►
From YouTube: Weekly Sync 2020-11-17
Description
Meeting Minutes: https://docs.google.com/document/d/16u9Tev3O0CcUDe2nfikHmrO3Xnd4ASJ45myFgQLpvzM/edit?pli=1#heading=h.auz3t2zhfizp
B
John
remember:
I
sent
something
on
the
github
chat
a
while
back
about
the
sources
error.
I
tried
to
fix
it
using
the
imports
that
you
would
send
and
I'm
still
getting
the
same
error.
A
This
error
here-
anomaly,
detection
name
sources-
is
not
defined.
B
A
A
B
A
A
Yeah
this
one
okay,
this
is
not
the
master
version
of
the
docs,
so
I
think
what
happened?
Is
you
copy?
I
think
maybe
you
loaded
the
page
once
and
then
you
loaded
the
other
page,
because
this
one
see
this
one
imports
sources.
So
I
think
I
think
what
happened.
Was
you
copied
the
imports
from
the
master
version
and
the.
B
A
And
yeah
and
the
rest
of
the
code
from
the
latest
release
so
because,
if
because
the
this
version,
which
is
the
latest
release
this,
this
shows
that
we're
importing
sources.
But
if
you
go
to
the
docs
for
the
master
branch,
it
shows.
A
Yeah
well
also,
I'm
not
sure
if
the
rest
of
these
are
going
to
be
the
same,
because,
let's
see
look
at
the
rest
of
your
code
here,
yeah
well,
the
rest
of
your
code
is
probably
following
the
following:
the
I
mean
sorry,
the
methods
can
we
see
the
the
training.
B
A
B
B
Have
it
yeah
0.3.7.
B
A
Okay,
so
yeah
okay,
so
it's
installed
from
the
gate,
repo
so
and
we
should
probably
okay
yeah.
We
should
probably
have
it
say
something
when
it's
installed:
okay,
so
yeah
you're
running,
you
should
be
running
off
of
the
master
branch,
in
which
case
you're
going
to
want
to
make
sure
that
you're
running
you're
doing
the
and
I'll
paste
this
in
that
in
the
median
here,
so
you're
going
to
want
to
make
sure
that
you're
running
off
of
I
just
pasted
it
in
the
into
the
chat
on
the
meat
call.
B
B
Soil
clone
that
repository.
That's
fine.
A
Well,
no,
you
have
it
you've
got
it
you've
got
it.
You've
got
it
correctly.
I'm
just
saying
I'm
saying
you're
going
to
want
to
so
click
on
this
link
that
I
just
sent.
A
Yeah
so
yeah
click
on
that
right
and
so
you'll
notice
that
this
this
url
it
says
df
dfml,
master
right,
and
so
this
is
basically,
if
you're
working
off
of
the
the
version.
If
you're
working
off
of
the
like,
if
you've
installed
the
ffml
using
the
git
repo
you're
going
to
want
to
make
sure
that
you're
on
this
version
of
the
documentation,
the
one
that
has
the
dfml
slash
master
and
so,
if
you
scroll
down
a
little
bit
yeah.
So
this
is
this
tutorial
has
been
slightly
updated
and
it
looks
like
when
you
copy.
A
That
said
colon
sources,
context
which
tells
me
that
when
you
were
originally
writing
the
the
method
body,
as
you
were
looking
at
this
document,
because
this
says
colon
socialist
content,
so
you're
going
to
want
to
make
sure
you
have
that
train
plot.
That
block
that
you
just
copied,
because
that
corresponds
to
the
code
that
you
have
checked
out.
B
So
yeah
so
right
now
what's
here,
should
I
rewrite
my
code
using
the
one.
A
B
A
B
Sources:
sources,
comma
sources,
context.
A
Well,
so,
okay,
so
one
second
here,
so
this
code
block
here
that
you
have
copied
you
copied
it
from
the
latest
released
version
right.
That
was
the
the
other
tutorial
that
we
were
looking
at
before
I
sent
you
the
link
right
now.
I
think
what
you
need.
What
you
need
to
do
is
you
need
to
take
that
so
go
back
to
the
tutorial.
We
were
just
looking
at.
A
Right-
and
so
this
is,
this
is
the
version
of
the
tutorial
for
the
master
branch,
which
is
whenever
you
wrote
the
train
method.
You
were
looking
at
this
right,
but
when
you
went
back
and
you
copy
pasted
the
imports
you
were
looking
at,
the
latest
release
version
right.
So
what
we
need
to
do
is
we
need
to
look
at
the
imports
block
from
this
version
of
the
documentation
and
just
copy
paste
that
block
just
because
there
may
be
other
things
that
changed
in
there
right.
A
So
we'll
just
replace
yeah
so
use
this
and
then
replace
the
stuff
that
you
have
in
your
file
with
this
and
then
you
should
be
good
to
go.
B
B
A
A
B
A
B
A
Okay,
okay,
so
yeah
it
looks
like
you
may
have
also
so
that
was
in
so
now
the
now
you're.
Looking
at
the
error.
A
Thank
you
very
much,
so
this
is
in
the
the
the
accuracy
method
now
yeah
so
and
that
may
have
been
one
of
those
things
where
yeah.
Okay,
all
right.
A
Predict,
okay-
and
I
think
this
is
this-
is
another
one
so
that
the
train
method
got
copied
from
the
new
tutorial
and
the
accuracy
and
predict
methods
got
copied
from
the
old
version
of
the
tutorial
when
you,
when
you
did
this
so
so,
we'll
just
need
to
just
reference.
The
new
version
of
the
tutorial
here.
A
A
Yes,
this
is
the
new
version
with
it
when
you
see
slash
master
in
there.
That
means
you're
working
off
of
the
master
branch,
and
so,
if
you
have
the
the
checked
out
version
of
dffml,
then
you're
working
off
of
the
master
branch,
if
you're
installing
it
from
from
pi
pi
rather
than
installing
it
you
know
from
the
git
repo,
then
then
you
use
the
other
version.
B
B
I'll
try
and
rewrite
the
part
of
my
code,
that's
causing
it
and
hopefully.
A
B
A
All
right
great,
so,
let's
see,
let
me
just
write
this
stuff
down.
A
On
that,
so
we
need
to
have
clear
documentation
on
the
difference
between
the
released
version
of
the
docs
and
the
slash
master
version
of.
A
A
Yeah
right,
so
what
we
need
is
actually
that's
actually
what
we
want
to
happen,
yeah,
which
is
good,
because
that
would
be
bad
if
it
wasn't,
but
what
we
need
is,
let's
see,
so
we
probably
need
this
somewhere
in
contributing,
maybe
somewhere
like.
I
don't
know
where
what
do
you
guys
think
maybe
in
like
getting
set
up
here,
or
this
is
more
like
what
you
need
to
do-
I'm
not
sure
where
this
might
go.
Actually
it
might
be
sort
of
right
under
contributing
right
here
or
yeah.
This
isn't
ideal,
there
would
be.
A
A
This
is
the
time
right
here
where
now
they
go
as
soon
as
you've
done
this
command.
These
commands
now
you're,
not
looking
at
you
know,
you're,
not
looking
at
the
main
version
of
the
docs.
Now
you
need
to
be
looking
at
the
master
branch
version
of
the
docs,
because
now
you're
sort
of
with
the
rest
of
the
development
rate
is.
A
Or
is
there
a
better
time
like
I
don't
I
mean,
I
know
the
contributing
docs
kind
of
are
a
little
convoluted
but
yeah,
where,
where
would
be
the
best
place
for
people
to
know
to
get
this
information?
Does
anybody
have
any
suggestions
other
than
here?
I.
A
Yeah,
that's,
I
think
I
think
that's
pretty
much.
That's
the
only
place
I
can
think
of
where
you're
you
know
where,
if
someone's,
if
someone's
looking
at
this
documentation,
then
that's
pretty
much
where,
where
that
dividing
line
is
so
so,
let's
see
mention
this
after
the
get
g
around
the
git
clone
on
the
contributing
oops
getting
started
to
work
onto
your
from
all
page,
and
this
would
be
what
is
this
called
devon
bye.
E
C
C
So
isn't
there
a
badge
that
looks
good
up
there
from
there
for
the
documentation?
Sorry,
what
is
isn't
there?
A
badge
for.
A
The
documentation
there's
not
a
badge
for
the
documentation,
no
there's
the
tests,
the
coverage,
there's
core
infrastructure,
getter
and
then
pi
pi,
I
I
don't
know
would
yeah.
I
mean
I'll
look
for
a
badge.
If
I
find
one.
A
A
I
think
I
think
you
mean
ooh
yeah,
it
might
be
good,
oh
actually,
you
know
what
would
be
bad
would
tell
people
is
that
we
could
change.
A
To
be
like
you
know,
as
soon
as
we
we
finish,
we
do
a
release,
we
change
the
version
number
to
like
hyphen
the
commit
or
something
I
think
we
can
do
yeah.
I
think.
C
A
C
A
Okay
yeah:
let's
do
that
because
that's
yeah,
that's
a
good!
That's
a
good
way
of
doing
that
so
need
to
make
sure
we
have
clear.
A
Identification
of
what
version
of
the
docs
we're
on
so
in
the
sidebar,
it
should
say
x.y.z,
at
least
for
the
released
version
for
dev
version,
it'll
be
x,
dot,
y
dot
c
and
then
the
short
git
commit
sha.
A
That
should
that
should
keep
us
all,
at
least
at
least
at
least
it'll,
be
slight
they'll,
be
somewhere
other
than
the
url
bar,
because
yeah
I've,
I've
often
found
myself
in
the
wrong
version
as
well.
So
it'd
be
good
to
have
something
that
that
points
that
out.
A
B
A
A
Okay,
cool
and
okay
yeah,
so
all
right,
great,
so
you're
you've
faced
an
issue.
We've
fixed
that
we'll
write
some
test
cases
and
try
to
submit
a
pr
by
this
weekend,
great.
A
The
config
part,
like
those
structures
and
what
all
that
is.
B
B
There's
there's
some
really
clear
documentation
for
record
and
a
couple
of
other
modules.
That's.
A
A
good
point
yeah,
I
don't
think
we
have
that
so
let's
see
documentation
on
config
so
and
do
you
have
any
questions
in
particular
right
now,.
B
Yeah
so
remember,
I
told
you
in
a
last
meeting
that
I
wanted
to
allow
users
to
choose
the
size
of
their
validation
set,
like
I
I'd
put
a
10
of
the
total
training
set
by
default,
but
in
the
future,
I'd
like
to
allow
users
to
change
that.
So
I
was
talking
to
saksham
and
he
told
me
that
it's
possible
to
do
that.
Config
and
I
was
just
wondering
how.
A
Okay,
let's
see
oh
yeah,
I
remember
this.
So
let's
see
okay,
I
did.
A
Yeah,
so
so,
and
and
what
I
mean
basically
all
you
so
so,
okay,
so
yeah,
we
need
we'll
need
to
have
a
tutorial
on
config,
so
we
need
to
make
a
tutorial
on
config,
so
essentially,
just
as
a
at
a
high
level
classes
at
a
high
level.
The
config
stuff
is
a
wrapper
around
data
classes
and
their
class
fields.
So
you're
really
you're
really
probably
interested
in
this
where's.
The
stamp
field
thing
oh
yeah
here,
yeah.
A
Yeah,
no,
we
needed
a
separate
tutorial.
It's
we've,
we've
sort
of
we've.
We've
we've
done
some
nasty
things
to
data
classes,
so
it's
it's
not
similar
enough
really,
but
we'll
so
we'll
make
a
tutorial,
but
but
for
as
far
as
how
you
would
do
that,
let's
see,
I
think
we
can
come
up
with
a
quick
example
here.
For
you
I
mean
it
really
just
boils
down
to
like.
If
you
want
another
field
like
if
you
want
something
configurable,
you
are
actually
here's.
How
do
I
best
explain
this?
A
Okay,
so
from
general,
am
I
sharing
my
screen?
Can
you
guys
see
my
screen?
Can
everyone
else
see
my
screen.
A
Okay,
all
right,
so
yes,
I
am
sharing
my
screen.
Maybe
here
I'll
try
to
stop
and
or
wait.
No,
it
says
it's
paused,
okay.
It
was
probably
a
still
image.
Sorry,
okay!
I
don't
know
why
it
does
that
all
right,
but
okay,
so
here
from
a
more
general
perspective
of
of
okay,
we
don't
have
a
tutorial,
but
from
a
more
general
perspective
of
how
how
would
one
figure
out
what
to
do
with
this?
A
If,
if
there's
no
tutorial
right
until
we
can
write
one,
I
would
go
into
your
dffml
source
code
and
do
a
get
grep
and
then
look
for,
like
you
know,
whatever
you're
looking
for
right
here
and
so
in
this
case,
you're
looking
for
you're
looking
for
things
that
that
have
to
do
with
with
config
right
the
act,
config
decorator,
so
I'd
put,
you
know,
grep
and
show
me
10
lines
after
is
capital
a
anything
that
says
at
config,
and
so
then
I
now
I'm
presented
with
a
bunch
of
examples,
essentially
right
and
actually
we
might
make
that
a
little
bit
longer
and
we'll
make
it
like
25..
A
So
now,
essentially,
I'm
looking
through
all
of
the
dlfml
source
code
and
I'm
looking
at
all
of
the
times.
This
was
used
right.
So
we
can
kind
of
use
this
as
as
examples
right
for
how
we
might.
We
might
do
different
things
and
the
first
thing.
A
Here's
here's
one
thing
to
know,
but
the
first
thing
were
seen
right
here
is
that
you
can,
when
you're,
using
types
that
are
non-primitive
data
types
like
or
non
things,
things
that
are
possibly
mutable
like
a
list
or
a
class
or
something
you're
going
to
actually
need
to
use
this
default
factory
and
and
you'll
see
a
few
of
these,
but
you
basically
just
say,
define
this
empty
function,
which
is
a
lambda
and
it
returns
an
instance
of
the
thing
that
you
want
and
the
reason
that
this
happens
is
because,
if
you
were
to
set
the
default
to
this
clock
to
a
to
an
empty
list,
then
every
single
time
this
an
instance
of
this
field
was
created.
A
It
would
be
the
same,
it
would
point
to
the
same
empty
list
and
then
with
the
default
factory.
It
creates
a
new
list
every
time
and
the
next
thing
is
obviously
okay,
so
default
values
right.
So
if
you
wanted
a
default
value,
you
just
say
default
equals
whatever
it
is.
Unless
you
have
this
case,
where
you're
using
something
that's
mutable,
then
you
do
a
default
factory
and
you
you'll
find
you'll
find
lots
of
examples
here.
This
is
your
non
primitive
data
type
state,
yeah,
non-pro,
well,.
A
B
I
understand
what
primitive
and
non-narrative
data
types
are.
The
empty
list
part
that's
just
for
the
non-primitive
data
types
right,
yes,
say
I'm
working
with
something:
that's
an
integer
or
a
float.
A
Yeah,
so
basically
the
the
yeah
exactly
so,
if
you're
working
with
something
that
doesn't
have
properties
attached
to
it.
Right,
if
you
can
say
if
yeah
well-
okay,
that's
not
necessarily
true,
but
if
you're
looking
at
like
basically
integer
float
string,
bytes
there's
a
couple
other
ones
like
I
think
a
tuple
is
also
immutable,
but
those
those
types
of
things
you
can
set
as
default,
and
it's
gonna,
it's
gonna.
It's
gonna,
throw
you
an
error.
A
If
you,
if
you
do
the
wrong
thing,
the
only
thing
it
won't
throw
you
an
error
for
is,
if
you
set
it
to
be
an
instance
of
a
class
or
a
class
yeah
an
instance
of
a
class,
because
it
just
doesn't
know
how
to
check
for
that.
So,
if
you're
thinking
about
setting
some
something's
default
to
a
class,
you
have
to
do
default
factory,
but
otherwise
you
just
use
default
right.
A
And
so
in
this
case,
if
this
is
some
kind
of
integer
value
or
a
float
value
that
you're
using
here,
like
maybe
a
float
for
a
percentage
right
of
a
split,
then
you're
going
to
use,
then
you're
just
going
to
say
default
right
and
the
way
that
this
works.
Is
you
specify,
so
you
don't
have
to
necessarily
specify
field,
you
can
just
say
the
config
and
then
the
type
equals
the
default
value.
But
we
really
would
like
to
say
fields
here,
because
the
field
is
going
to
be.
A
The
first
argument
to
the
field
is
always
the
help
text.
So
what
is
what
is
this
thing?
Give
it
a
description
so
that
the
rest
of
us
know
what
this
is
supposed
to
do,
because
that
help
text
is
actually
going
to
show
up,
and
let
me
find
some
worse
where.
A
This
is
just
yeah,
get
grip
and,
and
if
you're
not
familiar
with
the
rep,
this
is
basically
gonna
become
your
new
best
friend,
because
this
this
is
just
the
best
thing
ever.
If
you're,
looking
at
a
code
base
and
you're
trying
to
figure
out
what's
going
on
in
there
gitgrep
is,
is
a
very
good
way
to
do
that.
It
searches
through
all
the
files
that
have
been
committed
into
the
code
base
so
and
I'll
put
this
in
the
media
minutes.
So.
G
Get
crap
for
a
better
understanding.
You
can
also
show
the
help
command
like
a
dffml
edit
records
dash
h.
That
will.
G
A
Yeah
so,
and-
and
this
is
this
is
so
this
this
works
for
some
some
things.
This
ends
up
showing
like
this
some
things
it
does
not.
So
this
is
a
good
example
of
something
that
it
does
show
from.
The
command
line
is
the
command
line,
so,
basically
all
of
the
command.
The
command
line,
commands
are
use
config
and
then
all
of
the
classes
use
config,
and
the
idea
is
that
if
everything
uses
the
same
way
of
configuration,
then
you
can
configure
things
through
the
command
line.
A
You
can
configure
things
through
the
http
interface
or
you
can
configure
things.
You
know
as
regular
python
code,
and
so
this
is
an
example
of
what
happens
when
you
know
we
we
did
the
edit
command
and,
let's
see
where
was
the
edit
command.
A
I
think
it
was
towards
the
top
of
this
right
yeah,
so
the
edit
command
here
right,
so
dataflow
config
features
sources
right
and
now
we
do
dash
h,
and
we
see
that
this
is
the
help
text
right,
that's
being
displayed
here
and
then
it
shows
us
what
the
default
values
are
and
then,
if
you
go,
for
example,
to
the
to
the
documentation-
and
you
were
to
look
at
the
plugins
page-
and
you
looked
at
you
know,
the
yeah,
like
the
the
json
or
the
csv
source
here,
you'll
see
that
all
of
the
configuration
parameters
are
documented
here
and
like
let's
see
we'll
go
to
different
master
plugins
model.
A
Yeah,
the
edit
is
in
the
command
line,
so
command
line
ones
will
show
up
correctly
here,
but
you
can't
really
say
like,
and
actually
I
think
we
used
to
be.
This
was
a
long
time
ago.
This
is
my
list
models.
A
Does
it
still
work
hold
command?
Yes,
it
does
still
work.
Okay,
yeah,
it
just
prints
their
docs,
though
okay
we
haven't
this.
Is
this
hasn't
been
updated
in
years
now?
So,
let's
see
so,
but
you
can
see
it
so
so
the
essentially
the
point
of
using
the
config
structures
is
that
you
can
display
it
anywhere
or
parse
it
in
coming
from
any
different
kind
of
location,
right
command
line,
or
so
you
know
some
kind
of
json
or
something
or
just
as
a
regular.
You
know
when
you're
instantiating
these
these
these.
A
When
you're
instantiating
this
stuff,
you
can
just
pass
pass
it
as
you
would
to
a
normal
python.
You
know
object,
and
so
let's
see
yeah,
and
then
this
is
in
the
docs.
You
can
see.
This
is
the
hell
messages
here
so
that
it's
all
you
know,
sort
of
sort
of
displayed,
and
actually
I
was
thinking
we
should
probably
make
this.
This
is
a
side
note,
but
we
should
we
should.
A
We
can
take
this
and
and
turn
it
into
regular,
numpy
doc
format
and
that
way
it'll
show
up,
as
it
usually
does
in
under
the
api.
You
know
like
under
regular
arguments
that
you'd
see,
for
example,
yeah
something
you
know:
it'll
be
format.
We
can
make
it
so
it's
formatted
like
this,
but
that's
just
sort
of
a
side
side.
Note,
let's
see,
does
that
sort
of
give
you
does
that
give
you
does
that?
Give
you
enough
information
there:
okay,
cool,
yeah,
okay,
okay!
A
So,
let's
see
we
should
also
figure
out
how
to
convert
configs
into.
We
should
figure
out
how
to
convert
that
configs
into
num
pi.
A
A
All
right,
so,
let's
move
on
all
right
so
and
we
got
a
little
sidetrack
from
what
we're,
what
we're
usually
supposed
to
do
there,
but
that
is
good.
Okay,
so
documentation
config
all
right!
So
let's
go
through.
Let's
go
through
everybody
else,
real
real,
quick
here
and
and.
A
Okay,
let
me
just
copy.
A
F
On
what
were
you
just
saying
like
what
what
we
are
gonna
miss?
I
was
about.
A
Oh,
let's
because
seagull
has
to
go
so
I
just
wanted
to
check
in
and
let's
see
yeah
I
just
want
to
see.
Is
there
anything
so
there
you
had
issues
with
the
the
pip
install
command.
Is
there
anything
you're
blocked
on
at
the
moment
that
we
should
circle
back
with
you
and
and
update
you
to
unblock
you
or.
A
Oh
and
I
was
going
to
give-
I
was
going
to
update
the
the
auto
sk
learn
docs,
because
you
were
thinking
about
maybe
using
something
that
that
was
that
was
auto
ml.
So
I'll
try
to
try
to
do
that,
all
right
cool!
I
don't!
I
just
wanted
to
make
sure
that
we
we
checked
in
with
you
if
you
had
to
go
all
right.
Well,
thanks
for
stopping.
D
A
Yep
see
ya,
have
a
good
one
all
right!
So,
let's
see
we're
gonna
update
the
audio
next.
So
thinking
about
using
auto
sk
learn
for.
A
I
can't
remember
what
I'm
just
gonna
use
that
for
so
think
about
using
auto
sk
learn,
so
we
need
to
update
the
docs,
because
this
is
another
one
that
didn't
get
toxic.
Oh,
that
was
part
of
that
test,
so
it
needs
also
tested.
A
Docs-
and
I
think
that
was
a
part
of
this
issue
here-.
A
C
Yeah,
so
if
you
look
at
the
issue
just
a
second
892.,
so
I
have
completed
most
of
them
yeah
for
transformers
the
issue
was
the
download
loop.
We
talked
about
earlier,
oh
yeah,
that's
right!
So
I
couldn't
run
those
tests
locally.
E
C
And
we
haven't
tested
them
for
windows:
okay,
okay,
for
pytorch;
nothing
is
working.
Okay,
that
might
just
be
the
version
issue
yeah,
and
it
just
has
this
weird
tendency
to
repeat
the
tests
somehow
ndf
and
by
torture.
If
you
see
the
logs
for
pythons,
I've
pasted
them
down
below
it
just
starts
the
test,
then
pauses
and
then
I
don't
know.
Oh.
A
C
A
C
A
A
Oh
is
this:
this
is
probably
because
I
just
updated
the
spacey
model
model,
spacey
dffml
model
spacey,
because
you
you
pulled
recent
changes
right.
A
A
C
A
Okay,
great
okay,
yeah,
so
I
think
we
can
toss
this
test
ner
integration,
because
this
is
the
same
stuff
that
is
getting
done
by
the
that
it's
getting
done
by
the
console
test
stuff
now,
so
we
can
probably
just
drop
that
that
test
case
and
the
related
or
let's
see,
yeah.
Okay,
it's
doing
oh!
No!
This
is
a
separate
thing,
but
now
this
is
a
separate
thing.
A
A
Yeah,
okay,
so
that's
good
to
know,
let's
see,
and
we
did
we
get
this
down
in
the
in
the
so,
let's
see
so.
A
So
space,
let's
run
failing
because
of
tests
of
calls
to
bash
scripts.
This
is
not
the
same
as
the
doc
doc
test
or
the
console
test
for
the
doc
string.
C
A
A
A
Oh
yeah,
the
back
ticks,
oh
yeah,
okay,
so
that's
something
where
that's
a
good
point,
so
I
think
that's
something
where,
because
I
I
looked
this
up
and
I
found
out
that
it's
actually
the
up
arrow,
that's
the
the
multi-line
separator
on
windows
rather
than
the
backslash
for
for
linux
and
bash
and
stuff.
So
we
may
need
to
we.
This
may
be
one
of
those
things
where
we
have
that
group
tab
thing
and
we
just
change
it
or
it
may
be.
A
I
think
I
think
what
we
may
want
to
do
is
we
may
want
to
modify
this,
maybe
a
sphynx
thing
where
we
go
in
and
we
say
when
we
add
this
stuff,
programmatically
yeah,
I'm
not
sure,
I'm
not
sure
this
is
yeah.
This
is
so
okay.
So,
just
just
to
recap:
where
I'm?
What
I'm
thinking
here
is
like
you're
saying
so
on
windows
you,
you
can't
paste
in
these
commands,
because
backslash
doesn't
work
as
to
separate
lines.
I
remember
it
being
that
way.
I
can
try
it.
A
A
A
So
it
is,
this
is
true
yeah.
So
this
is.
This
is
the
one
on
windows.
Is
this
little
up
carrot
and
so
yeah?
Okay,
so
we're
gonna
need
to
and
this
these
are.
These
are
obviously
oh
look.
It
jumps
the
sidebar
there.
That's
weird
okay
yeah.
So
this
is
one
of
those
things
where
I
was
saying
like
I'm
not
sure
how
we're
going
to
do
this
when
we
do
move
to
console
like
with
the
console,
so
the
console
test
plugin
will
need
some
way
of
knowing.
A
Should
I
run
this
code
block
on
linux,
or
should
I
run
this
code
block
on
windows?
Actually,
okay,
wait!
Wait!
Sorry,
one
step
back,
so
the
console
test,
plugin
parses,
those
backslashes
and
it
treats
those
it.
It
actually
parses
in
that
command
and
then
figures
out
what
like
it'll
run
each
so
the
console
test
plug-in
runs
each
one.
A
Removes
the
backslide
yeah,
it
essentially
removes
the
backslashes
and
and
it
but
it,
and
then
it
runs
everything
as
a
subprocess
right,
so
it
doesn't
actually
call
out
to
bash
or
whatever
the
shell
is
it
it.
Basically,
it's
like
a
it's
its
own
sort
of
shell.
It's
like
it's!
It's
a
light
implementation
of
a
shell
because
it
supports
pipes.
C
A
Other
things
like
that
and
okay,
so
basically
what
we
could
do,
and
so
then,
what
we
could
do
is
we
could
basically
say:
okay
for
things
that
are
not
in
a
group
like
the
problem,
is
that
every
single
example
that
has
backslashes
in
it
will
need
to
be
changed
to
carrots
to
up
carrots
for
windows
right.
So
we
probably
want
to
do
that
programmatically,
somehow
right
and
have
it
be
like
right?
We
probably
want
to
have
something:
that's
like
a
swinx
extension
or
a
modification
to
our
existing
sphinx
extensions.
A
That
goes
in
and
says.
Oh
okay,
I
found
an
example
right
and
this
example
has
backslashes.
Let
me
put
let
me
like
automate
the
creation
of
this
linux
and
mac
os
versus
windows
tabs
here
and
then
create
within
the
windows.
Tab
do
the
same
command,
but
with
with
with
with
carrots,
right
up
carrots
instead
of
backslash.
So
how
do
you
push
the
docks
right
now?
Like?
Do
you.
A
No,
so
the
docs
have
their
own
ci
job,
but
yeah,
so
the
docs
have
their
own
ci
job,
but
you
can
run
them
locally.
Well,
what's
up.
A
C
C
A
Up
yeah,
exactly
that's:
why
that
it's
set
up
this
way,
yeah
if
you're
curious
about
it,
it's
all
in
the
run,
docks
then
yeah.
A
Yeah,
let's
see,
let
me
just
write
down
some
of
the
stuff
we
talked
about
here,
so
we
have
an
updated
list
of
plugins
that
work
and
have
issues.
A
A
A
The
appropriate
place
for
this,
I
think
there
may
be
multiple
appropriate
places,
but
I'm
not
sure
it
may
be
good
to
say
on
this
page
somewhere,
hey,
here's
all
the
ones
that
are
supported
and
not
supported.
A
Now
that
that
also
sounds
like
it
may
be
prone
to
not
being
updated,
because
we
may
forget
to
update
it
another
place
that
I
think
it
should
definitely
be
is
is
under
each
plug-in
right,
it
should
say:
hey,
pip
install,
you
know,
we
say
pip
install
whatever
right,
but
we
should
have
something
that
says
hey
by
the
way
this.
This
won't
work
on
windows
right.
So
that's
you
know
important
to
know,
so
we
should
definitely
put
it
on
the
on
the
plugins
page,
where
we
might
see
each
one
right
and.
A
A
Yeah,
however,
you
want
to
do
it.
I
just
think
I
think
yeah,
you
know
it
would
be
nice.
If
that's,
I
think
you
got
to
just
do
what
I'm
saying
right.
It
would
be
nice
if
we
could
see
it
all
in
one
place
and
then
you
know
they
should
each
individually
have
them
as
well
right,
because
if
somebody
goes
and
tries
to
do
this,
then
you
know
they
should
know
when
they're
presented
with
pip
install.
A
But
this
hey,
this
ain't
gonna
work
so
okay,
so
we
need
to
add
this
somewhere
in
the
docs,
where
we
list
all
the
plugins
that
work
work
wow.
I
can't
spell
today
work
for
each
platform.
A
Lets
the
user
know
if
or
what
platforms
that
plugin
can
be
installed
on
and
then
reference
the
list
or
the
slash
table
showing
support
from
the
main
from
the
docs.
A
Installation
right
so
because
we
so
wherever
you
end
up
putting
that,
make
sure
that
just
make
sure
that
when
we
say
you
know,
support
varies
by
which
plugins
you
install
like
here's,
the
list
of
ones
right
or
I
guess
replace
this
sentence.
That
says
we
don't
currently
have
a
list.
A
A
Yeah,
provided
we
have
complete
test
cases,
and
I
mean
I
think
I
think
we
do.
I
think
the
main
the
main
hurdle
that
people
are
going
to
run
into
when
they
try
to
do
this
on
windows,
is
that
you
know
the
command
line.
Examples
are
not
what
they're
thinking,
but
we
have
some
python
examples,
and-
and
you
know
that
was
also
the
other
thing
that
we
need
to
do.
Is
we
need
to
have
more
python
examples
right?
That
was
one
of
the
things
that
you
had
brought
up
a
while
back.
E
A
C
A
C
Yeah,
so
we
are
checking
for
whopping
rabbit
that
c
make
and
boost
is
installed
or
not,
but
the
documentation
doesn't
mention
that
anywhere.
So
it
always
fails
on
the
ci
as
well
as
locally.
If
we
run
dff
ml
service,
dev
install
and
it
won't
install
wobble
rabbit.
But
if
you
go
inside
wobble
rabbit
and
do
pip
install
dash
e
dot,
then
it'll
install
the
rabbit
and
all
the
tests
would
work.
Fine.
A
So,
let's
see,
let
me
write
that
in
my
notes.
So
why
are
we
checking
for
dependencies
and
plugins.
A
Okay
c
making
boost
yeah.
I
wonder
why.
Why
was
that?
Let's
see
you
get
blame.
Well,
we
know
it's
my
fault,
but
let's
see
check
for
boost
very,
very
helpful
common
message.
Sean.
Thank
you
very
much.
A
Yeah,
I
wonder
why
this
was,
I
know
at
some
point.
It
was
like
hashem
had
run
into
this
stuff
a
long
time
ago,
but.
C
A
Okay,
so
everything
that's
being
installed
from
conda
is
under
see.
Okay,
let's
also,
I'm
not
sure.
Let's
take
it
out.
A
A
Okay,
so,
okay,
here
here,
actually
I
think
here's
now
I
remember
what
was
going
on.
So
this
is
saying
that
if
whipple
rabbit
is
not
installed
already,
then
you're
going
to
need
c
make
and
boost
is.
That
is,
and
I
I
think,
that's
that's,
because
yeah
if
oppo
rabbit
is
not
installed
on
already
then
you're
going
to
need
cmake
and
boost
to
to
run
the
pip
install.
Was
that
correct
or
oh,
is
this
okay?
A
A
A
I
have
no
idea
why
we
would
have
been
doing
that.
That's
weird
september,
oh
september.
Oh,
that's
what
happened?
Okay!
So
when
we
did
this,
when
we
initially
did
this,
they
didn't
provide
compiled
packages
which
meant
that
if
you
installed
it,
you
installed
it
from
source,
because
this
was
in
this
was
four
months
ago.
So
in
september
they
uploaded
compiled
versions
of
all
of
them,
but
before
that
you
had
to
compile
it
from
source,
which
meant
that
you
you
had
to
have
the
had
you
had
to
have
boost
in
cmake.
C
A
A
Checks
what
happened
was
that
oppo
wabit
started
or
didn't
release
compiled.
A
Packages
for
previous
releases,
which
was
3.8
or
8.8.1
just
so
that
we
remember
everything:
okay,
all
right
any-
was
there
anything
else
on
there
that
I'm
not
capturing
okay.
So
we
got
the
link
to
this
issue
so
we're
tracking
stuff
in
there
and
we're.
C
A
A
Okay
and
the
answer
is
check,
dot,
ci,
depths,
dot,
sh
and
the
docker
file,
so
these
need
to
be
combined
at
some
point.
A
We
risk
wow,
there's
a
lot
of
wind
going
by
my
house.
The
entire
tree
is
losing
all
of
its
leaves.
This
industry
has
a
lot
of
leaves,
looks
like
it's.
Raining
leaves
out.
There
need
to
be
combined
at
some
point.
We
risk
them
getting
out
of
sync,
okay,
so
and
let's
just
open
that
up
real
quick.
So.
A
A
Okay
down
here
I
don't
know.
A
Okay,
this
is
the
incorrect.
The
comment
got
screwed
up
on
this
one:
okay,
yeah
now
here
install
dependencies,
no
okay!
No,
this
is
interdependencies.
A
Okay,
never
mind
all
right
so
in
this
file,
basically
you'll
see
it
right
so
anytime
on
to
install-
and
also
this
is
another
place
where
get
grep
is
your
friend
so
I'll
get
grapple
will
tell
you
wherever
we're
installing
packages
with
conda,
oh
and
then,
and
then
the
other
thing
is
that,
with
the
console
test
plug-in
that
that,
if
you're
running
under
a
conda
environment
already
when
you've
run
console
tests,
then
it'll
create
a
conde
environment
rather
than
a
virtual
and
buy
and
all
the
installs
will
be.
A
Honda,
installs
or
no
they'll
still
be
pip
installs.
They
just
install
into
the
conda
environment,
okay,.
A
Well,
it
depends
on
what
the
users,
basically
everything
will
work,
no
matter
what
the
users
are
doing
right
if
you're
a
user,
that's
ins
that
has
a
conda
environment,
then,
when
you
install
packages
it
just
installs
into
your
condo
environment
right,
if
you're
a
user
that
doesn't
use
conda,
then
you
know
you
just
won't
even
notice
right.
A
So
if
you
need-
and
I
think
this
is
one
of
those
things
where
this
is
probably
something
that
needs
to
be
noted
on
the
plugins
page-
actually
yeah-
I
think
this
is
something
I
came
across
yesterday
and
I
already
forgot
about,
which
is
that,
if
we're
doing
like
wopel
rabbit
or
dolph
or
pi
or
spacey,
we
need
to
tell
people
that
they
need
to
install
this
using
conda
they're
that
they
need
to
install
spacey
using
conda
before
they
run
the
install
right.
A
I
think
we're
missing
that
right
now,
so
people
will
would
may
try
to
install
it
and
then
not
realize
that
that
you
know
they.
They
really
have
to
install
this
using
conda
before
they
can
run
the
pip
install
of
this
module.
Why
why.
C
A
C
A
F
A
C
A
Well,
that's
great:
let's
see
spacey,
I
wonder
why
it
was:
let's
check
this
out
download
files.
Okay,
it
looks
like
these
guys
release
wheels
too.
So
yeah
there's
no
reason
for
this
to
be
installed
with
condo.
Why
are
we
installing
it?
Maybe
it's
probably
another
one
of
those
things
where
whenever
this
happened,
they
didn't
release
july
yeah
july,
I'm
not
sure
when
we
originally
got
spacey
support,
get
log
dash,
p,
marl,
spacey.
C
A
Yeah,
I'm
not
sure
why
why
himachu
added
it
using
yeah,
I
mean
I
guess
this
one.
We
don't
need
it.
So
it's
just.
We
can
remove
that.
So
this
is.
A
A
Yeah,
I
think
he
used
condo
for
a
lot
of
things,
and
so
that's
probably
what
happened-
and
I
know
there
was
one
there
was
one
thing
that
needed:
conda
that
didn't
like
there
was
something
that
needed
conda
at
some
point,
and
so
we
had
to
go
through
this
giant
mess
to
try
to
get
conda
working
in
the
ci
and
compatible
with
other
things,
and-
and
I'm
not
that
it
may
just
be
the
case
that
you
know
some
things
have
updated
on
on
pi
pi
beyond
needing
conda
right
and
I
think
it
may
have
been
wobl
rabbit
or
something
I'm
sure
we
can
look
through
the
weekly
minutes
and
find
out.
A
But
we
won't
do
that
right
now.
C
A
Yeah,
I
think
you're
right
so,
let's
create
an
issue
to
track.
So
let's
create
an
issue
to
track
conda
installed
packages
and
see
if
we
can
stop
installing
things
using
conda.
This
would
speed
up
our
ci
too.
C
A
A
Into
that,
because
that
took
a
long
time,
but
so
yeah
we
talked
about
that
last
time.
If
anybody
wants
to
they
talk
about
caching
and
docker,
you
can
you
can
watch
the
last
recording.
So
let's
create
an
issue
to
track.
It
looks
like
most
packages
are
now
releasing.
A
Compiled
versions
through
pi
pi,
we
may
not
need
the
we
may
be
able
to
get
rid
of
the
conda
slowness
in
the
ci.
Okay.
Great.
So
can
you
create
an
issue
for
that
yash.
C
No,
a
little
a
small
question
like
what.
What
is
there
in
the
secrets
for
get
like
the
fml
that
you
are
using
for
the
plugins
in
the
script.
A
Okay,
so
basically
I
mean
you're,
so
you
can
see
them
all
in
I
mean
you
can't
see
the
secrets,
but
you
can
see
see.
So
let
me
just
it's.
A
You
can
see
them
all
in
because
every
secret,
if
you're,
gonna
use
it
it
has
to
be
used
within
the
workflow
yaml
so
and
this
needs
so
there's
some
updating
that
needs
to
happen
here,
but
the
main
thing
is
that
they
will
release
themselves
to
pi
pi.
So
there's
one
thing
for
the
docks,
which
is
there's
an
ssh
key
that
pushes
the
docs,
because
it's
the
way
that
organization
is
configured
is
essentially
you
can't
push.
A
Oh,
I
might
be
able
to
change
that
now
actually,
but
they
used
to
they
used
to
they.
Had
this
whole,
this
whole
mess
where
you
couldn't
push
things
with
token
off
to
an
organization,
so
it
wouldn't
work
so
I'd
add
an
ssh
key,
so
that
was
for
the
docs
so
that
it
could
push
the
docs,
but
I
might
be
able
to
get
around
that
now
and
then
all
of
the
packages,
so
actually
there's
okay.
So
if
you're
curious
about
this,
maintain.
A
Okay,
so
if
you're
curious
about
this,
there
is
some
documentation
on
it
here,
but
basically,
when
we
create
a
new
package-
and
I'm
sorely
have
not
done
this
in
a
while,
so
we
need
to
make
sure
that
we
do
this
before
release.
This
is
one
of
the
release
process,
things
which
is
all
documented
here,
but
basically
I
have
to
go
in
and
change
the.
I
have
to
make
sure
that
we
have
a
secret
in
that.
So
I
have
to
go
into
pi
pi.
A
We
have
to
upload
the
package
because
you
can't
do
this
until
you've
uploaded
the
package.
If
we
didn't
have
a
package
yet
right
upload,
the
package
go
into
pi
fi
create
a
token
for
pi
pi.
That
allows
us
to
release
just
that
package
and
then
go
into
the
github
secrets
and
add
that
token,
as
a
secret
within
the
github
secrets.
A
And
then
what
will
happen
is
that
whenever
we
increment
the
version
number
on
the
package,
if
we
increment
the
version.py
it'll
release,
the
new
version
of
the
package
to
pi,
pi
and
that's
sort
of
the
workflow
is
there's
some
there's
a
lot
of
automation
around
the
release
process.
But
but
it's
still
not
completely
automated.
Yet
and.
C
A
Far
as
I
remember,
yeah,
okay.
A
Yes,
it's
just
gonna
blow
up
rc
I
mean
our
ci
is
already
like
incredibly
full
of
stuff
right,
but
I
would
think
okay,
so
here's
the
thing
is
see
if
you
can
figure
out.
Conda
first
see
if
you
can
figure
out
honda
first,
because
that'll
cut
down
the
ci
times
right
and
then
and
then
yes
add
them,
add
them
as
as
targets.
I
think.
A
Think
yeah
I
found
out.
I
saw
recently
that
you
can
actually
do
you
can
change.
A
Os
to
the
matrix
exactly
so
yeah
I
would,
I
would
say,
try
to
do
that.
The
one
thing
is,
though,
that
yeah,
I
guess,
if
you're
running
in
git
bash,
then
maybe
the
ci
script
will
work.
Anyways
the
dot
yeah.
A
A
Okay,
awesome,
yeah,
okay,
that
would
be
really
cool
and
then
so
yeah.
The
first
thing
would
be
see
if
we
can
get
rid
of
conda
and
then
and
because
that
would
speed
things
up
and
then
yeah
enable
them
within
the
regular
matrix
would
be
great,
oh
and
then
the
last
secret.
Let's
see
we
have
secrets
for
all
the
packages
where
the
github
pages
deploy
stuff,
and
then
we
have
the
code
cup
token.
So
that
is
oh,
that's
pretty
much
all
around
release,
stuff,
yeah,
so,
okay,
great
wow.
A
G
D
Yeah
this.
A
No,
I
don't
think
that's
necessary
right
now.
I
mean,
I
think
I
think
that
sounds
like
something
we
can
do
later
right.
So
as
long
as
we
can,
you
know
keep
the
tests
passing
after
this
right.
If,
if
for
some
reason
that
ends
up
being
that,
we
now
have
a
failing
test,
then
yeah
we'll
have
a
problem,
but
because
I
think
you
know
going
into
this,
you
had
the
test
passing
for
all
of
the
all
of
the
plugins
right.
A
So
you
know
you
just
want
to
make
sure
that
we
still
have
a
passing
test
going
out
of
it
right.
A
Yes,
okay,
cool!
So,
let's
see
working
through
through
accuracy,
score
phase
five.
B
A
Not
sure
yet
the
main
goal
is
just
to
keep
that
test
passing.
If
it's
not
passing,
when
all
other
changes
are
done,
then
we
should
implement
within
phase
five,
all
right,
yeah
you're,
closing
in
here.
Okay,
anything
else.
You
want
to
talk
about
at
all
or.
A
All
right,
cool,
saksham.
G
Oh
yeah,
so
I
just
just
before
joining
the
meeting.
I
looked
at
your
greater
message
about
you.
You've
plugged
in
some
links
for
the
context
patch
and
I
haven't
have
time
to
look
at.
C
A
I
was
trying
to
remember
any
like
unique
string
that
I
may
have
written
like
as
a
comment
or
some
code
or
anything
and
because
it
was
gripping
through
all
of
the
branches,
so
it
would
just
show
like
I
put
cap
and
there's
a
comment
that
says
cap,
and
so
there's
like,
like
every
there's
for
every
single
branch
that
I
have
added
from
everyone's
different
remotes.
So,
like
everyone,
everyone
who's
worked
on
the
project
and
then
all
of
the
branches
they've
created
it
showed
up
with
one
search
result
for
cap.
A
In
the
comment-
and
I
was
like,
oh
god,
I'm
never
gonna
find
this.
I
finally
remembered
what
the
hell
the
branch
name
was,
and
I
rebased
it.
So
I
think
that
this
should
be.
I
rebased
it.
So
if
you
go
and
you
look
at
ben's
ben
sex
scan
without
the
underscore
updated,
so
this
is
updated.
But
if
you
look
at
this
one,
this
is
the
one
that
was
not
rebased
on
top
of
master.
A
So
this
is
the
old
one
from
may,
and
so
I
tried
to
update
the
I
tried
to
give
this.
This
patch
should
cleanly
apply
with
cherry
with
cherry
pick,
because
I
rebased
everything,
but
it
may
or
may
not
be
working
because
I
haven't
tested
it.
So
all
I
did
was
rebase
it
so
that
you
should
be
able
to
cherry
pick
it
and
not
have
to
deal
with
the
rebaseness
of
it,
but
that
may
not
have
been
you
know
the
best
plan.
A
So
let
me
know
how
it
goes
all
right,
all
right,
yeah
but
yeah,
and
then
also
the
other
thing
about
this
is,
of
course,
I
think
this
breaks
stuff
because
there
is
yeah
this.
This,
I
believe,
is
an
api
breaking
change
so
because
it
says
it
no
longer
uses
the
keyword.
Arguments
is
the
reuse
things,
so
this
change
might
break
things
which
create
a
new
orchestrator
context
and
attempt
to
reuse
networks.
A
A
Okay,
hey
yeah,
this
remap
operation
might
be.
I
mean
things
test
tests
will
probably
fail.
I
I
think,
but
I'm
not
sure
if
there's
a
test
for
this
remap.
A
The
input
network
context
so
and
I'm
pretty
sure
the
change
for
this.
So
let's
see
so,
for
example,
anything
involving
a
sub
flow,
and
so
I'm
not
sure
I
think
ogen
did
some
stuff
on
that
too.
So
good.
But
let's
just
paste
this
in
here
grab
subflow,
okay,.
A
Self.Subflow
operation,
dataflow,
config,
dataflow,
so
okay,
but
that
doesn't
actually
okay
great,
I
think
that'll
be
your
only
issue.
Bim
dfml
operation
output
and
I
think
the
patch
is
basically
just
reuse,
equals.
A
All
right,
I
think,
and
then
the
last
thing
is
that
I
think
we
had
somebody
who
was
thinking
about
okay.
Let
me
link
to
that.
I
just
totally
left
that
out
of
the
meeting
minutes
didn't
I.
A
A
A
Okay
notes
on
context,
capping,
t
patch
and
then
let
me
just
copy
that
sentence
there.
All
right.
A
G
A
All
right,
okay,
I
guess
that's
fine!
Then
there
we
go
we're
good
all
right.
So,
let's
see
all
right,
then
the
last
thing
does
anybody
have
anything
else.
Otherwise,
I'll
just
cover
the
console
test.
Stuff,
real,
quick
here
so
or
you
know
I'll,
just
give
a
brief
once
again,
a
brief
thing
about
it.
So
I'm
writing
the
documentation
right
now.
I
think
I
fixed
this
yeah.
A
So
thanks
for
pointing
that
out,
gosh-
and
I
fixed
this-
I
believe-
okay
great,
so
the
whole
thing
is
basically
just
like
this
is,
and
just
as
a
recap,
this
is
how
we're
testing
that
the
tutorials
and
basically
they
so
I
got
it.
I
got
it
working
so
that
where
is
it,
I
got
it
working
so
that.
A
Commits
console
to
stocks,
let
me
see-
oh
no,
we've
already
got
it
in
here.
So
here's
an
example.
Basically
this
is
the
model
slr
in
the
in
the
library,
the
main
library.
So
this
was
how
I
converted
the
docstring
into
being
tested
by
this
new.
You
know
the
console
test
stuff
and
then
basically,
we've
got
this.
A
I
also
fixed
the
data
types
and
basically
this
is
the
extent
of
how
you
test
it.
You
basically
say:
okay,
import,
run
console
test
and
then
you
await
run
console
test
and
you
pass
it
the
class
that
you
want
to
test,
and
so
what
this
is
going
to
do
is
it's
going
to
let
us
write
doc
strings.
You
know
that
look
like
that.
Look
like
this,
that
are
restructured
tests
text
and
then
we
just
put
little
test
blocks
or
test
options.
A
So
you
put
this
test
option
on
the
code
block
and
literally
include
directives,
and
then
that
means
that
they
will
get
tested
or
if
it's
a
file,
then
it'll
get
written
out
into
the
temporary
directory
and
the
and
it'll
you
know
we
have
this
thing
that
basically
acts
as
like
a
it's
kind
of
like
a
lightweight
shell
implementation,
where
we
parse
these
blocks,
and
then
we
run
this
as
a
command,
and
it
knows
how
to
do
things
like
pipes
and
so
here's
an
example
of
a
little
or
literal
include
where
you
say
test,
and
then
it
copies
this
slr
file
so
that
you
can
run
it
then-
and
this
is
all
sort
of
in
this
documentation
here-
and
it
provides
a
little
example
and
then
you
can
also
use
it
so
that
it
includes
so.
A
A
This
is
the
arguments
and
this
is
the
options
and
then
it'll
do
the
same
for
code
blocks,
and
this
will
be
the
content,
and
so
there's
a
lightweight
parser
in
dffml.util.testing.parser
you
can,
when
you,
you
can
run
it
as
a
module
and
it'll
run
well,
you
will
be
able
to,
and
that's
this
that's
this
branch
when
it's
merged
and
it'll,
contain
this
documentation
and
so
you'll
be
able
to
run
it
as
a
module
and
you
just
pass
it
a
file
and
then
it'll
it'll
it'll
basically
run
that
file.
A
You
know,
run
the
examples
in
that
file,
so
this
is
sort
of
a
the
documentation.
Here
is
also
the
test
case,
and
so,
when
you
run
this
command,
it
runs
these
blocks
here.
A
So
it'll
show
you
how
you
know,
and
you
can
run
processes
as
daemons
in
the
background,
so
background
processes
you
can
compare
you
can
you
can
tell
it
to
run.
You
know,
run
this
this
command
until
the
output
is
whatever
this
compare.
Output
function
returns
true
and
there's
some
things
with
things
like
this,
which
is
python
code,
that
it'll
evaluate-
and
you
can
tell
it
to
ignore,
ignore
the
return
status
of
this
command.
A
Just
only
continue
once
the
output
comparison
is
is
correct,
if
you,
if
you
run,
if
you
run
things
as
daemons,
you
you
tell
it,
you
know
you
give
it
some
string
as
the
unique
key
so
that
if
you
run
another,
if
you
decide
to
replace
that
with
a
different
process
in
the
future
right,
so
here
we're
running
a
different
version
of
the
server.
So
we
control
seed
this
server
and
now
we're
starting
another
server
on
on
a
different
port
here
or
on
the
same
port
right.
A
So
we
had
to
kill
this
server
and
when
we
say
damon
with
the
same
key
here,
which
is,
we
use
the
port
number,
it
kills
the
guy
that
was
previously
existing
and
then
starts
this
as
now.
The
new
background
process,
so
things
like
pipes
are
supported,
but
ampersand
ampersand
are
not
supported
because
it's
just
like
a
simple
implementation
of
a
shell
and
I'm
going
to
document
sort
of
how
it
works.
In
case
you
want
to
extend
it
later.
A
You
put
commands
on
separate
lines
like
this
and
this
here's
an
example
of
basically-
and
I
thought
this
is
really
interesting-
how
they
correct
github,
actually
correctly
syntax
highlights
this.
So
basically
you
can.
If
you
have
a
code
block
you
can
put,
you
know
you
can
put.
We
just
saw
it
here
with,
or
we
just
saw
it
where
was
it
yeah?
We
just
saw
it
here
with
this
csv
file,
these
csv
files
here,
but
the
idea
is
that
you
have
some
content
and
then
you
say
hey
actually.
A
I
want
this
out
written
out
as
a
file
and
to
get
anything
to
happen.
You
need
to
put
test
on
it
or
else
it
won't
do
anything,
and
usually
the
other
thing
that
I
should
mention
with
in
this
example.
Here
I
put
a
python
file,
but
we
really
want
to
use.
Literal
include
for
all
of
the
python
files
anytime
we're
going
to
use
a
python
file.
A
We
should
actually
create
a
file
within
the
main
repo
and
then
literal
included,
and
we
want
to
do
that
because
we
want
to
make
sure
that
we're
still
using
if
you
put
python
in
one
of
these
code
blocks
the
auto
format
or
black
is
not
going
to
be
able
to
read
it,
whereas
if
you
put
it
in
its
own
file,
then
we'll
make
sure
that
everything
you
know
that'll
ensure
that
everything
gets
that's
formatted
correctly
and
yeah.
So
basically,
this
is
also.
This
is
not
working
at
the
moment,
which
is
why
I
haven't
you
know.
A
I've
got
got
some
other
stuff
that
I'm
working
on
and
I'll
try
to
fix
this
by
the
end
of
today
or
tomorrow,
but
yeah
that
last
command
isn't
working.
And
then
I
got
a
few
more
things
to
clean
up
here
and
I'll
ping.
You
all
again
for
for
another
review.
This
is
sort
of
meant
as
a
preliminary
thing,
any
questions
or
comments,
things
that
should
be
covered
in
here
that
oh
and
then
this
is
about
setting
up
the
environment.
A
So
you
can
provide
a
setup
function
to
basically
say:
hey
run
these
things
before
you
test
the
document,
any
anything
else
that
we
should.
You
guys
think
we
should
cover
in
here
other
than
sort
of
internals,
which
I
have
yet
to
cover
anything
that
that
you're,
that
you've
thought
about.
I'm
wondering
how
x
works
with.
A
Yeah-
and
I
think
I
got
I
got
those
links
here,
so
this
is
constant
pyo.
It's
just
internal
links
here,
yeah,
so
I
I
just
updated
that
so
I
got
those
now.
So
thank
you
for
that
yeah,
cool,
yeah
and
so
I'll
circle
back
and
ping.
A
You
guys
for
another
review
there,
because
I
want
to
make
sure
that
you
know
yeah
I'd
like
to
make
sure
that
this
is
correct
before
we
get
it
in
the
main
version
of
the
docs,
because
if
people
write
the
other
thing
is
that
I
want
to
update
the
model
tutorial
to
basically
take
the
tests.
You
know
I
want
to
still
have
the
tests,
but
I
also
want
to
have
a
section
that
says
you
know
under
the
packaging
that
says:
hey
how
to
write
the
documentation
and
here's
how
you
test
the
documentation.
A
You
know
you
import
the
console
test,
run,
console
test
and
and
now
you're
testing
your
documentation,
because
I
think
that's
that's
the
one
gap
we
had
in
there
at
the
moment
is:
we
didn't,
have
documentation
on
how
you
write
the
documentation
for
your
models
and
that'll,
hopefully
get
us
better
model
contributions,
because
they'll
have
docs
all
right
cool.
So,
let's
see
okay.
So
let
me
just
put
this
in
a
meeting
when
it's
gave
brief
overview
of
console
test.
Docs.
A
Because
it
runs
console
commands
because
well
so
this
is
not
a
good
good
name
for
it.
If
we
can
think
of
a
better
name,
let
me
write
that
down
too.
Can
we
think
of
a
better
name.
A
A
Yeah
because
there's
the
the
idea
was
that
there's
the
doc
test
plug-in
for
spank,
sphinx
and,
and
so
this
is
for
testing
console
commands.
So
this
is
another
console
test.
Why
is.
A
Named
verbs,
okay,
cool
is
this
test
console.
A
Commands
console
commands-
or
let's
just
put
console,
commands
and
I'll
put
this
in
the
description
too,
because
that's
a
good
point:
why
is
it
called
this?
There
is
another
sphynx
extension
that
tests
doc
strings
called
doctest.
A
All
right
cool
all
right,
thanks
everyone
and
have
a
great
week,
and
hopefully
we'll
have-
I
mean
the
way
things
are
going.
I
think
we
have
this
one.
I
got
to
get
this
documentation
in
there
so
that
that's
in
the
next
release,
so
that
if
we
have
contributors,
they
know
how
to
contribute
that
and
write
their
model
doc
strings,
and
then
I
think
we
also
just
have
the
main.
The
main
thing
is
that
I
need
to
set
up
some
scan.
C
A
Let's
see
like
this,
so
setup.py
requirements.
How
should
I
deal
with
this?
Should
I
just
move
it
to
setup
common
like
every
other
module?
Yes,
I
think
the
answer
is
if
there
is
a
setup
common.
Yes,
there
is
a
setup
common
in
which
case
yes,
you
should
move
it
around
here,
all
right
cool,
so
I
will
get
back
to
him
on
that.
I
didn't
see
that
comment.
So
great
looks
like
he's
he's
well
on
his
way
there,
so
that
hopefully
gets
us
we're
on
our
way
to
the
next
release.