►
From YouTube: Weekly Sync 2020-06-30
Description
Meeting Minutes: https://docs.google.com/document/d/16u9Tev3O0CcUDe2nfikHmrO3Xnd4ASJ45myFgQLpvzM/edit#
A
A
So,
let's
see,
let
me
let
me
let
me
sort
of
I
need
to
sort
of
get
organized
I'm
working
on
so
I'm
a
little
bit
I'm
a
little
bit
swamped
right
now,
I
think
yeah,
it's
probably
that's
it.
The
thing
is
it's
probably
the
same
error,
so
it's
just
like.
Can
I
I'll
try
that
if
that
is
that
the
only
thing
that's
left.
A
Okay,
so
all
right,
well,
wool,
so
I'm
I'm
gonna
do
this
I'm
gonna
do
the
thing
I
was
doing
where
I
make
the
agenda
here
and
then
we
talked
about
it.
So
we'll
get
everybody's
points
first,
for
what
we
want
to
talk
about
this
week
and
then
we'll
talk
about
that.
One
first
so
and
I
might
just
make
the
note
on
shirt.
Config
not
really
needed,
and
this
is
get
er
not
really
needed
for
get
er
right.
A
C
A
A
A
A
All
right-
and
that's
let's
see
so
yes,
so
that's
done,
talked
about.
A
B
D
A
A
A
Right
sorry,
so.
C
A
A
E
A
A
It's
not
critical
that
we
get
this
fixed
right
away.
It's
just
you
know,
I'll
put
the
version
range
on
and
whenever
we
can,
that's
don't
be
good,
so
note
that
okay,
the
camera
was
reminded
of
that.
Oh
because
you
said
that
test
named
out,
okay,
all
right
and
then
the
other
thing
you
said
section
was
that
you
want
to
figure
out.
How
do
we
use
an
operational
multiple
times
within
a
data
flow
right,
so
the
same
operation,
but
we
want
multiple
usages
of
it.
A
A
C
F
E
A
A
They
keep
changing
their
API
I'm,
seeing
yeah,
let's
see,
but
the
funny
thing
is
pip,
usually
won't
upgrade
tip
usually
doesn't
upgrade.
If
you
have
like
a
major
number
change,
I
thought.
So
it's
weird
that
it
upgraded
from
our
to
dot
something
to
three
I.
Don't
know:
okay,
we'll
fix
transformers.
There's
no
issue.
Oh.
E
A
A
A
B
A
Yeah
I
would
say
before
we
do
that,
though
we
should
figure
out
this
plugin,
because
you
know
the
Windows
stuff
is
yet
the
well
and
wasn't
there
also
the
issue
of
the
pipes
right
so
yeah.
So
there's
the
issue
of
the
pipes
and
so
I
think
the
first
step
here
is:
let
me
make
a
a
project
board
and
we'll
figure
out
won't
do
windows.
A
A
A
B
A
A
A
So
it
basically
just
goes
through,
and
so
you
can
see,
we've
got
this
plugins,
so
we've
got
you
know,
DF
of
all
plugins
and
then
we've
got
D
little
service
dev
and
in
plugins
we
just
list
out
all
of
the
you
know
all
of
the
plugins
that
we
have
right
and
then
so
that
we
have
this
complete
list
that
we
can
go.
Do
the
install
from
and
and
here
all
we
do,
is
basically
build.
A
A
Alright,
this
we
build
this
giant
command
right,
so
it
looks
at
all
these.
It
takes
these
turns
them
into
paths,
and
then
it
builds
this
giant
pipe,
install
command
right,
so
prefix
and
then
e
in
front
of
every
single
one
of
them
so
that
they
all
stall
install
on
development
mode.
So
it's
just
like
going
through
and
running.
You
know
it
install
prefix
or
not
prefix.
If
you
didn't
use
he
user
and
then
model
tensorflow
or
something
like
that,
and
it
just
does
all
of
them
right
so
model
Sai
kit,
and
so
let's
see.
A
A
B
A
B
A
A
Command
is
confusing
when
there
are
errors,
yeah,
I,
I'm,
seeing
I
I,
see
why
this
would
happen
is
because,
especially
some
of
these
now
have
some
more
complicated
appendices,
and
so,
if
we
don't
have
the
requisite
libraries
installed,
it's
gonna
blow
up.
So
it's
confusing
when
there
are
errors,
so
we
should
try
to
install
each
package
individually.
A
A
A
A
Another
thing
that
I
think
might
be
helpful
on
this
one
is
the
is
just
a
is
having
that
we
have
that
depth
dot
s
H
script,
that
we're
using
in
the
CI
to
install
all
the
dependencies,
and
maybe
we
should
we
should
probably
we
should
probably
make
it
so
that
someone
could
run
that
script
on
their
machine
right
and
then
end
up
with
their.
You
know
a
fully
set
up
machine
with
all
the
dependencies.
A
E
A
A
And
let's
try
to
get
this
in
before
the
next
release,
son,
because
that
will
be
well
I
guess
if
people
are
working
on
it
it
doesn't
really
matter
when
it
happens.
So
let's
see
yeah
we'll
just
you
know.
We
don't
want
to
done
that,
not
saying
it
doesn't
really
matter
if
it's
in
the
next
release
or
lot,
not
because
people
will
be
running
this
order
on
master,
ok,
ok
and
then
what
else
are
we
saying
here?
So
oh
yeah,
it's!
A
Okay,
do
you
guys
also,
you
guys,
also
see
the
value
in
this
one
or
do
you
think
maybe
there's
a
better
idea
to
do
this?
One
I
mean
we
could
make
this
one
like.
No,
can
we
make
this
one
its
own
command?
We
could
make
this
one
something
under
service.
Stuff,
I,
don't
know,
I,
don't
know
if
we
want
to
do
that
or
just
a
script.
What
do
you
guys
think?
A
G
A
A
A
I
think
this
one
was
almost
done.
I
think
I
had
a
patch
for
this.
Let's
see
alright
great
good,
good,
good
catch,
yesh,
ok,
so
and
then
the
documentation
plug-in
so
we've
discussed
this
basically
I
will
reference
the
project
board
here
for
Windows,
because
this
we're
gonna
kind
of
lump
this
all
on
your
Windows
support
and
convert
to
issue.
Oops
got
no
I
mean
have
these
dumb
formatted
issues
got
doing.
Ok,.
A
B
A
D
D
A
A
D
F
So
let's
say
we
have
a
sentence
and
we
want
to
remove
the
stop
words.
Basically,
like
is
Mr
a
and
all
these
kind
of
words.
So
we
have
a
operation
for
I
will
make
one
operation
for
that,
and
the
sentence
will
be
clean
by
this
operation.
Then
we
have
another
operation
that
will
map
all
the
words
to
M
buildings.
That
is
numbers
basically
every
odd
numbers.
F
A
A
A
A
So
in
these
cases
here
so
let's
take,
for
example,
the
first
one
right.
So
is
this
something
where
you're
gonna
use?
One
of
those
do
you
have
did
I
can't,
remember,
don't
do.
Is
there
some
code
in
one
of
those
models
that
you
alright
great?
Is
there
some
code
in
one
of
the
models
we
have
already,
that
does
cleanup
of
words
like
this,
or
is
this
something
new
you're
gonna
write?
A
D
A
A
E
A
All
right
and
then
this,
let's
see
okay,
so
this
one
and.
F
E
A
Okay,
it
sounds
to
me
like
with
this
first
case.
This
is
pretty
simple.
I
mean
this
is
basically
just
the
function
that
does
this
and
then
you
put
a
top
on
it
right.
So
writing
that
operation
is
it's
not.
This
is
trivial.
This
one.
This
looks
like
something
where
we're
going
to
have
a
config
and
the
config
object
for
the
operation
is
going
to.
You
know
going
to
contain
that
dictionary.
That
has
that
embedding
mapping
and
then.
A
F
F
A
All
right,
okay
and
then
so
basically,
the
flow
here
is
that
we
have
a
you
know.
We
have
a
you
know
a
bunch
of
sentences
and
our
these
are.
We
is
this
like,
so
we're
looking
at
a
classifier
right,
and
so
this
is
something
like
the
sentiment.
Analysis,
classification,
right,
okay,
so
essentially,
you're
gonna
have
the
first
operation
to
remove
the
stop
words
and
then
you're
gonna
have
the
operation
to
do
the
embedding,
and
then
you
train
the
model
right,
yeah,
okay,
so.
F
F
A
A
E
F
It's
not
labeled
protein,
so
basically,
just
like
you
have
a
board
and
then
you
can
have
a
vector
that
is
three-dimensional.
So
it
is
something
learned.
So
if
you
have
read
about
word
to
that
SIBO
models.
A
word
to
that
is
a
pretty
standard.
If
you
heard
about
that,
so
basically
the
it
will
convert
the
word
to
a
vector.
So
it's
word
to
its
name.
C
C
A
A
A
A
Okay,
I
think
this
is
kind
of
a
different
use
case
here,
but
I
think
I
do
I,
do
see
what
you're
saying,
because
what
he's
doing
is
within
the
operation
he's
the
operation
itself
is
going
to.
You
know,
do
each
word
in
the
sentence:
it's
gonna
look
it
up
and
and
have
some
file
that
has
you
know
a
vector
of
numbers
associated
with
that
word
and
then
it's
going
to
output
a
new.
A
F
A
Yeah,
so,
if
you're
looking
at,
if
you're,
if
you're
saying
like
what
is,
if
you
just
took
this
feature
data
in
a
JSON
file
right
and
fed
it
to
the
tensor
flow
model,
as
you
know,
each
feature,
what
is
is
the
feature
that
you're
only
feeding
in
one
feature
and
that
feature
is
that
array
of
arrays
or
is
the
there
is
each
feature?
The
word.
A
F
You
have
to
simplify
this,
we
can
see
like
if
you
have
a
sentence.
We
can
have
a
five-word
sentence.
10
word
sentence,
so
we
will
take
a
world,
we
can
say
unknown,
and
this
end
the
longest
sentences.
Ten
words
and
all
the
small
and
dense
will
be
that's
right,
unknown,
unknown
unknown
that
all
the
places,
okay
and
and
and
note.
A
F
A
A
A
Yeah,
that's
exactly
what
has
about
insane
yeah
because
yeah.
This
looks
like
something:
that's
a
model
config
option
right
to
me.
It
sounds
like
right,
so
yeah.
So
let's,
let's!
Let's
do
that?
Okay,
so
because
they're
else,
yeah
we're
gonna
have
to
create
a
bunch
of
features
and
then
the
models,
basically
just
gonna,
see
I,
don't
support
more
than
this.
Many
features
right
and
the
user
is
gonna
have
to
yeah
okay.
A
A
Okay,
so
so
the
data
flow
is
to
have
operation,
one
clean,
the
sentence
operation
to
does
the
encoding,
and
then
we
feed
it
to
the
model
for
training
of
prediction.
Okay,
so
then
this
does
end
up
as
a
pretty
direct
mapping
to
what
was
done
with
miss
I
missed,
because
you're
gonna
basically
modify
the
feature
in
place.
There
modify
their
feature
in
place,
you
know,
but
you're.
A
A
A
E
A
A
A
A
Okay
yeah,
but
this
isn't
something
that's
well
documented
here,
unfortunately,
I'm,
sorry,
where
is
it
actually?
Okay?
Here
it
is
yeah.
This
is
just
like
a
it
basically
wraps.
It
takes
your.
This
is,
let's
see
what
is
it?
Oh
yeah.
It
takes
a
data
flow
and
then
basically
it
looks
at
the
output
of
the
data
flow
and
the
data
flow
is
basically
just
gonna,
be
your
output
operation,
probably,
and
then
it
just
looks
at
the
output
of
that
and
in
every
maps,
whatever
you
want
it.
So
this
is
not.
A
A
F
A
E
A
A
C
A
A
A
A
G
A
A
E
A
Okay,
so
regressive,
and
what
else
was
I
saying:
oh
yeah,
what
to
do
next?
Okay,
let's
see,
let's
actually
just
check
this
milestone
here.
So
basically,
I
am
going
to
be
getting
the
compliance
test
done
today,
because
today
is
the
last
day
that
I
can
use
the
the
static
analysis
tool
that
I
have
automation
around,
so
we
are
releasing
next
week.
So,
let's
see
about
what
we
need
to
get
done
here,
Hritik
method
should
take
sources.
That
is
a
very
good
one.
A
This
would
be
a
very
good
one
to
do
so
like
if
I
know,
I
know
you
want
to
do
model
stuff.
So
if
you
want
to
do
model
stuff
great,
if
you
have
just
interest
in
if
if
you
could
possibly
do
this
one,
this
would
be
great
so
cuz,
you
know.
Obviously,
if
we
need
to
release
so
we
need
things
to
be
sort
of
cleaned
up
right.
G
A
Cool
great
thanks-
and
this
should
be
basically
I
mean
this
is
kind
of
it's
gonna-
be
a
little
bit
tedious,
but
it's
basically
just
go
through
and
and
make
sure
that
we're
using
sources
which,
with
features
in
each
of
these
model
and
each
of
the
models
right
and
and
so
the
trick,
is
basically
making
sure
that
we
we
don't
have
copy-paste
errors
on
on
our
sources,
which
features,
call
and
predict
because
actually
no
yeah.
It
should
pretty
much
always
just
be
self
top
features
or
self,
not
features,
duh
config
names,
because
basically
I
don't
know.
A
I
think
I
might
not
have
said
this.
But
what
I
realized
was
that
we
just
like
pass
records
to
predict
and
there's
no
guarantee
that
any
of
those
records
have
have
the
features
that
we
that
we're
asking
to
model
the
to
train
on
and
well
that
that
can
obviously
make
it
blow
up
and
weird
and
interesting
ways
rather
than
just
having
those
sources
which
features
which
now
has
that
validation
on
it
to
explode
and
say,
didn't
have
any
of
those
features.
A
So,
let's
see,
let
me
and
this,
but
this
one's
probably
gonna,
be
quick
for
you.
So
let
me
like,
let
me
give
you.
Let
me
give
you
some
some
other
stuff
here,
so
looking
for
what
to
do
next,
Wow,
okay,
well,
I
can't
believe
I
was
I.
Haven't
looked
at
that
3.8
release
milestone
in
a
while
we're
actually
getting
pretty
close.
There
I
thought
it
was
the
weekend
now
it
today.
It's
not
the
waking
of
God
I
have
a
bunch
of
to
do
this.
A
Don't
write
a
and
I
still
have
Google,
okay,
so
looking
for
what
to
do
next,
that
would
be
a
good
one
and
then,
let's
see
source
D
F
yeah.
We
need
to
do
that.
Let's
see
integration,
I
think
I'll,
probably
tackle
the
integral
rest
of
the
integration.
I
think
we
talked
about
this
agony
of
stuff
to
do
so
and
then
channel
config,
okay,
ooh
and
then
let's
see,
did
it.
Oh
I
have
one
more
thing:
I
tell
you
guys
more
about
stuff
like
so,
let's
see
enable
Windows
and
CGI
HTTP
testings
updating.
A
Let's
check
out,
oh,
you
know
what
might
be
good
here
is
well,
this
is
XG,
but
this
has
weird
dependencies.
It's
not
critical,
yeah
plot
ml
would
be
good,
I.
Think
plot
MO
would
be
good,
I
think
as
a
next
target.
You
know.
So,
let's
see,
let's
look
and
make
sure
that
it
actually
looks
as
good
as
I
really.
A
A
A
A
A
Yeah
I
don't
really
see
an
obvious
way
to
wrap
this.
It
sounds
like
okay,
so
maybe
looking
at
this,
if
this
is
like
so
it
looks
like
they've
got
this
thing
where
they're
they're
running
on
top
of
Karis
here
right.
So
the
right
approach
for
this
might
be
to
do
Karis
first
like
something
that
that
runs
on
top
of
Karis
and
then
basically
say.
If
you
have
flat
MLM
stalled,
you
know
optimized
to
be
a
plateau
mill.
A
Let's
see
does
that
make
sense
what
I
was
saying
that
so,
basically,
because
this
the
extent
of
this
code
right
here
is
basically
just
using
Karis
and
then
they
say
you
know
plot
a
mile,
Karis
back
and
right
and
then
I,
don't
think
we
see
any
more
specifics.
Other
than
compiled
might
be
I'm,
not
sure
whether
that's
a
cursing
or
a
plot
Amell
thing,
but
I'm
kind
of
just
assuming.
That
might
be
a
one.
F
F
A
A
A
A
B
A
A
A
G
A
A
So
I
would
just
start
it
with
something.
That's
like
you
know.
I
would
make
this
I
know
I
mean
I,
don't
know
what
the
hell
is.
Let's
see.
Oh
this
is
that
data
set
okay,
I,
see
okay
and
then
vgg
19
I
assume
is
some
kind
of
implementation
of
that
model.
For
that
data
set
okay,
so
yeah
I
guess
you
could
just
do
a
straight
wrapping
of
this
model
here,
basically
just
like
this
and
then,
of
course,
you'd
need
to
figure
out.
We
need
to
figure
out
what
the
format
well.
A
I,
guess
you
don't
really
need
to
care
about
the
format
of
the
data
set
just
for
the
test
cases
and
stuff,
but
once
we
get
farther
along
we're,
gonna
need
to
figure
out
how
to
how
to
you
know,
I
think
that'll,
probably
work
well
with
some
of
the
stuff.
Psalm
is
doing
and
say
you
know.
How
do
we
add
different
things
here
right
rather
than
just
the
cats,
so.
C
C
A
So,
and
it
looks
like
I,
don't
know
from
from
the
from
from
what
I'm
getting
from
this.
We
have
cats
and
not
cats,
so
we'll
probably
have
to
do
something
where
we'll
probably
we're
going
to
want
to
use
your
directory
source
and
stuff.
When
we
do
a
full,
like
example,
usage,
you
know
in
the
documentation
with
the
SH
files
and
stuff,
but
for
now
just
wrap
it.
F
A
F
F
A
A
I
see
so
we
support
all
carats
application
networks
from
current
version,
2.0
validated
networks
tested
for
performance
and
connected
correctness
as
a
part
of
our
containers,
integration
system.
Okay,
so
there's
just
no
there's.
No,
these
are
the
implementations
that
they
have
tested,
but
there's
no
there's
no
guarantees
on
the
other
ones,
but
technically
they
should
work.
Is.
F
F
A
B
A
B
A
B
E
F
B
B
What
happens
you
just
load
the
breeding
model
and
provided
the
weights?
You
already
have
saved
them
like
you
already?
Have
them
retain
model,
and
then
you
just
remove
the
last
couple
of
layers
from
those
preteen
models
and
put
on
some
other
layers
on
Keros
and
they
are
just
ready
to
use
for
other
applications.
Okay,.
F
F
A
F
F
A
Is
what
so
you're
saying
that
so?
Okay,
so
that,
when
we're
instantiating
this
VG
g19?
What
we're
saying
is
that
it's
loading
in
the
model
with
its
layers
and
then
it's
loading
in
the
weights
and
then
and
then
compiled,
is
doing
some
sort
of
like
what
is
compiled
in
here.
I'm,
not
familiar
with
Tara.
B
A
B
A
A
F
A
That's
I
mean
that's
that's
the
thing
is
like:
is
it
a
typo
or
is
it
not
a
typo
right
because
I
think,
because
that
doesn't
yeah
what
doesn't
make
any
sense?
Let's
see
because
well
then
the
thing
is
that
they
pass
the
X
train.
A
E
A
B
B
A
A
Yeah
I
mean
I,
guess
the
question
here
is
like:
can
we
get
these
like?
Do
we
know
how
to
get
these
trained
models
to
do
what
you
did
Himanshu
and
drop
the
non?
You
know
irrelevant
layers
and
stuff
and
then
use
them
for
other
things
right,
because
there's
really
no
use
to
us
and
wrapping
something
that
that
just
you
know,
does
some
sort
of
predefined
data
set
right.
We
want
people
to
be
training
their
own
data
sets.
A
B
F
A
A
Gonna
do
when
we
yeah.
We
want
to
make
it
so
that
I
mean
that
the
main
domain
goal
when
we
wrap
that
is
going
to
be
make
it
easy
to
use
from
the
command
line
right,
because
that's
sort
of
one
of
our
main
interfaces
here
is
using
things
from
the
command
line
and
then
secondary
to
that
would
be
things
like
the
HTTP,
API
and
stuff.
It's
going
to
have
the
same
interface
on
top
of
right,
so
so
yeah.
A
A
A
A
A
A
Let's
see
all
right,
we're
gonna
I'm
going
to
take
this
offline
because
you've
got
we've
spent
a
bunch
of
time
on
this.
So
basically,
if
you
want
to
do
X
G
boost
there's
some
code-
that's
up
there
for
now
like
if
you're
just
like
you
know
you
run
out
of
stuff
to
do
in
your
you're.
Looking
for
more
or
more
models
like
then
X
cheapest
could
be
good.
A
A
A
G
A
A
Let's
see
ok,
so
priority
wise
then
come
on.
Give
me
that
priority
wise.
Then
it
would
go
like
this.
I
would
say,
and
it's
not
gonna
have
to
it
that
whatever
so
alright
so
and
then
guys
so
where
we've
gone
us,
is
that
pretty
much
it
for
you
suit
on
shoe?
Is
there
anything
else
you
want
to
talk
about
today,
yeah.
G
A
A
G
D
A
A
D
A
D
A
A
A
D
A
A
A
A
D
D
A
D
A
All
right
there
we
go
so
we
got
something
so
pretty
cute
value
for
salary;
okay,
so
yeah
we're
getting
a
debug
output
here.
Interpretation
message:
blank:
okay,
so
examples,
dataflow
chat,
BOTS,
clients,
okay,
so.
A
So
message
return
message:
a
CDL
to
get
value;
okay,
redirect
us
to
do
an
STD
out;
okay!
Well,
oh,
oh!
Yes,
okay!
So
this
is
what
I
was
gonna
say.
So
the
thing
is
that
it's
not
gonna
be
in
standard
out
because
check
this.
So
when
you
do
CLI
you're
actually
going
to
get
the
results,
it's
it's
gonna
just
be
you're.
Just
gonna
get.
A
A
See
so
this,
so
this
is,
let's
say:
where
is
seal
I
called
seal?
A
call
is
in
Maine,
so
yeah.
So
what
happens?
Is
we
call
CLI
from
the
main
method,
which
is
basically
the
thing
that
gets
called
like
as
the
entry
point
for
the
console
scripts
entry
point,
which
is
what
you
know:
Python
can
turn
now
any
function
into
something
that
gets
called
from.
A
The
console
is
like
a
binary
right,
so
it
basically
grabs
the
result
and
then
says
like
this
is
what
we
talked
about
last
week
with
suction
I'm
having
pretty
printing
still
outputting
this
stuff.
So
if
the
result
is
command
output
override,
then
it's
not
going
to
dump
or
if
it's
none
or
display
help.
Otherwise
it's
going
to
jump.
Dump
the
JSON
as
its
gonna
dump
the
date
JSON
the
standardout
right.
A
A
D
A
Right,
okay,
so
I
think
we're
good,
then
we're
anything
yeah,
okay,
so
and
then
it
format
at
the
whole
damn
thing
so
just
basically
switch
switch
main
or
CLI
from
underscore
main.
Okay,
so
let's
say
yeah
because
or
unless
you
want
to
you
know
like
if
you,
if
you
want
the
results
like
as
records,
then
you
can
keep
it.
But
you
know,
if
you
don't,
then
switch
if
you
just
want
that.
You
probably
just
want
the
JSON
output
to
just
dump
back
for
now
right
so
which
I.
A
D
D
D
D
A
D
D
D
A
D
A
D
D
A
A
D
D
A
A
G
D
A
Yeah,
so
I'm
not
I'm,
not
entirely
sure
about
about
that;
sometimes,
usually
they
they
will
expire,
even
if
you
use
them
like
if
you're
still,
if
you
make
another
post
request
right,
your
your
initiative,
ala
dates
it
on
the
initial
connection,
I'm
assuming
not
on
the
not
on
the
not
on
just
the
the
stream
and
kicks
you
out
after
that.
Long
right,
I,
don't
know,
I
mean
this
is
all
an
assumption
right.
So
we
don't
we
don't
know,
but
I
would
I
would
venture
to
guess
that
your
post
request
that's
initiated
to
post.
D
A
A
A
A
D
D
A
A
D
A
A
A
I'm
trying
to
understand
whether
that
is
yeah,
okay,
personal
access,
token,
because
I'm
trying
to
understand
whether
we
need
to
do
that
o
auth
flow,
because
both
is
usually
you
know
when
we
do
OAuth
it's
because
we're
trying
to
have
our
service.
Our
service
has
been
delegated
permissions
of
a
user
to
post.
On
behalf
of
that
user.
Right
and
let
me
I
gotta,
grab
my
charger
and
in
this
case
we're
not
really
hosting
like
we're,
not
really
put.
We
don't
really
want
to
post
on
behalf
of
our
user.
A
D
D
D
B
D
A
Answer:
yeah
I,
don't
know
if
we
should
use
that,
because
if
you
look
at
the
maintenance
on
that
thing,
obviously
it
hasn't
been
touched
for
a
while.
You
know
it's
probably
stable,
but
it's
also
not.
It
also
is
not,
as
it
does
not
have
a
sink
io
support,
so
we
could
use
you
know,
running
executor
and
stuff,
but
for
what
we're
doing.
D
A
A
A
D
A
D
C
A
C
C
A
C
A
G
C
A
C
A
A
So
when
you
support
operation
instances
for
the
CLI
create
command,
so
if
we
see
equals
so
we
need
to
do
something
where,
let's
see,
we
need
to
do
something
where
we
can,
like
you,
know,
sort
of
like
how
we
have
those
labeled
or
like
the
like
yeah.
We
have
labeled
models
or
labeled
sources.
Where
we
say
you
know
blank
equals
CSV
or
something
right.
So
we
need
something
or
it's,
like
instance:
name
equals
operation,
name.
A
C
A
C
A
A
A
E
E
A
Oops
new
tabs
and
spaces
and
spaces
and
tabs
damn
it.
Okay,
hopefully,
Gator
fixes
this
for
us,
don't
make
it
a
giant
mess
and
it's
all
right.
Okay,
all
right!
I'm!
Sorry,
that's
a
giant
tabby
mess,
let's
see,
but
there
is
actually
way
we
can
fix
this
real,
quick.
All
right
is
that
pretty
much
all
you
need
it
there.
Yes,.
H
A
E
A
A
A
A
Cuz
you're
gonna
do
operations
equals.
You
know
the
basically
OP
dot
name
up
unless
and
you're
probably
going
to
want
to
make
so
for
load
operations.
Let's
the
operations,
you're,
probably
gonna,
want
to
make
this
a
dictionary
here
right
and
then
to
do
if
you
see
an
equals.
So
this
should
be
a
key
value.
A
If
you
not
equals
use
operation
named
instance,
name
otherwise
use
instance
name
and
then
let's
see,
you're
gonna
need
to
call
you're
going
to
need
to
call
data
flow,
dot,
Auto
flow
or
whatever.
That
method
is
because
auto
will
do
the
flow,
and
if
you
specify
the
operation,
instances
I,
don't
believe
it
or
no
I.
Think
as
long
as
flow
is
not
specified,
it's
going
to
yeah
I
think
as
long
as
flow
is
not
given.
A
No,
it
does
it
goes
through
and
it
says
it
grabs
the
so
this
might
be
something
you
need
to
do
to
is
look
at,
because
I
think
this
allows
you
to
pass
implementations
and
I
do
believe
we
load
implementations
if
they
exist,
so
you
might
want
to.
You
might
want
to
just
add
this
logic
to
to
the
anit
method.
A
A
Plantation
set
default,
update,
operation,
name
elephant:
let's
see,
Southside
operations
update
operations.
Ok,
it's
already
going
to
do
that
for
you
so
soft,
all
right,
yeah,
it's
gonna,
don't
be
fined
so
basically
just
map
the
equals
the
instance
names
and
then
the
rest
of
that
command
should
work
all
right.
Anything
else
from
anyone
today.