►
From YouTube: Weekly Sync 2021-06-15
Description
Meeting Minutes: https://docs.google.com/document/d/16u9Tev3O0CcUDe2nfikHmrO3Xnd4ASJ45myFgQLpvzM/edit#heading=h.uiu53zd8h957
B
Have
fixed
all
the
issue
the
issues
related
to
models?
Okay,
so
right
now,
like
all
the
models,
are
passing
okay
test
for
the
models.
So
I
think
we
can
merge
that
and
also
I
have
also
started
working
on
the
sql
and
accuracy
score
thing:
okay,
great.
A
B
A
Okay,
so
yeah,
so
the
the
let's
just
check
the
I
think
this
style
with
style
passing.
So
what
I
want
to
do
is
I
want
to
make
sure
that
we
aren't
having
okay,
so
style
wasn't
passing
okay,
so
we
should
fix
that
oops,
okay
and
then
I
wanted
so
I
wanted
to
check.
A
I
noticed
that
these
guys
are
failing
and
I
believe
it's
because
of
that
stupid
unit
test
thing.
Yeah.
B
A
I
saw
somebody
else's
project
today
was
saying
something
about
that.
I
was
like
at
least
it's
not
just
okay,
so
this
one
is
maybe
okay,
so
this
was
failing,
but
it
looks
like
maybe
it
didn't
even
get
to
the
point
where
it
was
able
to
have
that
problem.
So
my
my
question
here
really
is
like.
A
Is
there
something
within
these
set
of
commits
that
that
that
creates
this
issue
or
is
its
or
was
this
issue
already
existing
here
now?
My
my
guess
is
that
we're
going
to
find
the
issue
if
we
fix
this
so.
A
B
So,
actually,
in
the
high
level
dot
five
high
level
file,
I
actually
changed
some
of
the
code
there,
because
when
we
actually
run
the
model
to
get
the
accuracy
there
is
this
accuracy
method
in
the
high
level
right.
So
in
that
we
are
actually
instantiating
the
model
context
and
the
accuracy
score
context.
B
So
there
like,
when
we
had
first
merged
this
into
the
into
the
the
merge
here.
A
Yeah,
okay,
let's
see
so.
B
Or
else
I
I
don't
feel
so,
there
are
many
changes
to
it.
A
Yeah,
so
let's
just
take
so
okay,
so
we're
experienced-
let's
just
let's
just
sort
of
re:
okay,
so
we'll
we'll
reset
ourselves
a
little
bit
here.
So
we
have
okay
so
on
this
commit
here
so
on
origin,
slash.
A
So
currently,
so
currently
on
origin,
slash
this
really
long.
Actually,
staging
rebates
master
was
this.
This
was
the
one
okay.
A
Great
at
commit-
let's
just
make
sure
we
record
all
this,
because
this
can
get
confusing
oops
that
commit
we
did
or
were
not
seeing.
A
The
weird
unit
test
issue,
which
is
this
right:
okay,.
A
All
right
on
the
branch,
so
we
should
also
say
we
were
seeing
a
different
error.
A
Which
is
this
import
error?
So
the
question
that
I
have-
and
I
think
you
know
just
to
restate
this
basically
for
recording
and
otherwise
is
you
know,
we've
we've
run
into
this
area.
A
million
times
like
all
of
a
sudden
unit
test
is
trying
to
call
a
function
because
I
don't
know
it
thinks
a
function
is
a
test
case
class.
A
So
for
whatever
reason
it's
trying
to
instantiate
it
as
a
test
case,
and
we
don't
really
have
much
much
much
hope
of
fixing
that
from
what
I
understand
so
so
anyway.
So
we
want
to
decide
what
we
want
to
do.
Is
we
want
to
see
if
there's
one
commit
you
know
between
this
pr
and
the
what
exists
that
would
have
caused
that
maybe
now
my
guess
is
that
this
would
exist
here.
A
So
basically,
if
we
were
to
check
out
this
branch,
my
guess
is
that
we
would
find
and
fix
this
issue
that
we
would
find
the
same
the
same
problem
so
but
we'll
yeah.
So
we'll
try
that
and
then,
if,
if
we
find
the
same
problem,
then
we'll
just
merge
it
and
we'll
fix
the
we'll
fix
it
later.
But
if
we
don't,
then
you
know,
then
then
we
have
to
deal
or
if
we
don't,
then
we
know
that
it's
something
within
these
set
of
commits
in
the
new
pr
right.
A
Okay,
so
all
right,
you're
all
mad
about
accuracy,
great
okay,
that's
the
right
error.
D
A
A
Okay,
so
all
right,
so
that
was
sort
of
separate
error,
separate
error,
all
right.
So
none
of
these
things
look
like
that
error,
which
is
good
because
we're
mainly
concerned
with
that
one
error
right
now,
so
we
can
say
in
their
notes,
so
fixing
this
error
seemed
to
result
in
different
errors,
not
the
unit
test,
accuracy,
error
above
okay.
So
now
we
know
that
so
now.
A
A
When
so,
we
should
say
win
run
with
python,
oh
crap,
sorry,
so
what
we
now
know
is
we're
basically
well,
you
know
one
step
farther
into
our
3.8
migration.
A
A
A
Let's
look
for
any
place
where
we
were
doing
that
and,
let's
maybe
just
change
it,
so
coverage
run.
Okay,
so
coverage
run.
Setup
py
test
is
going
to
be
a
problem
there,
because
coverage
run
takes
a
file,
not
a
module.
A
A
B
C
A
All
right
so
we'll
just
try
to
change
this
and
push
it
up
so,
and
this
is
how
just
to
show
you
guys
how
I
usually
do
this,
so
this
is
the
command
that
I
use
here.
So
I
use
I
try
to
do
this
fine,
so
I
try
to
do
find
on
everything
and
then
escape
that
period.
A
I'd
use
the
find
command
grab
all
the
five
python
files
and
then
I
do
sed,
which
does
like
a
find
replace.
So
you
can
give
it
this.
You
know
this
slash
s
syntax,
if
you're
familiar
with
that,
like
a
regex
type
thing
so
and
then
we
wanted
what
dash
m
unit
test
so.
E
C
A
Let's
see
and
then
on
this
one
with
dots,
the
ci1
dash
run,
so
these
coverage
ones.
So
this
guy
here
and
this
guy
here
are
going
to
be
probably
problems
for
us,
so
we'll
make
sure
that
we
we
deal
with
that.
So,
let's
just
check
out
coverage.
I
just
want
to
show
you
guys
how
I'm
going
about
this,
because
thought
process
was.
C
A
Done
all
right,
so
coverage
run
dash
m
discover
dash
v,
so
that
should
be
fine
now,
so
let's
just
try
it
we're
going
to
try
this
and
we're
going
to
see
if,
if
this,
as
the
all
of
that
looks
good
to
me,
stop
me
if
it
looks
wrong
to
anybody.
C
C
A
And
python
3.9
dropped
support
for
setup
by
test.
A
Hopefully
the
ci
will
run
and
then
let's
just
do
the
black
check.
A
A
So
updated
record
resources,
that's
ridiculous!
I
use
like.
Did
we
even
touch
that
that's
dumb?
I
don't
think
so.
A
Basically,
oh,
and
this
is
like
from
pre
even
that
like
so
this
is
what
we'll
do
is
we'll
fix
this
style
issue
once
we
merge
that
branch
in
because
I'll
just
take
that
and
I'll
do
like
what
we've
been
doing
and
I'll
just
well
I'll
re,
we'll
we'll
rebase
it
in
and
then
we'll
fix
up,
we'll
fix
up
that
commit
itself.
So
let's
just
make
a
note
that
this
is
the
offending
commit
and
then
we'll
all
right
so
wow.
A
This
has
been
a
very
complicated
process,
good
job
with
this
all
right
so
commit
as
a
s
black
formatting
issue,
we'll
fix
that.
A
all
right
great,
so
we
so
we
changed
over
all
the
test
running
commands
to
unit
test
discover
rather
than
sort
of
py
test
we're
re-running
ci
to
see
if
that
helps
all
right.
Anything
else
on
this
right
now.
A
A
That's
good
to
know:
that's
very
sketchy,
what
the
hell
windows,
it's
just
running
the
command
and
failing
and
then
telling
us
the
test
is
fine.
What
what's
up
with
that?
Maybe.
A
Maybe
yeah
well,
that's
really
not
helpful
for
the
purposes
of
testing
what
the
hell!
Well,
that's
really
concerning!
So
let's
definitely
make
an
issue
out
of
that.
I'm
glad
we
checked
hell.
Now
I'm
worried
that
we
have
failing
windows
tests
in
the
master
branch.
How
were
we
supposed
to
know
thanks
a
lot
bill
gates,
geez.
C
A
So,
let's
just
make
okay,
so
ci
windows
failing
tests,
do
not
cause
ci
job
to
fail.
That's
that's
a
big.
A
A
All
right,
great,
okay,
so
the
rest
of
this
we're
just
wow,
okay,
great
everything
else,
is
passing
all
right,
so
we'll
cross
our
fingers
that
well,
let's
see
what
happened
here.
That
was
actually.
A
A
A
C
C
C
C
Okay,
let's
operate
x
reference
for
assignment
that
should
be
an
easy
fix
as
well.
Let's
see
those
are
probably
from
the
same
thing:
aren't
they.
C
B
A
Yeah
phase
eight
all
right,
so
I
think
that
yeah,
okay,
so
this
one's
an
easy
fix.
This
is
just
add
to
add
to
this
console
test.
A
Docs
yeah,
this
all
needs
to
be
restructured.
What
we're
talking
about
the
okay
wait?
No!
This
needs
to
be
added
to
we're
talking
about
restructuring
all
of
this
testing
for
the
documentation
right
now,
all
right.
So
this
will
all
change
hopefully
soon,
because
we're
trying
to
split
out
everything
into
the
second
party.
C
A
A
A
And
it's
a
csv
source,
so
you
know
it
is
correct.
So
let's
see.
A
A
Yeah
well,
the
no
async
part
should
just
should
just
be
a
proxy
for
high
level,
because
if
we
look
in
no
async
yeah,
all
of
these
should
just
be
async.
Io
runs
on
the
high
level
functions
and
then
modifications
to
the
docs
strings.
A
Yeah,
for
the
sake
of
to
do
replace
this
with
so
when
we
do
the
type
checking,
so
we
have
type
hinting
on
a
lot
of
things,
but
we
don't
have
it
on
everything.
A
There
was
a
talk
on
this
at
pycon
recently,
it's
basically,
it
should
be
a
way
to
validate
this
type
hints
at
runtime.
I
believe
yes,
I
think
it
is
okay.
A
Yeah,
my
pi,
I
think
you
knew-
and
I
think
we
mentioned
my
pi
in
here
or
where
was
oh
pi
type.
Okay.
Well,
I
guess
I
don't
know
where
my
why
my
pi
wasn't
included
in
this
list.
Oh
yeah,
actually
there's
my
pie
for
some
reason:
it's
down
here,
but
not
there.
A
A
So
if,
for
some
reason
you
know
you
end
up
something
doesn't
get
caught
by
the
static
type
or
checker
it'll
catch
it
at
runtime.
It
does
do
you
know
if
my
pi
can
do
that
as
well
or
because
then
we
could
just
use
one
thing.
A
Okay,
all
right:
well,
they
may
they
may
add
that
at
some
point,
so
we'll
just
keep
our
eye
out
for
that
as
well.
So-
and
this
is
why
this
is
sort
of
like
you
know,
this
is
a
long
thing,
because
we've
got
a
bunch
of
places
where
this
is
not
the
case.
It's
static,
type,
checking
and
maybe
dynamic
through
something
like.
C
Pie,
damn
tick
c
issue
number.
A
36
issue
36
wow,
okay:
this
has
been
around
for
a
while,
so
we'll
just
say
you
know.
If
not,
I
think
we
also
have
a
bug
here
where
this
should
be.
A
A
A
No,
this
wasn't
it
all
right.
They
were
supposed
to
be,
or
was
it
the
double
context?
Entry
yeah?
Okay,
so
this
is
not
getting
tested
so
because
this
okay,
so
what
I'm
saying
here
is
basically
the
this
code
must
not
being
tested
because
the
this
would
break
so
see
if
we
go
in
here
and
it's
given
model
and
accuracy
score.
Oh
it's
because
it's
train.
A
Train
model
now
this
should
break
too
right,
because
we're
saying
if
the
instance,
if
the
model
is
an
instance
of
model,
then
create
the
model
context
otherwise
create
this
mctx
so
basically
enter
the
first
context,
then
into
the
second
context
and
then
use
the
you
know,
use
the
second
context.
However,
this
mctx
variable
shouldn't
be
defined.
A
You
know
if,
if
there
is,
if,
if
you're
passing
a
model
context,
so
that's
what
makes
me
concerned
that
this
might
not
be
getting
run.
A
A
I
thought
there.
I
think
I
keep
trying
to
create
one,
and
then
I
get
sidetracked
so
where
was
make
sure
context
can
be
kept
open.
Okay!
Well,
this
is
not
a
very
descriptive
issue,
so
in
high
level-
okay,
let's
just
link
to
it.
A
So,
let's
just
raise
this
raise
this
error
for
now
and
that
way
we
should
catch
the
places
that
we
didn't
remember
to
change
like,
for
example,
if
we
run.
A
So
if
we
run
this
test
case
again
that
was
failing,
we
should
yeah.
We
should
get
this
type
error
now
yeah.
So
now
we
should,
and
now
we'll
know
that
we
need
to
go
change
that
test
case
and
people
will
know
that
you
know
that's
a.
A
So
anyways,
I
think
I
think,
let's
see
so
that
will
give
us
a
bunch
of
that.
Will
what
all
were
the
test
cases
that
failed.
So
these
guys
were
probably
all
offenders
and
then
this
thing,
where
there's
no
path
lib
is
confusing,
but
so
basically
I
think
when
I
push
this
like
you
know
this,
when
we
push
so
I'll
push
this
right
and
then
that
should
show
us
in
the
ci
where
there's
other
issues.
A
And
then
this
guy-
and
we
should
probably
rebase
this
in
somewhere,
so
high
level,
accuracy
fix
type
annotation
of
accuracy.
C
A
All
right
so,
let's
see
what
happens
here
and
then,
hopefully
we
should
have
a
more.
I
think
I
think
we
will
end
up
with
a
few
places
that
we
need
to.
You
know
a
few,
a
few
places
like
that
one
test
case
where
we
we
just
need
to
update
that
and
hopefully
the
rest
of
the
test
pass
and
when
that
happens,
then
we
can
merge
it
all.
A
And
then
we
are
very
close
to
merging
this
into
master,
which
is
very
exciting,
and
let's
just
make
sure
that
the
other
I've
noticed
that,
let's
see,
let's
make.
C
A
Fingers
crossed
okay,
so
it
did
run
the
test
great.
So
hopefully
we
should
have
some
more
helpful
results
in
a
second
here
all
right
and
then
we're
going
to
record
this
as
a.
C
A
A
This
is
why
I
need
to
refactor
all
this
we're
talking
about
with
the
second
party
plugins
refactoring,
about
all
of
the
all
of
the
testing
of
the
documentation
and
the
doc
strings,
and
all
that
into
some
more
generic
infrastructure
so
that
all
the
plugins
can
be
tested
in
in
in
one
way
like
so,
if
you
did
like
dfml
service
dev
create,
and
then
you
created
a
plugin
right
and,
and
you
had
some
documentation
that
you
wrote
in
there,
then
you
know
you
might
have
this
one
test
case
file
that
basically
just
goes
and
runs
all
of
that
stuff.
A
E
C
E
A
I
don't
know
what
happened
to
the
body
there,
all
right,
okay,
I
think
we
are
good
on
this
stuff
for
now,
thanks
that
took
a
while.
So
all
right.
So
what
else
do
we
have
today,
then?
So,
it's
so
sudha.
Do
you
have
anything
else.
B
Yep,
I
have
a
small
thing
to
ask:
okay,
so
for
the
accuracy
score
for
the
sqm
accuracy
scorer,
we
have
two
things:
one
is
the
scorer
which
will
actually
use
the
models
context,
the
model's
default
scorer
and
other
one
is
the
sql
metric
scorer,
which
I
will
be
implementing
from
scratch,
so
so
for
the
the
scorer
which
we
will
be
implementing
for
the
model,
the
default
model.
A
I
think
within
this
I
could
learn
models.
I
would
not
package
it
as
its
own
plugin,
so
so
add
this
under
the
model,
so
I
could
see
wpy
or
something
right.
B
A
B
All
right,
actually,
I
have
been
working
on
it,
but
it's
not
completely.
Okay,
okay,
that's
fine!
So,
basically,
yes,
so
the
thing
is
we
have
in
escalant
we
have
two
scorers
two
types
of
scorers,
one.
A
B
The
one
is
which
actually
uses
the
model's
default
score
method
and
another
is
the
escalant
metrics,
which
is
which
I
will
be
implementing
from
scratch.
B
E
B
B
I
can
ship
it
with
the
model
itself,
but
the
one
where
I'm
implementing
the
scalar
dot
matrix
the
the
new
scorer,
which
I'm
implementing
should
that
also
come
with
this.
B
B
A
A
Scikit
do
model
scikit,
or
should
we
put
it
in
dfmo
accuracy
scikit,
so
the
scores
could
be
used
with
non-psychic.
Are
these
psychic
models,
so
they
might
want
to
be
in
their
own
package
at
some
point.
A
Because
so,
let's
just
say,
because
we
are
actively
going
through
the
second
party
transition,
let's
not
create
another
plugin.
At
the
moment,
we.
A
We
we
might
want
to
do
this
later
at
the
end
of
the
day.
You
know
if
there
is
there's
so
there's
there's
something
to
be
said,
for
you
know,
there's
a
there's,
a
certain
code
based
size
that
comes
with
having
you
know
like.
If
we
split
out
you
don't
have
to
download,
you
know
you're
only
downloading
the
scores
right.
If
you
want
to
use
the
scores
right
and
that's
just
the
abstractions
around
the
scores
now
there
is
also
some
sort
of
you
know.
There's.
C
A
Like
a
trade-off
to
be
had
where
okay
like,
if,
if
the
psychic
scores
end
up
being
like
500
lines,
you
know,
is
it
really
worth
creating
their
own
package
for
right?
Maybe
yes,
maybe
no,
but
we'll,
but
that's.
I
think
it's
it's
a
great
great
thinking.
It's
exactly
like
the
way
that
we're
trying
to
think
about
this
right.
I
think,
let's
just
let's
just
not
do
it
right
now,
just
because
it
I
think
it'll
be.
A
You
know
more
more
trouble
than
it's
worth
to
do
at
the
time
definitely
definitely
worth
considering
in
the
future.
So,
let's
maybe
let's
maybe
create
an
issue
to
track
that
too.
So
so.
C
A
Accuracy
scikit,
should
we
split
out
psychic
accuracy.
A
Models
and
then
I'll
link
to
here.
A
A
A
2018:
okay,
11
28,
11,
28,
perfect,
okay,.
C
So
I
want
some
input
on
this.
Like
is
it
quite
oh,.
A
And
let
me
just
quickly
address
the
so
this
one
I,
where
did
I
go
log
time,
decorator,
all
right,
so
this
one
right.
So
there
is
an
issue
and
I
keep
forgetting
to
fix
this.
What
happened
to
it?
I
think
I
forgot
again
damn
it.
I
should
just
open
a
polar
question.
I
then
I
won't
okay.
So
the
thing
is
that
in
the
refactor
of
where'd
it
go
in
the
refactor
of
the.
A
Where
is
it
docs.
C
Struggling
here
where's
the
stupid
thing.
A
All
right
so,
basically
there's
the
docs
are
not
the
doc
tests
are
not
running
is
the
problem,
and
this
is
so
so
I
don't
want
to
merge
that
until
they're
being
run
in
the
ci,
because
because
obviously
then
we
don't
know
if
it's
working
or
not
because
it's
not
running
it's
not
running
the
tests
either
it's
actually
not
run
anywhere.
A
Oh
basically,
the
docks
I
haven't
merged
the
docks
one.
Yet
because
what
the
hell,
because.
A
Because
there's
a
there's
there's
a
problem
with
the
the
way
that
the
tests
are
being
run.
Where
is
this
stupid
thing?
Great?
Okay,
I
forgot
to
see
I
keep
so.
The
thing
is
I
keep
going
to
do
this
and
then
there's
then
there's
a
problem,
and
then
I
don't
finish
it.
So
where
is
this
so
we
look
at
and
let
me
just
look
at
master
intel
all
right,
so
docs
or
test
doc
strings.
A
What
the
hell
I
swear!
I
just
did
this
okay,
so
here.
So
this
is
the
problem.
Basically,
you
see
look
at
this
so
the
way.
So
this
is
creating
a
test
case
class
and
then
it's
assigning
methods
to
it,
and
actually
I
think
this
is
part
of
the
3.9
thing.
That's
what
happened?
Okay,
whatever
sorry!
A
The
problem
is
that
this
is
actually
creating
a
tuple,
because
I
think
I
I
probably
put
an
extra
something
and
then
it
formatted
with
black
and
then
now
it
looks
like
this
and
so
essentially,
instead
of
creating
a
you
know,
this
function
returns
a
function
which
becomes
the
test
case
to
run,
and
so
this
is
supposed
to
do
the
dock
test.
The
python
dock
test,
but
instead
it's
a
tuple,
so
it
doesn't
actually
run
the
test
case.
So
none
of
the
test
cases
are
being
run
right
now.
So.
A
C
A
Let's
see
what
happens,
I
think
there's
some
bugs,
which
is
annoying.
I
was
gonna
push
it
and,
but
now
you
guys
know
so
I
think
I
keep
bringing
it
up
in
the
mentor
media.
Now
I
keep
forgetting
to
say
it
here,
but
this
is.
This
is
a
thing
that's
going
on
so
yeah.
I
had
hesitate
to
push
it
because
it's
going
to
break
the
ci
in
the
master
branch,
but
it
should
probably
just
be
breaking
it,
but
that
now
that
just
makes
things
confusing
so
anyways.
A
Let's
just
see
what
happens
here,
okay,
so
this
is
why
this
has
not
been.
I
yeah
I
meant
to
respond
to
you
and
then
I
kept
trying
to
fix
the
bug,
and
then
I
forgot
to
say
anything
so,
oh
yeah
this.
What
the
hell!
This
is
weird.
This
cash
download
on
pack
archive
stuff
is
not
making
any
sense
because
it's
trying
anyways
this
is
the
bug
that
I
ran
into
and
then
I
I
sort
of
you
know
I
hit
a
wall
there
and
then
these
guys
are
all
so
yeah
there's
bugs.
A
Basically,
so
let
me
just
create
an
appear
for
this.
A
I
think
this
is
on
the
this
is
on
the
computer
with
the
hard
drive.
That's
failing!
That's
what
happened
to
test
product
strings.
C
A
For
test
dock
string
all
right,
so
basically
you
could
grab
this
change
and
then
try
to
run
just
your
tests,
but
I
have
a
strong
feeling
that
this
is
the
thing.
So
I
have
a
strong
feeling.
This
won't
work
without
unit
test
mock
stuff
so
because
see
where
was
it
yeah?
So
I
don't
know
if
this
is
going
to
work,
and
this
is
why
it
had
held
off,
because
I
was
hoping
that
it
doesn't
really
help
you
to
not
have
the
ability
to
test
it
right
for
me
to
say
anything.
A
Where
is
it
right,
yeah?
So
just
the
fact
that
it
says
exactly
one
seconds:
it's
probably
not
going
to
be
what
it
is
right
it
usually
it
it
sometimes
drifts
right,
so
we
might
need
a
a
mock.
Yes,.
A
A
Okay,
now,
hopefully
I
won't
lose
that
again
and
you
guys
know
what's
going
on
there,
so
you
could
grab
that
and
you
could
apply
it
here.
If
you
wanted
to
try
it
or
you
know
you
could
just
make
make
that
same
change
and
then
try
to
run
the
test
case.
So,
let's
check
out
because
then,
if
we
do
this.
C
A
Time,
okay,
yeah,
so
okay,
so
it
won't
output
the
logs
actually.
So
that
might
be
a
good
thing.
Anyways,
let's
see
simple
function
took
one
second,
I
wonder
if
there's
a
way.
A
Yeah
yeah:
let's
see
it,
logging.
A
Yeah,
okay,
so
it
does
it,
it
is
so
it
is
obviously
it
is
doing
it
and
it
is
the
right
time.
So
I
wonder.
A
Yeah
I
mean
you
could
just
take
it
out,
you
could
just
sort
of-
and
this
might
be
this
would
sort
of
solve
all
our
problems
right
away.
You
could
do
something
like
this,
where
you
say.
Oh
I've
got
a
bunch
of
white
space,
so
you
could
just
say
something
like.
A
A
A
Set
the
level
until
it's
logged
standard
out.
A
We're
just
getting
rid
of
the
white
space.
This
check
has
purposely
worked
until
so
now
we
can
try
this.
A
Basic
config
stream,
equal
system
level
equals
login.
So
here's
another
thing,
I
don't
know
if
you
guys
have
seen
this.
I
think
I
forgot
to
but
this
scripts.
Where
is
it
scripts.
A
Yeah:
okay,
what
all
right!
I'm
sorry.
D
A
C
A
A
However,
that
doesn't
seem
to
work
for
some
reason,
so
I
I
I'm
not
sure,
I'm
not
sure
what
we
should
do
here,
but
we
do
need
to
make
sure
that
it
passes
something.
So
I
guess
what
we
could
do
is
we'll
just
go
back
to
this
way
and
we
can
just
do
it
this
way
where
we
can
say
hey,
you
know
this
is
an
example
right,
and
you
should
see
this
if
you
have
login
enabled.
A
A
A
Didn't
say
I
mean
it
said
that
it
didn't
it.
You
know
it
said.
It's
basically
said
it's
not.
You
know
it
might
be,
because
you
know
I'm
not,
I'm
not
sure
that
really
doesn't
make
any
sense
to
me
why
it
wouldn't
work,
but
either
way
you
know
it
didn't
so
it
didn't
work.
We
can.
A
Well,
so
there's
a
lot
of
different
things
that
have
I
mean,
there's
a
lot
of
different
kind
of
test
environments
that
we
we
have
so
and
and
and
so
let's
see
so
there's
basically
three
there's
the
doc
test,
which
is
this
right,
we're
looking
at
doc
tests
right
now
and
then
there's
the
regular
test
cases,
which
is
you
know
like
the
one
that
you'd
written
that's
up
there.
So
the
regular
test
case
is
just
you
know.
That
would
be
this
right
and
then
there's
the
console
tests
and
the.
A
Things
like
the
restructured
text
files
so
like
where
we
just
saw
that
let's
see
tutorials
models
or
double
context
entries,
so
this
stuff
here
right
with
like
the
colon
test
on
it,
so
restrict
your
text.
Files
with
that
or
you'll
see
it.
You'll
see
it
within
some
of
these
within
some
of
the
docs
string
classes
as
well.
A
So
like
this,
so
so
yes,
there
are
multiple
testing
environments
and
yeah.
I
guess
do
what
what
else
do
like
what
what
questions
you
have
about
that.
D
Yes,
but
I'm
getting
really.
A
D
A
Okay
yeah,
so
essentially
we
have.
There
are
multiple
test
environments,
you
know,
there's
the
three
and
they
do
operate
differently,
and
so
we
we've
done.
We've
done
a
lot
of
things
to
try
to
make
them.
You
know
more
integrated
into
the
main
unit
testing
environment,
but
it's
not
you
know
they
still
have
their
differences.
The
one
that's
going
to
be
the
most
straightforward
is
always
you
know
the
regular
unit
tests
things
get
a
little
bit
different.
You
know
they
get
a
little
bit
weirder
as
you
get
into
the
doc
tests.
A
We're
still
leveraging
the
regular
doc
test
infrastructure
from
python
we're
just
that's
all
in
the
standard
library
we're
calling
out
to
it.
The
console
tests
are
a
custom
thing,
but
yeah.
So
so,
basically
there
are
some
quirks
and
I'm
assuming
you
probably
can't
hear
everything
I'm
saying
but
but
you
know
you
can
look
at
that,
you
can
look
at
the
recording
for
hopefully
a
clearer,
sound,
so.
A
Okay,
great
so
basically
yeah,
there's
the
three
things.
Yes,
yes,
okay,
perfect
great.
Do
you
have
any
questions
on
any
of
that
or.
C
A
Yeah,
that's
nice
all
right
great
and
let
me
just
sort
of
commit
this
and
push
it
up
so
and
then
you
understand
so
basically
what
I,
what
I
was
saying
here
is:
it
looks
like
we
weren't
successful.
I
don't
know
redirecting
the
standard
out.
Didn't
work,
first,
try
so
whatever
we
would
have
had
to
wrap
it
and
probably
do
some
things
to
make
sure
it
always
says
one
second.
So
this
is
actually
sort
of
an
easier
approach.
We
can
just
say:
hey
enable
logging.
A
You
should
see
this
in
the
output
and
that
way
we
don't
have
to
do.
Comparisons
and
and
hacky
stuff
to
you
know,
do
unit
test
mock
so
hopefully
that
that
fixes
that
so
I'm
just
gonna
rebase
this
and
push
it
back
up.
That's
not
what
I
wanted
to
do,
but
whatever
and
I'll
just
fold
that
into
there.
A
All
right
and
I'm
just
gonna
rebase
all
this
into
one
thing
here.
A
So
all
right,
so
I
got
rid
of
the
white
space
and
then
change
the.
C
A
A
C
A
All
right
so
remove
white
space
and
move
login
to
separate
quick.
C
A
What
do
you
want?
Do
you
want
a
guide
discussion
here
or
you
want
me
to.
D
What
I
was
hoping
to
do
was
use
as
a
cheater's
arcade,
unpack
thing,
but
the
later
on
it
would
have
been
difficult
to
add
new
formats.
So
I
switched
to
this
supported
archive
formats,
dictionary
thing,
and
then
I
thought
that
I
will
just
put
those
classes
in
that
would
handle
like
four
tar
files.
There
is
a
tar
file
class
and
built
in
standard
library
and
four
zip
files,
similarly
and
all
in
all,
but
then
I
realized
that
they
cannot
be
used
in
in
the
same
way
like
they
have
different
interfaces.
D
So
I
created
a
base
class,
which
is
this
one
like
the
helper
base
class,
which
is
basically
an
abstract
base
class.
So
whenever
you
need
to
implement
a
new
method
new,
what
what
whatever?
What
I
should
call
it?
A
new
format
to
support
just
use
this
base
class
and
implement
those
two
methods
and
all
other
methods
depend
on
those
two
methods.
So
it
becomes
uniform
for
all
them.
A
We
don't
okay,
so
let's
see
okay,
so,
first
of
all,
let's
make
sure
that
everything
derives.
Okay,
let's
see,
let's
see,
okay,
so
the
standard
pattern
that
we
follow
is
to
have
everything.
Every
object
be
like
one
of
those
base,
configurable
objects,
and
so
let's
see,
if
we
were
to
look
at
for
example,
model
so
model
is
so
base.
Data
flow,
facilitator,
object,
context
and
base
data
flow
facilitator
object,
so
this
is
so
this
is
sort
of.
A
So
we
try
to
make
everything
follow
this
pattern
because
we
don't
know
if
at
some
point
it
might
not
follow
this
pattern.
Basically,
we
try
to
make
everything
follow
this
pattern,
because
this
allows
us
to
do.
This
basically
makes
it
so
that
if
we
run
into
problems
like,
let's
see,
I'm
not
not
able
to
describe
the
silicone.
So
by
always
following
this
pattern,
it
makes
all
of
our
code
like
consistent,
like
everything,
is
consistent
to
the
way
every
class
is
used
throughout
the
entire
code
base.
So
everything
has
a
config.
A
A
You
know,
network
access
or
something
right,
because
you
have
two
two
spots
where
you,
where
you
can
do
asynchronous
io,
you
can
do
it
on
the
first
context,
entry
which
is
sort
of
something
that
is
alive
for
the
lifetime
of
the
object
or
you
can
do
it
on
the
second
one,
which
is
usually
alive,
for
you
know
a
smaller,
a
smaller
lifetime.
A
You
know
the
lifetime
of
an
operation
so,
for
example,
like
with
the
model
the
first
context
entry
is,
is
you
know
when
the
model
is
instantiated
and
so
that
we
might,
you
know,
load
load
the
directory,
and
this
is
you
know
where,
where
we
might,
for
example,
with
the
with
the
file
formats.
This
is
where
we
might
load.
A
We
will
probably
use
this
we'll,
probably
versus
so
we
will
probably
do
within
a
inter
method,
we'll
probably
do
the
loading
of
of
the
of
the
of
the
data
from
the
archive
right
and
put
it
into
directory.
A
So
so
the
thing
is
what
you'll
find
is
so
what
you'll
find
is
everything
starts
out
as
as,
as
you
know
like?
Oh
it
just
does
this,
and
it
just
does
this
and
then
all
of
a
sudden
you're
like
okay?
Well,
what
if
the
zip
archive
you
know
needs
to
take
certain
arguments?
For
example,
some
of
these
archives
are
compressed
at
different.
Let's
go
to,
let's
see,
what's
a
good
example
here,
so.
A
So
say
I
wanted
to
create
a
zip
file
say
I
wanted
to
store
my
model
in
a
zip
file,
but
I
wanted
to
do
it
in
compressed
with
bzip2
right
so
now
now,
if
I
had
so,
if
I
don't
follow
the
config
the
standard,
config
format,
I
now
don't
have
a
way
to
specify
that
that
it's
a
be
like
that
I
wanted
to
do.
It
would
be
zip2
right
because
I'd
need
to
you
know
it
would
be.
The
extension
would
be
dot
zip,
but
I'd
need
to
pass
that
that
was.
D
A
D
So
what
do
you,
what
you're
trying
to
say
is
all
the
helper
classes
should
have
a
conflict
so
that
we
can
pass
them
certain
parameters.
If
you
want
to.
A
Yes,
yes,
so
basically,
so
as
long
as
everything
has
a
config,
everything
can
be
every
as
long
as
everything
follows
this
pattern,
everything
can
be
configured,
you
know
through
through
any
interface,
and
when
I
say
any
interface,
I
mean
the
python
api,
the
command
line,
api,
the
http
api.
A
You
know
all
of
that
stuff
right,
because
we
have
to
support
many
different
interfaces
and-
and
we
need
the
ability
to
make
sure
that
you
know
they're
all
they're,
all
they
all
all
of
the,
because
all
of
these
classes
get
get
configured.
A
lot
of
this
project
is
actually
configuration
and
and
then
the
way
that
we
configure
the
different
plugins
and
stuff
like
that,
because
you
know
we
have
to
standardize
everything
to
be
able
to
work
across
different
different.
You
know
different
frameworks
and
stuff
and
underlying
things.
A
So
so,
basically
we
always
try
to
if
you're
gonna
create
new
classes.
You
know
maybe-
and
this
might
be-
this
might
be
a
place
to
create
a
new
class
it
might
not.
But
if
you're
going
to
create
new
classes,
then
you
should
follow
that
format,
and
that
also
mean
that
also
it
all
the
other
thing
that
it
does
is
it
allows
us
to
hook
into
the
plug-in
system.
A
So
you
will,
you
will
register
pretty
much
everything
gets
registered
as
an
entry
point
and
that
way,
if
you
have
you,
know
different
archives
that
are
not
a
part
of
the
main
so
now
say
say
there
is,
for
example,
rpm
file.
There's
you
know
say
say
you
have
something
stored
in
an
rpm
and
that's
not
one
that
you
can,
but
it's
not
one
you
can
pack
into,
but
you
can
unpack
at
least
with
the
python
module.
A
A
Plugins
and
you
know
you
can
you
can
use
various
third-party
libraries
as
as
you
will
right,
and
so
that's
why
we
always
follow
that
same
pattern
in
this
case.
A
Was
let's,
can
you
can
you?
Can
you
send
me
where's
your
proposal
here?
I
don't
have
it
handy,
can
you
send
me
the
copy
of
it
real,
quick,
yes,
shall
I
send
it
on
twitter
yeah.
D
C
A
A
See
I
think
that-
and
this
is
sorry
I'm
sorry
I
haven't
had
my
meeting
with
you
guys
yet
for
our
one-on-one-
I've
been
extremely
busy
with
things
okay,
so
yeah!
Okay!
No,
I
don't
remember
what
I
was
thinking
here.
So
the
one
thing
that
we'd
like
to
get
to
eventually
is
the
use
of
like
more
operations
rather
than
you
know,
creating
plug-ins
for
everything,
because
operations
become
like.
A
You
know
you
don't
have
to
implement
the
plug-in
type
for
everything.
If
you
just
have
an
operation
so
and
and
the
operations
are
they
already
follow
that
pattern
so.
C
A
You
all
right,
so
I
think
what
what
you
should
focus
on
is
really
just
doing
like
a
a
decompress
and
decompress
right
almost
as
as
functions
rather
than
than
methods,
because
what
you
can
do
is
you
can
take.
Let's
see,
let's
see
here
so.
A
We're
we're
running
pretty
long
today,
so
do
you
guys
want
to
have
another
meeting?
Would
you
guys
want
to
keep
going
or
do
you
want
to
have
another
meeting
on
later
in
the
week
on
friday,.