►
From YouTube: NuPIC Development Progress Review, May 15, 2015
Description
No description was provided for this meeting.
If this is YOUR meeting, an easy way to fix this is to add a description to your video, wherever mtngs.io found it (probably YouTube).
A
Hello
doback:
this
is
a
new
pic
development.
Progress
review
form
a
15
2015
a
lot
of
things
to
talk
about.
We
skipped
our
last
review
in
favor
of
an
email
review,
because
I
wasn't
feeling
well
so
on
this
review,
I'm
going
to
go
over
a
bunch
of
problems
that
have
been
happening
and
what
we're
doing
to
try
and
fix
them,
and
then
we'll
talk
about
a
couple
of
good
things,
but
mostly
will
be
talking
about
problems.
A
First
thing:
I
will
take
a
quick
look
at
the
roadmap.
This
is
the
high-level
roadmap
that
that's
got
put
together,
I'd
like
to
point
out
that
serialization
is
at
the
very
top
of
this
list,
we're
still
trying
to
work
through
this
on
our
other
sort
of
dynamic
roadmap.
That
is
again
the
first
thing
that
pops
up
here
so
there's
a
lot
of
stuff
to
do.
If
you're
looking
for
something
to
do,
we
need
help
doing
this.
The
serialization
tasks,
there's
lots
of
examples
of
things
that
have
already
been
serialized.
A
A
A
So
let's
talk
about
Travis
builds,
so
we've
had
a
problem
this
past
week
more
like
two
weeks
with
our
build
time.
So,
as
you
can
see,
there's
a
builds
over
time
with
the
most
recent
builds
on
top
we
hit,
we
started
to
increase
our
builds
pretty
significantly
at
some
point.
We
figured
out
what
it
was
and
we
have
a
fix
in
place
and
I.
A
So
here's
the
PR
are
actually
the
issue
that
defines
it
and
there's
PR
that
has
a
fix
that
has
decreased
the
build
for
quite
a
bit
that
the
thing
that
bothers
me
is
we
didn't
realize
this
was
happening
until
we
had
so
many
red
builds.
A
lot
of
this
is
my
fault,
because
I'm
the
one
that's
supposed
to
be
monitoring
this,
but
frankly,
I
have
not
been
paying
attention
very
well
recently
because
of
other
responsibilities.
A
So
I'd
like
to
have
some
better
type
of
early
warning
system
in
place,
so
that
these
don't
pile
up
like
this
and
then
all
of
a
sudden
I
realize
oh
everything's
broken,
but
we
do
have
a
PR
to
fix
this
whoops.
This
isn't
it
and
it's
a
I'm
just
waiting
for
somebody
to
give
it
a
thumbs
up.
So
once
that
is
fixed,
this
is
a
heck
of
a
lot
better
and
all
it
all
it
does
is
switched
from
the
Python
to
the
C++
spatial
pooler
implementation
and
that's
generally,
what
our
problem
was
anyway.
A
So
it's
not
a
big
deal,
but
we
need
to
get
that
fixed
okay.
So
another
problem
that
I'm,
having
is
you
might
have
noticed
that
usually
we
don't
have
this
status,
that
we're
at
the
core
is
behind
master,
because
we
have
these
automated
PRS
that
come
up
that
create
that
get
created
by
the
tooling
server.
Those
aren't
getting
created
anymore.
So
there's
a
bug
there.
A
So
if
you
need
new
pic
or
to
get
synced
to
master,
you
need
to
create
that
PR
yourself,
I'm
working
on
a
fix,
I'm,
not
sure
exactly
what's
wrong,
but
I
I,
look
I,
dug
into
the
tooling
server
a
bit
and
found
a
lot
of
stuff
that
was
going
wrong.
So
just
one
of
those
things
is
here
is
a
visualization
of
all
of
the
github
events
that
being
sent
from
github
to
the
tooling
server
and
the
red
ones
are
our
requests
that
are
never
closed.
A
A
This
is
really
because
just
bad
design,
the
github
event
processing
was
sort
of
a
tack
on
process
for
the
tooling
server,
but
it
has
become
one
of
the
most
important
things
that
it
does
so
I'm
going
to
sort
of
re-evaluate
how
that
whole
thing
works.
It's
pretty
nasty
now
I
figured
out
a
much
much
better
way
of
doing
it.
It
shouldn't
take
too
long.
It's
just
a
little
bit
of
tedious
work
base
get
done
and
that
should
fix
the
problem
with
the
missing
PRS.
That
should
fix
the
problem
with
missing.
A
All
of
these
get
events
that
are
coming
from
github,
and
it
also
should
fix
this
other
problem
that
shows
up
when
I
go
through
the
server
logs
for
tooling
server,
which
I
see
this
every
step.
Every
few
hours,
the
chili
server
crashes
and
restarts
so
I
need
to
give
it
a
little
TLC
like
I,
said:
I
have
not
been
paying
much.
A
Attention
is
to
look
so
recently
because
it's
just
been
sort
of
running
and
crashing
in
the
background
and
not
letting
me
know,
what's
going
on
so
I
will
be
working
on
that
aside
from
the
tooling
and
build
problems,
we
do
still
have
quite
a
few
open,
pull
requests.
I'll
see,
we
have
32
open,
pull
requests,
23-year
overdue,
I
said
atlanta
and
last
week's
email
I'm
going
to
try
and
come
up
with
a
potential
process
change
or
some
type
of
something
to
try
and
clean
these
up.
I
hate.
Having
so
many
pull
requests
open.
A
A
A
So
that
caused
a
problem,
because
anybody
who
has
a
new
pic
check
out
older
than
two
months-
and
they
continue
to
build
off
that
if
they
have
not
pulled
down
that
new
pic
or
binary
file,
they
will
get
an
error
and
it
just
will
not
exist
because
it's
been
flushed
from
s3,
so
I've
increased
that
to
a
year
at
this
point
and
Scott's
proposed
to
fix
to
work
around
it
entirely
when
those
tar
balls
don't
actually
exist.
This
is
also
affecting
our
wheels
for
linux.
B
A
Okay,
so
those
so
the
good
good
news
is
we
have
a
new
release
of
new
pic
after
all
that
bad
news
and
the
other
good
news
that
I'd
like
to
share
is
that
the
hackathon
is
got
122
people
rsvp'd.
So
there's
going
to
be
a
lot
of
people
here,
so
yay
eight
spots
remaining
it's
in
15
days,
I
think
it's
15
days,
14
days,
18
hours,
50
minutes
and
43
seconds
we're
going
to
start
hacking
if
you're
going
to
the
hackathon.
A
A
Registered
here-
and
you
can
say,
hey
I'm
looking
for
teammates
and
I'm
interested
in
doing
this
or
that-
and
you
can
see
everybody
else-
who's
who's
already
registered
quite
a
few.
You
can
try
and
connect
with
them.
You
can
start
conversations
and
all
this
jazz
like
a
little
social
networking
site
for
packet
on
stuff,
so
that's
fun
to
be
good
tool
for
us
to
use,
and
let's
see
there
was
one
other
thing.
I
think
I
wanted
to
point
out
and
now
I
just
can't
remember
what
it
was.
I.
A
Wrote
all
this
stuff
down
and
I
have
one
other
thing:
I
camera
it
was
so
I
guess
I'm,
just
gonna.
Let
it
go.
Oh,
oh
I
remember
what
it
is
now,
okay,
so
some
of
the
reasons
that
I've
been
sort
of
neglecting
some
of
these
tooling
processes
and
haven't
been
quite
as
available
as
usual
is
I've
been
working
a
lot
on
while
the
hackathons
one
thing.
The
other
thing
is.
A
We
have
a
significant
amount
of
code
at
momenta
that
has
been
private
for
quite
a
while
that
we're
planning
on
open
sourcing
very
soon
it
will
be
gplv3,
because
it's
all
stuff
that
depends
directly
upon
new
pic.
This
will
be
our
grok
for
IT,
analytics
code
base
and
as
well
as
a
another
project
that
we've
been
working
on,
that
we
have
a
bit
much
of
a
fuss
about
right,
but
it's
pretty
cool
I,
don't
have
a
date
for
this
and
we're
still
actively
working
on
it.
A
At
this
point,
it's
just
legal
licensing
crap,
but
it's
it
will
happen.
Hopefully
it
won't
happen
for
the
hackathon.
I
will
send
information
out
on
the
on
the
mailing
list.
I'm
sure
it'll
be
something
and
in
the
newsletter
at
some
point,
but
it's
just
a
more
application
code
of
different
things.
You
can
do
with
new
pic
and
I
think
they're
pretty
neat
apps.
So
I've
been
working
on
that
just
the
licensing
issues,
mostly
so
in
case
you're
wondering
why
all
this
stuff
is
broke
and
matt
hasn't
done
anything
about
it.
A
B
I'll
have
a
question:
I,
don't
know
if
this
is
the
right
forum
for
it
or
not,
but
I've
been
taking
a
look
at
Rafe
sellers,
JavaScript
port
yeah
that'd,
be
something
to
discuss
here
or
is
that
a
different
thread?
No.
B
Anyway,
I
sent
an
email
to
Raven
about
contributing
to
his
project.
I
didn't
get
a
response,
not
sure,
so
what
I'm
trying
to
do
is
take
his
his
port,
which
is
a
port
of
a
port,
so
he
ported
JavaScript
a
version
of
java
script
from
HTM
java
and
I'm
interested
in
getting
that
configured
to
work
on
server
side,
JavaScript
on
a
like
a
node
or
iOS
j/s
and
I,
just
kind
of
wonder
what
the
right
approach
should
be.
B
A
B
I'm
not
sure
actually
I,
don't
I,
don't
know
if
you
have
much
contact
with
Rafe
or
not
I've
been
able
to
get
in
touch
with
him.
Oh
I.
A
Mean
it,
you
should
work
that
out
with
him
I
don't
have
I
I
mean
I,
read
his
messages
on
the
mailing
list,
but
but
I
haven't
looked
at
his
port
at
all,
but
but
it's
up
to
him.
You
know,
however,
he
might
just
say
no
just
take
and
run
with
it
and
do
whatever
I
don't
really
care
for
contributions,
in
which
case
you
might
program
it
a
lot
differently.
But
if
he's
willing
to
accept
your
changes,
you
might
want
to
do
it
end
up
in
another
fashion.
So
it's
up
to
you
about
them.
A
Sure
anything
else
from
from
the
office
all
right.
Okay,
thanks
guys,
can
you
hear
me
now?
I
can
hear
you
now
all
right,
I'm
looking
for
the
hackathon!
That's
all
I
got
yes,
you
too,
so
so
I
I
didn't
make
much
of
a
fuss
about
this,
but
I
did
go
up
and
chain
and
updated
the
schedule
so
1,
I'm
really
excited
about
the
session.
We're
going
to
have
music
theory
will
have
a
special
guest
musician
for
this
music
theory
discussion.
A
Last
time
we
had
like
this
discussion,
the
subject
matter
expert
on
neurology
doctor,
richard
cantera,
something
similar
to
that
will
have
a
subject
matter,
expert
on
music
theory
and
also
has
experience
with
artificial
intelligence
programming.
So
it
should
be
really
interesting.
My
the
goal
I
want
to
do
is
to
have
some
solid
ideas
on
how
we
can
create
a
media,
encoder
and
I.
Think
we
need
some
some
knowledge
of
music
theory
to
to
do
that
properly.
So
that
should
be
pretty
pretty
cool
and
that's
it
take
care
you
on
the
mailing
lists.