►
From YouTube: NuPIC Development Progress Review - Apr 3, 2015
Description
Bi-weekly progress review. I'll go over some of the milestone and label re-organization I've done over the past couple weeks.
A
Hello,
new
big
welcome
to
the
new
pic
development
progress
of
you
april
third
2015.
I
am
Matt
Taylor
and
in
a
new
meta
offices
we
have
some
music
engineers,
Cafferty
and
Austin
Marshall
Institute
I.
We
also
have
David
Ray
lead
on
HTML
java
coming
to
us
from
Chicago,
so
without
further
ado,
I'm
going
to
share
my
screen
and
we're
going
to
go
review
some
tickets
and
stuff
okay.
A
So
we
had
a
little
bit
of
github
instability
over
the
past
week
or
two
because
of
the
denial
of
service
attacks,
so
that
broke
our
pipelines
up
a
little
bit.
That
is
the
reason
why
regression
tests
were
failing
for
a
while,
but
it
looks
like
from
the
new
pic
side.
Things
are
working
so
pipe.
Our
CI
stuff
is
green
at
the
moment
and
I
do
want
to
look
over
open
pr's,
because
there's
there's
a
ton
right
now.
A
A
So
we'll
see,
if
that's
label
is
useful
and
continue
using
it
as
we
go
along,
but
I'm
not
going
to
go
over
all
of
these
PRS,
because
I
have
commented
on
them
in
the
past
couple
of
days
trying
to
get
some
more
movement
on
them.
There
is
a
lot
of
ongoing
work
currently,
especially
on
encoders
for
Captain
proto,
serialization
config.
So
that's
what
these
all
these
bottom
ones
are
about
a
lot
of
those
merge.
A
So
let's
jump
to
the
roadmap
up
here.
I
I
brought
up
a
conversation
last
week
or
the
week
before
about
changing
the
roadmap,
changing
the
milestone.
Specifically,
we
talked
about
this
at
our
last
review
meeting
to
be
less
topical
and
more
chronological,
so
I've
done
that
I've
updated
all
the
labels
on
all
of
our
repos
and
and
the
milestones
so
there's
now
only
three
milestones-
and
this
is
the
proposal
I
put
on
the
Packers
mailing
list
and
got
some
suggestions
from
mark
and
Scott
about
it.
A
So
there's
a
bug
fixed
milestone,
which
currently
has
two
issues
in
it
that
are
both
assigned
to
me.
One
of
them-
I,
don't
know
if
it's
I,
don't
think
it's
our
problem,
I
think
it's
a
user
problem,
but
I'm
still
waiting
for
response,
the
other
one
I
am
trying
to
get
a
decent
installation
of
fedora
on
my
VMS,
not
cooperating
with
me
in
virtualbox,
so
I'm
still
working
on
that,
and
then
we
have
the
current
development
milestone,
which
is
the
next
minor
release.
Just
as
a
reminder,
we
are
on
021
once
I
I
clean
up.
A
This
fedora
bug
will
I'll
release
022
and
all
current
development
is
going
on
in
030.
So
you
can
see
all
of
these
can't
produce
serialization
tickets
that
Austin
and
Scott
have
been
working
diligently
on.
So
thanks
guys
for
that,
that's
good.
Some
of
them
are
tagged
with
Help
Wanted.
So
you
can
see
a
bunch
of
these
have
been
cleared
out
already.
A
I
do
have
I
still
have
this
as
a
removal
of
dead
code
from
from
New
pick
tickets.
That
David
has
been
working
on,
so
these
are
still.
They
still
have
TRS
that
are
under
review
kind
of
still
going
back
and
forth
with
those
and
then
the
windows
stuff.
This
is
still
my
fault.
I've
not
made
any
progress
with
the
windows
stuff
it
just
it's
just
going
to
take
me
going
going
heads
down
and
turning
things
off
a
day
or
two
to
get
to
get
through
these.
A
There
is
a
another
ongoing
ticket
of
code
being
removed
from
the
decor.
These
are
just
cleanup
tasks
that
mostly
David
ragazzi
and
Mark
have
been
working
on,
so
those
are
in
the
current
minor
milestone,
just
because
it's
been
a
lot
of
ongoing
work
on
that
and
the
odd
one
remaining
already
talked
about
this
at
Bayer
integration
and
new
big
pipeline,
and
we
still
have
a
bunch
of
one-off
issues
that
are
currently
either
being
more
tour.
We
actually
some
of
these
I
need
to
clean
up
I.
A
A
A
Yeah
and
last
milestone
that
that
we
have
is
a
1
point
0,
and
this
is
a
capturing.
What
we
want
to
have
done
by
one
point.
Oh
I
meant
to
have
this
open
here,
but
I
did
send
out
a
survey
this
morning
or
nose
yesterday,
I'm
going
to
find
it
here.
It
is
and
let's
see
where
we
got
13
people.
Oh,
that's
not
what
I
wanted
want
to
see
the
live
form.
Oh
shoot,
there's
one
one
place
I
can
look
and
see
like
a
nice
bar
chart
or
a
pie
chart
for
this
eve.
A
You
voted
for
what
I
guess:
I
I,
don't
know
how
to
do
it
here,
but
well.
I
should
have
come
over
prepared,
I
had
it
open
last
time
so
anyway,
I'll
tally
those
up
and
report
them
on
the
mailing
list.
Once
I
once
I
find
the
the
right
link
to
it
all,
but
I
was
asking
a
asking
questions
about
what
we
think
should
be
included
in
the
one
point:
0
release
no
be
upgrade
the
model
serialization,
which
I
think
is
pretty
important.
A
I
think
most,
the
other
people
did
too
windows
compatibility
and
refactoring
swarming,
a
new
standalone
library,
so
once
we
give
this
a
week
or
so
for
people
to
respond,
oh
I'll
come
back
and
publish
those
results
on
the
mailing
list
and
that
will
inform
of
deciding
what
other
things
are
going
to
go
into
this
about
1.0
release
so
that
this
new
road
maps
name
is
basically
just
bug-fix
release.
Next
minor
release
next
major
release.
So
that's
the
thinking
behind
it
at
the
moment.
A
A
A
C
Well,
mostly
I've
been
working
on
the
network.
Api
there's
some
work,
I've
been
doing
for
cortical
for
cortical
I
owe
to
ease
kind
of
some
legacy
frameworks
and
to
try
to
transpose
the
network.
Api
work
I've
done
into
an
older
version
of
Java
and
there's
some
lessons.
I've
learned
from
that,
so
I'm
going
to
port
those
back
into
JDK
8
to
the
HTM.
C
You
know
to
the
new
mentor
version,
so
that's
kind
of
what
I'm
doing
right
now
and
I'm
going
to
shooting
for
the
end
of
April
to
have
that
done,
at
least
by
the
end
of
april,
so
that
there
will
be
a
month
between
the
time
people
get
to
review
and
work
with
the
first
version
and
the
first
hand
the
New
York
hackathon.
So
hopefully
we
might
get
some
people
using
it
during
the
hackathon
and
they'll.
Be
cash
rewards.
If
you
do
not
just
joking.
C
Anyway
and
I
have
we
have
a
community
member
right
now
aunty
data
who
is
working
on
sprucing
up
the
current
code
base
with
the
slf4j
logging
facility
that
we
just
put
in
you
HTM
de
chavez,
so
auntie
date
is
he's
kind
of
the
only
community
member
right
now,
who's
working
on
the
people
come
in
and
spurts.
So
he's
like
sign
me
up.
He
wants
all
of
the
all
of
the
issues
just
be
assigned
to
him,
so
oh
I
told
about
it.
C
I
would
only
give
him
one
at
a
time,
but
but
anyway,
so
he's
been
working
on
that
he
did
his
first
submission
last
week
and
things
look
pretty
good
with
that.
So
that's
about
it
right
now,
so
just
just
a
little
bit
of
community
involvement
and
working
on
the
network
API
and
which
I
want
to
get
done,
probably
definitely
by
end
of
April.
C
But
you
know
the
next
couple
of
weeks
and
I
saw
a
man
that
you
sent
a
you
responding
to
my
email
about
how
to
kind
of
integrate
the
work
back
into
the
main
trunk,
because
it's
going
to
be
a
huge
body,
you
know
I'm.
I
think
I
can
get
it
into
the
main
trunk
in
little
sections
that
you
can
test
independently.
So
I'm
going
to
try
and
do
it
like
that
that.
A
A
C
A
A
B
B
B
C
And
I
can
I
also
offer
a
comment.
The
meeting
there
was
some
mention
of
the
random
number
generator
work
and
zumba
tai
and
I
like
last
October,
went
back
and
forth
about
this
kind
of
because,
although
Mersenne
twister
is
the
implemented,
the
common
algorithm
implemented
in
Python
c++
and
now
there's
well,
there's
also
a
separate
library
in
Java.
There
are
some
fundamental
differences,
even
even
minor
things
like
the
floating-point
floating-point
resolution
is,
can
throw
things
off.
What
I'm
saying
is
is
that
the.
C
It's
it's
not
really
dependable,
even
though
they're
using
the
same
algorithm
is
not
really
dependable
to
to
decide.
Okay,
we
can.
We
can
use
this
rng,
give
it
the
same
seed
and
hopefully
come
out
with
the
same
thing,
which
is
why
I
suggested
a
facade
kind
of
treatment
where
we
run
it
in
Python
record.
C
The
numbers
that
are
generated
then
have
a
separate
file
that
can
be
loaded
that
can
be
loaded
from
a
rng
facade
and
can
just
spit
out
those
numbers,
because
they're
all
they're,
all
floating
point
numbers
and
you
know
they're
accessed
in
sequence.
So
when
you
run
the
test,
there's
a
list
of
numbers
that
are
going
to
be
generated
and
those
numbers
can
be
loaded
in
so
I
mean
that
was
my
suggest.
I,
don't
know
how
else
really
to
come
about
doing
it,
because
it
is
really.
A
B
A
B
C
B
C
So
you
know
eventually
when
we
do
get
other
ports
or
you
or
you
get
clients
which
will
hook
into
the
C++
version
or
something
like
that
mean
you
want
to
offer
different
platforms
for
that.
We
might
want
to
look
into
having
a
compliance
kind
of
kit.
You
know
it
Matt
you,
the
one
who
came
up
with
you,
I
do
so
I.
I.
A
Right:
okay,
okay,
okay,
let
me
share
my
local.
Have
one
last
thing
here:
going
back
to
the
milestone
and
label
things
that
I
was
talking
about,
the
labels
specifically
are
all
sort
of
different
I've
got
a
standard
standard
here
where
there's
word
colon
word
and
the
first
word
is
kind
of
the
type
of
label.
So,
for
example,
I've
got
priorities
as
priority
colon
and
a
number
status
status.
A
Colon
block
help
wanted
in
progress,
ready
or
won't
fix,
and
then
type
is
a
big
one
and
that's
really
sort
of
type
of
issue
that
we're
creating
is
it
represents,
see,
there's
also
a
cleanup
issue
of
a
build
problem
and
an
actual
bug,
administrative
tasks
etc.
So
I'm
hoping
that
when
you're
creating
issues
instead
of
having
this
huge
list
to
go
through
and
just
kind
of
point
and
peck,
you
can
just
type
okay,
the
subjective.
This
is
the
decoders
right,
and
this
is
a
type.
This
is
an
encoder
bug
right,
no.
A
It
should
be
easier
to
label
things
and
it'll
also
be
easier
for
me
for
tooling
to
create
filters
this
way,
if
I,
if
I
decide
to
do
that
so
and
I
think
the
sort
of
came
out
of
a
marks
comment
on
on
the
mailing
list.
So
thanks
for
that
input
mark,
but
I
think
this
should
work
out
pretty
well
hope
it
does
I'm
still
totally
open
to
comments
and
suggestions.
None
of
this
stuff
is
set
in
stone,
of
course,
so
yeah,
that's
that
that's
all
I
got
on
on
my
end
anything
else
at
all.