►
From YouTube: NuPIC Development Progress Review - Oct 3, 2014
Description
A
All
right,
hello,
new
pic,
we
are
live,
we're
having
a
new
development
process
review
meeting
and
I'm
matt
taylor.
We
have
in
the
nementa
office's
austin
marshall,
scott
purdy
and
prince
lesbirinsky.
I
think
jaden's
on
the
way
and
so
for
this
meeting.
What
I
would
like
to
focus
on
is
currently
open
prs,
because
we
have
a
lot
of
open
pr's
right
now,
so
mainly
it's
going
to
talk
about
the
work
that
is
in
progress.
A
So
the
current
initiatives
that
are
going
on
there's
several
there's
some
of
these
are
for
google
tests,
like
these
two
issues
are
have
to
deal
with
well
this
this
issue
is
google
test.
There's
another
issue.
It
has
to
do
with
the
google
test
somewhere.
A
A
So
let
me
pop
this
open
here,
so
we
can
have
a
quick
discussion
about
this
pr.
So
this
is
something
that
mark
worked
on
and
we
are
stuck
in
a
place.
A
Let's
see
what
I
wanted,
I
haven't
been
really
super
involved
in
this
discussion.
So
a
lot
of
what
I'd
like
is
to
get
an
update
on
the
progress
of
this
pull
request
for
this
entire
initiative
here,
for
those
of
you
guys
who
seem
to
have
a
better
finger
on
what's
going
on
here.
A
So
if
anybody
wants
to
volunteer
and
give
a
status
update
on
on
this
pr,
what's
holding
it
up,
what
work
needs
to
be
done?
That
would
be
a
great
place
to
start.
B
I'll,
let
austin
do
the
status
update,
but
I
think
we
haven't
been
working
on
that
on
it
this
week
and
maybe
next
week
we'll
be
able
to
come
back
to
it.
But
we
should
at
least
if
mark's,
working
on
it,
which
at
least
make
sure
he's
not
getting
blocked
on
us.
A
Yeah
mark
kind
of
works
in
spurts
if
you've
noticed
he'll
come
in
and
work
for
like
three
days
and
go
a
few
weeks
and
then
come
into
work
a
bunch
in
over
three
days.
So
a
lot
of
the
prs
that
are
currently
open
are
assigned
to
him,
as
you
can
see
over
here
on
the
right
and
I've
gone
in
and
commented
on
pretty
much
all
of
them
and
asked
him
what's
what's
the
status
under
this
or
that
so
the
plan
is
to.
A
C
Yeah,
so
the
problem
is,
is
basically
the
same
as
we
discussed
two
weeks
ago
at
the
last
review.
C
But
what
had
happened
was
the
siblings
plus
11
changes
were
completed,
they
were
merged
into
nuke
core,
but
new
pick
was
not
compatible,
so
we
were
trying
to
work
to
get
new
pick
compatible
and
where
they're
incompatible
is
seems
to
be
the
intersection
of
osx
and
swig
and
steeling,
and
we're
able
to
get
it
working
such
that.
C
If
part
of
it
was
compiled
with
ceiling
and
the
other
part
of
it
was
compiled
with
gcc,
it
could
all
build,
but
then
there
were
still
some
linking
issues
and
even
to
get
it
to
that
point,
it
required
quite
a
bit
of
a
hack,
so
not
making
a
whole
lot
of
progress.
We
just
backed
off
on
that
reverted
those
changes
and
thrown
the
back
burner
so
that
we
could
figure
out
what
the
problem
was.
C
In
the
meantime,
I've
not
been
able
to
work
on
it
because
I
was
working
on
other
internal
fundamental
stuff,
but
I
think
mark
might
have
been
able
to
work
on
some
of
it.
I
just
don't
think
that
it's
it's
ready
yet.
A
Okay,
I
mean
the
core
issue
here.
If
I'm
not
mistaken,
is
that
we're
doing
c
compilation
stuff
in
both
of
our
repository
builds
and
if,
if
we
could
simplify
that,
so
that
it
only
happens
in
new
pick
and
we
do,
we
eventually
have
a
binaries
published
that
could
be
pulled
in
by
the
new
pick
builds.
Then
we
wouldn't
be
having
this
problem
right.
A
That
is
the
direction
we
want
to
go
yes,
yeah,
but
it's
not
like.
We
have
all
the
time
or
the
bandwidth
to
do
that
right
now,
or
else
we
would
just
do
that
right
now,
right,
okay,
so
that
makes
sense,
so
this
is
still
open
and
ready
for
work
mark
if
you're
watching
welcome
any
of
your
comments
on
here.
Aside
from
the
last
comment
that
you
had
we'll
be
able
to
work
on
this
coming
up
in
the
next
couple
of
weeks,
this
is
blocking
up
several
other
issues.
A
Getting
the
google
tests
running
in
the
new
pick
build
utensil
has
a
pr
for
this.
He
already
did
the
work
in
new
big
core,
which
is
has
been
merged.
I
believe,
if
it
has
been
merged,
I
think
I
can
no,
no
it's.
This
is
still
open.
A
This
is
a
we
were
running
into
a
problem
with
this
too
right.
I
wanted
to
talk
about
that.
He
he
has.
He
presented
some
options
on
how
to
fix
this,
because
I
think
there
were
linking
problems
on
osx,
so
he
is
working
through
this
right
now.
So
it
looks
like
he
even
worked
today
on
this,
which
is
great
and
I'm,
if
you're
watching
utensil,
I'm
ready
to
try
and
test
this
again
once
you're
you're
ready.
A
So
if
that
works
and
that
gets
merged
and-
and
we
should
be
able
to
merge
this
one
excuse
me
after
we
get
the
c
plus
plus
11
thing
done.
So
that's
kind
of
the
cpis
plus
11
one
is
holding
up
the
google
test
one
and
then
there's
these
these
anomaly
tickets
here,
like
normally
strict
anomaly,
call
likelihood
that
are
both
marks
projects,
and
I
believe
the
state
of
both
of
these
are
just
builds
are
erroring
and
he
just
needs
to
put
some
more
work
towards
it.
A
No,
this
one
looks
like
it's
it's
passing,
but
he
has
marked
it
as
blocked
and
I'm
not
sure
why
so
I'm
waiting
for
some
feedback
from
him,
the
anomaly
strict
one.
I
believe
this
is
the
one
that's
not
quite
yeah.
There
is
an
error
on
this,
build
that
he
was
still
working
on
all
right.
So
that's
thanks
for
clearing
up
the
c
plus
plus
11
issue,
I'm
glad
we
hopefully
will
be
able
to
put
that
to
bed
over
the
next
couple
of
weeks.
Let's,
let's
have
a
little
swing
update.
A
So
I
know
basically
nothing
about
swig
at
all.
Mark
started
this
because
we
started
talking
about
how
we
needed
to
update
swag.
So
he
took
the
initiative
and
obviously,
as
you
can
see,
there's
over
a
thousand
files
changed
in
this
pr
which
really
sucks
to
see
that
many
changes
and
austin
has
been
helping
out
with
this.
But
at
this
point
I
don't
know
what
the
next
steps
are.
So
austin,
if
you
know
what's
going
on
here,
status
update,
would
be
good
yeah,
so.
C
C
A
Well,
it's
definitely
not
ready,
because
the
builds
are
all
erroring.
Well,
of
course,
so
I
don't
totally
know
what
the
state
is.
So
what.
A
By
the
c
plus
11
efforts-
oh
great,
I
hear
you
okay,
so
this
could
be
similar
to.
Let
me
I
should
at
least
mention
it
here.
A
So
it's
it's
very
possible
that
over
the
next
two
weeks
we
could
update
swig
you
get
c
plus
11
running
and
get
google
test
running,
which
I
would
be
totally
happy
with
that
all
right,
there's
also
one
other
thing
I
wanted
to
talk
about.
A
That
was
bothering
me,
which
was
trying
to
add
this
new
pic
linux,
64
back
to
the
travis
ci
build.
So
the
idea
behind
this
is
these
platform
repositories,
new,
big
linux,
64
and
new
pic
darwin
64..
A
They
just
contain
dependencies
beliefs
are
they
I
think
they're
already
compiled
dependencies
for
linux
and
darwin
darwin
being
os
x
and
we've
been
using
it
for
a
long
time.
We
use
them
internally
for
numenta
pipelines,
but
I
don't
think
the
community
generally
uses
them
very
much,
but
it
would
help
if
we
put
it
back
into
our
travis
ci
build.
A
It
might
shave
off
three
five
minutes,
something
like
that
on
our
build,
and
it
also
would
help
be
nice
just
to
continually
validate
that
these
repositories
are
are
properly
working
because
then,
currently
they
are
not.
So
I
tried
adding
this
back
and
realized.
We
had
a
problem
with
pill
and
pillow
pillow
is
or
excuse
me
pill
is
the
predecessor
to
pillow.
A
I
think
pillow
is
a
fork
of
pill
and
from
what
I
read
in
the
documentation,
they
use
the
same
name
space,
which
is
pil
as
far
as
python
name
spacing,
so
we
shouldn't
have
to
change
any
of
the
code,
but
when
I
try
and
compile
new
pic
on
top
of
linux
64,
I
can
continue
to
get
the
same
problems
with
not
having
this
lib
tiff
binary
on
the
platform
and
that
is
required
by
pillow
and
it's
not
easy
to
get
on
ubuntu.
A
The
ubuntu
version
that
travis
cis
is
working
on.
I
also
tried
to
create
a
pr
on
new
pick
linux
64.
That
removes
all
of
the
pill
stuff.
I
don't
know
if
I
did
this
right,
but
I
just
basically
removed
all
of
the
all.
A
The
files
that
I
found
related
to
pill
tried
to
build
new
pick
on
top
of
it
and,
interestingly,
although
pillow
supposedly
exists
in
here,
I
get
a
no
module
named
pill,
import
error
which
confused
me
so
that
that's
where
I'm
at
on
this
and
I-
and
I
I'm
not
sure
where
to
go
at
this
point
yeah
and
that
led
to
this
issue.
You
know
pill
depends
on
this
certain
library,
lib
tiff
version
5,
which
we
can't
get
access
to
on
in
the
travis
ci
environment,.
C
Yeah,
so
I
have
a
little
bit
to
add
to
that.
I
looked
into
it
last
week
earlier
this
week
and
I'm
a
little
bit
stumped
on
it
in
part,
because
I'm
not
sure
what
to
do
to
fix
it,
because
what
I
believe
happened
originally
is.
C
The
new
pic
linux
64
was
built.
It
was
either
built
on
ubuntu
or
sent
to
us
and
on
one
it
would
link
dynamically
against
lib
tiff.so.5
and,
on
the
other,
it
would
link
against
lip
tiff.4.
A
Including
us,
because
our
pipelines
are
sent
os
and
that's
probably
exactly
what's
going
on,
because
I
think
the
linux
64
was
compiled
for
for
us,
because
everything
is
working
in
our
pipelines,
which
makes
it
seem
like
this
lib
tiff
is,
is
available
in
our
in
the
centos
environment.
We're
using.
C
A
Yeah,
so
I
I
did
try
when
I
removed
all
the
pill
files,
I
did
try
to
go
into
image
sensor
and
import
from
pillow
instead
of
pill,
and
I
I
I
believe
I
got
the
exact
same
error,
except
from
no
module
named
pillow
was
the
import
error
that
I
got
but
truth
be
told.
I
don't
think
this
is
a
serious
issue.
A
I've
got
it
marked
as
a
p2,
because
this
having
a
new
pick
linux,
64
platform
repository
in
a
good
state,
I
think,
is
important,
even
even
even
if
we're
the
only
ones
using
it
currently,
if
we're
going
to
be
using
it,
we
should
at
least
have
it
of
confirmation
that
that
it
actually
works
in
the
environments
we
want
to
use
it
in
and
travis
is
one
of
those.
A
So
I'm
going
to
keep
these
open
on
as
a
as
a
p2
for
the
time
being
and
hope.
Maybe
we
can.
We
can
find
some
time
to
figure
out.
What's
going
on
there.
A
Okay,
let
me
see
if
there's
any
other
major
issues
this.
This
is
one
that
david
ray
just
created
and
it's
running
through
travis
right
now,
mark's
issues.
Oh,
let's
talk
quickly
about
the
windows
build,
so
this
is
I
I
want
to
get
this
richard
richard
crowder,
who
started
this
windows,
build
initiative
built
built
on
top
of
work.
I
think
that
david
ragazzi
had
done
in
the
past
and
he's
sort
of
recruited
he's
done
a
lot
of
work
here.
A
If
you
can't
tell
to
get
this
working
on
on
windows
and
he's
recruited,
another
guy
ahmed,
because
github
name
is
candeel
here,
it
is,
and
they
have
been
kind
of
working
on
this
together
talking
on
gitter
about
it
and
let's
they're
in
a
state
where
nothing's
broken
in
travis
on
linux,
which
is
great
or
os
x
and
condeel.
A
Has
this
branch
let's
see
here
it
is
candeel,
has
a
branch
that
is
actually
running
and
passing
in
app
conveyor,
app
conveyor
being
a
windows
ci
system,
so
he's
actually
got
it
running
and
building.
This
is
a
new
big
core,
an
app
conveyor
and
publishing
artifacts
up
to
to
that
that
platform
and
he's
actually
got
its
creating
a
nuget
module,
which
I
don't
know
if
we're
going
to
use
or
not.
I
just
thought:
hey,
that's
that's
cool!
That's
impressive!
That
you
know
we're
very
hands
off
on
this.
A
Numenta
is
not
concerned
about
windows
build,
but
the
community
obviously
is
and
they're
putting
one
together.
The
thing
that
does
concern
me
about
this,
I'm
I
personally
am
happy
as
long
as
if
they
get
a
windows
build
running
as
long
as
it
continues
to
run
in
travis
and
osx,
and
linux
I'm
happy,
but
scrolling
all
the
way
back
up
here.
There's
a
lot
of
files
changed
here
and
they're,
not
just
insignificant
changes.
A
So
this
this
is
something
it's
that
needs
a
a
good
strong
review
from
new
pic
reviewers
to
figure
out
how
much
of
this
is
just
to
ensure
that
that
they're,
all
changes
that
make
sense
and
github
is
really
slow,
and
you
have
a
huge
pr
diff
like
this,
so
I
don't
think
it's
even
worth
scrolling
through
here,
but
some
of
it's
just
app
conveyor
configuration,
but
there's
a
lot
of
c
plus
changes
in
here
as
well.
A
So
this
is
something
I
I
can't
really
review.
I
need
someone
with
the
time
and
the
knowledge
of
c
plus
plus
to
to
run
through
and
make
sure
all
this
makes
sense.
I
mean
it's
not
something
because
it's
so
such
a
big
diff,
not
every
single
line,
doesn't
have
to
be
reviewed,
but
at
least
the
files
need
to
be
reviewed
to
make
sure
that,
like
is
this,
this
is
probably
something
that
was
just
added
for
some
library
that
needs
to
be
included
for
windows.
A
We
just
need
to
verify
that,
first
of
all
that
the
licenses
are
okay
for
us
to
use,
because
I'm
not
certain
about
about
all
the
licenses
and
that
these
additional
dependencies
that
are
used
in
the
windows
build
are
not
going
to
cause
any
other
problems,
and
I
mean
these:
this
is
being
stored
in
external
linux
32
I
mean
it
and
then
there's
a
win,
32
full
holder
in
here
I
mean
I'm
not
even
sure.
If
that's
the
right
place,
we
want
to
put
this
stuff
it
just
it
just.
A
It
needs
a
good
review
from
someone
before
we
can
merge
it.
So
that's
my
main
concern
with
this.
C
A
A
I
think
those
are
the
major
things
yeah.
Those
are
all
the
major
things
I
wanted
to
talk
about.
I'm
not
going
to
talk
much
about
future
plans,
because
I
think
we've
got
more
than
another
two
weeks
of
work,
just
within
the
prs
that
we
currently
have
open.
So
let's
just
focus
on
that
focus
on
near-term.
A
If,
if
you're
actively
working
on
new
pick,
keep
an
eye
on
what
prs
are
open,
you
can
see
all
of
them
here
on
our
issue
thing
just
by
picking
type
pull
request
and
that's
that's.
It
type
pull
request.
B
Sure
matt
we
were
thinking
about
whether
or
not
we're
getting
any
value
out
of
the
swig
or,
if
it's
just
more
effort
than
it's
worth
so
one
option
is
to
just
have
a
new
pick,
chord
read
the
c
plus
and
then
have
the
language.
Implementations
use
their
native
c
extension
support
to
interface
with
it,
which
would
be
pretty
similar.
What
we
do
with
swift
now
for
python,
at
least
so
it's
something
we're
going
to
kind
of
look
at.
I
think.
A
Okay,
that
makes
sense.
I
have
no
problem
with
that,
so
that
would
just
be
a
more
manual
solution
that
we
would
have
to
write
code
for
ourselves
instead
of
using
the
swig
interface
yep.
A
Okay,
that
sounds
fine
to
me.
We
probably
want
to
bring
it
up,
at
least
if
you
want
to
hear
what
other
people
want
to
want
to
say
about
it
on
nuka
hackers
or
something,
but
I
don't
have
any
experience
this
week,
so
I
don't
have
any
input.
A
Okay,
thanks
for
joining
us-
and
we
will
talk
again
in
another
two
weeks,
so
take
care
have
a
great
weekend
and
see
you
on
getter,
the
mailing
list
or
wherever
you
end
up
in
our
mini
communication
channels.