.
There's
a
new
YouTube
video
I
talked
to
Jeff
and
suba
type
about
this
over
the
past
couple
weeks
and
they
agreed
to
post
this
internal
video.
We
just
this
is
just
sooo.
Bataille
opened
up
a
laptop
at
a
research
meeting
and
just
pointed
at
the
whiteboard
and
Jeff
did
an
overview
of
his
current
thinking
about
new
sensorimotor
inference
research.
A
This
is
going
to
kind
of
send
us
new
meta
into
a
research
phase
to
investigate
this
theory.
Try
and
do
some
implementation.
We
do
not
have
a
schedule,
we're
not
we
I,
don't
have
that.
We
don't
have
the
details
about
that.
We're
still
sort
of
working
it
out,
but
that's
going
to
be
so
they'll
find
out
things
as
we
find
out.
I
just
noticed,
subitizing
the
office
thanks
for
joining
us
ubat
I.
Do
you
have
anything
you'd
like
to
say
about
that
since
you're?
There
no.
B
A
And
I
think
that's
a
good
format,
so
we're
having
an
office
hour
on
tuesday
this
coming
tuesday
june
seventh
there's
a
post
about
it
on
HTM
forum,
and
this
posting
this
video
ahead
of
time
was
was
my
plan
so
that
the
community
could
percolate
on
this
a
bit.
If
you
have
any
questions,
ask
them
on
that
forum
post.
That
Jeff
made
about
with
the
video
link
and
I
will
collect
the
the
questions
that
we
think
of
the
best
and
and
we'll
try
to
address
those
at
the
office
hour.
So
that's
sort
of
the
plan.
A
B
The
one
another
thing
that
is
this
is
not
like:
it's
a
well
baked
done
deal
it's
very
much.
Research
in
progress
and
a
set
of
ideas
and
and
concepts
and
a
framework-
and
you
know
will
be
I
found-
will
be
the
old
answer
all
of
these
questions,
because
we
simply
just
don't
know
the
details
of
a
lot
of
them,
but
at
least
you
can
get
a
sense
of
how
Japan
and
the
rest
of
us
are
thinking
about
it.
A
Right
yeah,
that's
the
idea,
so
in
our
whole
initiative
of
trying
to
be
transparent
about
a
research,
that's
why
we
did
this
and
I
know
that
it's
sometimes
frustrating
to
some
community
members
who
want
to
get
into
the
research
and
where
it
is
a
little
bit
of
a
one-way
mirror.
At
times
you
can
see
into
our
research
repository,
but
can't
we
don't
necessarily
take
pull
request,
sir.
This
is
sort
of
the
same
thing.
A
B
A
C
Yeah
there's
a
there's
been
a
few
people
have
been
trying
to
get
it
running,
get
new
pic
1
and
multiply
recently.
Matthews
looks
it
before
I've
looked
at
a
few
people
have
tried
this
coded
hard
and
particularly
patty,
that
on
github
in
the
community,
nice
and
really
good
progress
of
getting
new
pic
or
running.
C
So
we
run
a
stage
at
a
moment
with
a
PR
new
pic
or
where
it
builds
all
of
New,
big
core
and
the
C++
side
and
go
spoonbills
the
bindings,
an
old
Lee
C++
unit
tests
and
they're
passing
on
on
Raspberry
Pi
there's
the
new
pic
side,
six
out
of
the
eight
hundred
or
so
unit
tests
are
not
passing,
and
that's
mainly
just
down
to
be
sick
d.
32-Bit
data
model
in
Linux.
So
there's
issues
there
that
we
need
to
to
look
at,
but
pretty
good
progress
on
getting
run
down
on
Raspberry
Pi.
A
C
Know
that
was
a
nice
thing
that
Scotland
Batali's
work
on
on
building
the
external
libraries
or
source
has
really
helped,
because
it's
it's
tied
it
up
a
lot
of
the
work
you
need
to
do
on
that
side.
So
it's
the
run
a
huge
amount
of
it.
There
aren't
many
changes
need
to
do
to
get
the
bill
to
working
right.
A
A
A
B
B
A
Here's
a
pro
tip
you
might
be
able
to
extract
a
data
set
from
riverview
in
a
in
CSV
directly.
That
would
be
accepted
into
that
app
contest
which
is
like
you
know,
five
minutes
worth
of
work,
so
I
haven't
looked
into
it,
but
it's
possible.