►
From YouTube: Weekly Sync 2020-04-28
Description
Meeting Minutes: https://docs.google.com/document/d/16u9Tev3O0CcUDe2nfikHmrO3Xnd4ASJ45myFgQLpvzM/edit#heading=h.pss1gp67y2cr
A
We're
it's
sort
of
like
an
ongoing
thing,
so
I
like
the
way
it
works.
Is
we
weave
right
up
basically
like
what
we
did
and
then
what
we're
going
to
do
and
that's
sort
of
like
feather,
is
to
the
performance
report.
It's
kind
of
this
this
day
they
have
this
new
system
and
we're
all
sort
of
like
not
so
sure
about
this
new
system
and
how
it
works.
That
I
mean
I,
think
I
think
it'll
be
ok
but
yeah.
Thanks
for
asking.
A
A
A
B
A
So
well
so
yes,
we've
we've
figured
this
out
and
I
think
I've
solved.
The
problem
I
may
have
created
some
problems
in
the
process,
but
we
will
see
the
tests
are
passing.
I
want
to
run
a
few
more
little
tests
before
I,
merge
everything
but
I
think
I
thought
I
fixed
every
single
example
here,
so
they
all
should
be
able
to
be
copy
pasted
into
into
their
own
file.
So.
A
Help
with
sub
flows
and
should
I
so
didn't
get
that
one,
but
I
have
this
other
reason,
I
didn't
do
that.
One
is
reason
this
one
did
get
done
is
because
John's
introducing
this
because
there's
going
to
be
a
new
function
in
high-level
WI,
called
run,
which
is
responsible
or
which
runs
data
flows,
and
so
I
was
going
to
use
that
function
in
doing
that
patch.
But
so
this
should
be.
Basically,
this
is
coming
down
the
pipe
and
as
soon
as
I
do
some
last
validation.
A
Here
then
I'll
go
ahead
and
merge
it
and
then
they'll
be
this
high
level
run
function
and
stuff
and
the
examples
will
be
fixed
and
what
else
what
else
happened
here?
We're
getting
close
to
600
commits
oh
yeah.
There
was
something,
though
there
was
a
race
condition
and
the
data
flows
green
things
up,
sometimes
I
fixed
that
yeah.
So
that's
the
only
other
interesting
thing
here.
All
right,
so
did
you
guys
have
anything
you
wanted
to
that
so
section
armor,
you
basically
stuck
on
the
on
the.
B
A
A
A
A
A
A
Okay,
so
there's
one
thing:
I
guess:
I
didn't
catch
this
when
we
did
review
of
this
guy,
but
so
similarly
to
how
you
do
this
after
an
a
exit-
and
you
create
a
new
I-
can't
do
that.
Well,
I
can't
do
a
suggestion,
but,
similarly
to
how
you
do
a
enter
in
it
and
I
basically
create
a
new
context
for
the
source.
You're
gonna
want
to
create.
So
so.
Well,
you
don't
create
a
new
path.
You
enter
the
context
of
the
source
here,
alright,
so
the
source
is
already
instantiated
when
it
exists,
something
config.
A
A
A
Alright,
so
with
the
double
context,
entry
pattern
right
so
we
enter.
This
is
where
we're
entering
the
first
context
of
this
a
enter
every
time.
You
start
the
width
block,
you're,
hitting
a
enter
method
and
every
time
you
in
the
with
block,
you
hit
the
a
exit
method.
So
what
we're
doing
this
is
basically
saying
you
know,
start
the
width
block,
the
async
with
block
on
the
source
and
then
call
assign
that
to
self
dot
source,
and
here
we're
basically
doing
the
second,
this
block.
A
A
A
A
A
A
A
Okay,
so
let's
see
so
you
create
the
source
context,
you
create
the
features,
I
believe
that
damn
I
keep
trying
to
hit
suggestion.
I,
don't
think.
There's
an
async
for
here.
I
think
this
is
gonna,
be
yeah
I'm,
pretty
dang
sure
that
this
features
is
just
a
these.
Just
is
just
a
list,
I'm
pretty
sure.
A
A
Okay,
this
one
doesn't
yeah
this,
isn't
that
helpful
right
now:
okay,
yeah
just
just
this-
is
for
typing
stuff,
so
yeah,
okay,
this
looks
good.
Let's
do
this
house
features
inputs,
correct
vision,
oh
okay,
we're
looking
for
all
the
records
and
records
so
grab
every
record
and.
A
A
B
A
A
A
A
There's,
okay,
so
just
just
just
not
to
confuse
you,
but
there's
I
just
want
to
let
you
know
about
something
that
we're
going
to
do
in
the
future.
So
there's
an
open
issue
to
have
this
memory
Orchestrator
dot
or
to
have
the
running
of
a
data
flow
except
a
asynchronous
iterator
as
the
as
the
input
and
when
we
get
that.
Basically,
what
we're
going
to
do
is
we're
going
to
be
we're
basically
just
going
to
wait.
Wait
for
each
we're
going
to
wait,
we're
going
to
have
the
run
method
itself.
A
A
Exponents
but
we'll
basically
end
up
with
like
when
the
orchestrator
runs,
we'll
pass
it.
This
generator
function,
that's
going
to
don't
worry
about
it.
It's
gonna
be
kind
of
hard
to
expand
it.
So
sorry
about
that,
but
okay,
so
for
now
what
we're
gonna
do
is
we
can
take
this?
Let
me
just
write
it
here:
real
quick,
because
I'm
gonna
need
them.
Okay,.
A
A
All
right,
so
this
is
basically
what
it's
going
to
look
like
here
is:
should
we
can
just
sort
of
check
this
and
make
it
one
I?
Don't
like
it
cleaner
or
not?
But
let's
see
what
it
looks
like
here
all
right,
so
we
passed
the
data
flow
to
run,
and
then
we
pass
a
dictionary
where
the
keys
are
going
to
be
all
the
record
keys
and
not
records,
but
record
so
using
for
record
records,
feature
name,
yeah,
feature
name,
okay,
so
we
say
here
give
the
first
arguments
the
data
flow.
A
A
There's
two
ways:
there's
two
ways
we
can
do
this:
we
can
the
way
that
I
was
hoping
to
do
this.
Is
we
take
the
oh?
We
can
actually
do
that.
Can't
wait.
Okay,
so
there's
a
couple
ways:
we
can
do
this.
We
can
go
through
this
Basin
for
it.
Let
me
just
write
the
other
one,
so
you
can
see
what
the
options
here
are.
A
We
could
go
through
the
records
and
for
each
record
we
just
give
we
do
this
right.
We
say
we
say:
okay,
we
we
loop
through
the
records
and
for
each
record
we
just
create
adjectives,
but
what
single,
dict
and
or
what
we
don't
even
need
it,
the
dict
we
can
just
say
we
can
just
give
it
a
list
in
this
case
right
and
we've
got
okay.
So
this
is
your
list
of
inputs
for
this
record
right
and
this
part
stays
the
same
and
we
do
I.
Guess
we'll
just
do
this
for
now.
So.
A
All
right,
so
we
go
through
all
the
features
and
we
create,
and
we
could
probably
you
know
you
probably
do
this
somewhere
else,
so
you're
just
doing
a
lookup
here
and
you
say:
okay,
give
me
the
value
is
going
to
be
whatever
you
know
the
record
stored
as
the
value
for
that
feature,
name
right
and
so
that
it's
going
to
be
this.
This
call
right
here
right.
So
if
I
call,
if
I
have
this
reload,
the
page
okay
well
come
on
all
right
whatever.
A
So,
if
I
have
this
record
example
with
the
features
of
dead
beef
and
I,
say
oops,
this
is
features
yeah.
If
I
have
this
record
and
I
say
feature
and
I
passed
the
feature
name,
then
I
get
be
afraid,
so
I
get
the
value
so
we're
gonna
go
through
and
we're
gonna
say
if
it
go
through
all
the
this
is
a
generator
that
basically
creates
this
input,
objects
and
yeah.
A
A
A
This
is
not
quite
the
ideal
because
we're
generating
this
giant
this
giant
dictionary
here,
but
we'll
solve
that
sort
of
in
the
future
by
by
just
passing
a
generator
that
sort
of
creates
these
lists
of
inputs
on
the
fly.
That's
what
I
was
trying
to
explain
earlier,
but
so
for
now
this,
where
we
basically
were
creating
the
orchestrator
once
and
we're
passing
every
single
record
to
it.
Now,
when
we
pass
every
single
record
to
it,
it
it's
going
to
evaluate
them
all
at
once,
but
they're
going
to
come
out
out
of
order.
A
They
may
not
come
out
in
the
same
order
that
they
came
in
and
basically,
whenever
they're
done,
we're
gonna
need
a
way
to
grab
the
record
from
the
you
can.
You
can
actually
create
a
context
instead
of
these.
What
this
does
is.
It
creates
a
what's
called
a
string
and
puts
that
context,
and
so
basically
all
of
these,
this
array
of
inputs
runs
and
the
Condor.
A
Basically,
this
array
of
inputs
is
not
allowed
to
coexist
with
any
other
inputs
that
are
for
different
record
keys,
so
this
array
of
inputs
is
only
allowed
to
intermingle,
with
other
inputs
that
are
supposed
to
that
are
better
with
in
this
record
keys
scope,
and
so
so
they
all
basically
that
what
ends
up
happening
is
that
they
all
run
it
once
and
we'd
have
to
greet
another
context
object,
because
this
is
creating
this
string.
Input
set
context.
A
What
we'd
want
to
do
is
we
want
to
create
a
record
input
set
context,
that's
something
that
has
been
also
an
open
for
a
while,
but
the
so
this
is
nice
because
it
evaluates
all
of
them
all
at
once,
but
it's
not
so
nice,
because
if
you
have
a
really
giant
source
now
we're
going
to
end
up
creating
this
dictionary,
where
we
pull
the
entire
source
into
memory,
which
is
like
the
point
of
this.
Is
that
we're?
Never
we
never
want
to
like
well.
A
Okay,
the
problem
is
right
now
the
problem
is
that
most
of
our
things
actually
do
pull
the
entire
source
into
memory,
but
in
the
future
the
idea
is
that
we
never
want
to
pull
the
entire
source
into
memory.
That's
not
something
that
we
that
we
want
to
be
doing
it's
just
something
that
we
happen
to
be
doing
right
now,
so
I
think
in
this
case,
we
probably
want
to
go
with
this
first
approach,
especially
since
we
don't
have
that
generator
support.
A
A
A
A
A
A
A
Yeah,
it
basically
takes
a
sub
flow
and
runs
that
sub
flow
and
then
renames
things
to
get
you
whatever.
You
want
the
right
names
to
be,
and
so
obviously
you
might
you
might
end
up
with
like
if
you
do,
if
you
call
get
single
and
stuff
like
that,
you
might
end
up
with
right.
If
you
call
get
single
so
you
rent,
you
run
some
sequence
of
operations.
A
You
rent,
you
run
some
data
flow
and
the
output
data
types.
Are.
You
know
they
be
out
the
definitions
that
you
care
about
using
as
the
features
now
might
be,
of
some
other
name
right,
that
it
isn't
the
original
feature.
Name
now
so
now
all
you're
going
to
do
is
you'd
want
to
run
get
single
where
you
pass
like,
for
example,
this
one.
We
wanted
to
format
our
string,
the
output
from
this
thing,
so
we
wanted
this
message
right.
So
we
want
to
turn
that
into
so,
oh
yeah.
We
want
to
call
that
response
right.
C
A
A
A
A
D
D
A
D
A
A
A
A
A
We
used
to
be
able
to
run
it
with
locally,
because
in
a
virtual
environment,
now
it
just
sort
of
like
slobbers
all
over
everything,
and
so
it's
not
really
the
best
to
run
things
locally,
but
you
could
sort
of
modify
it.
Basically,
if
you
wanted
to
use
I,
don't
know
if
you
can
use
so
you
could
create
a
wrapper
script,
which
runs
would
run
dot.
Css
run
that
Sh
within
a
docker
file
or
than
a
doctor
container.
The
Anaconda
wand.
A
Create
a
wrapper
escape,
which
would
run
the
CIA
rather
than
a
docker
container,
and
then
have
the
CI
or
have
github
actions
run
that
script
instead
of
Ronda
SH.
This
would
make
it
so
that
we
could
run
the
CIA
yes
locally
to
because
they're
in
containers
or
you
could
see
if
there's
a
way
to
install
anaconda
within
the
github
actions
environment
and
not
fix
the
issue
of
no.
A
No
way
to
run
CI
locally,
so
basically
you
could
just
now
that
you
know
that
anarchy
everything
works
with
anaconda.
You
can
sort
of.
You
have
two
options
you
could.
You
could
use
that
docker
file
where
you
or
you
use
that
docker
image
where
you
know
everything
works
and
you
can
make
it
so
that
the
CI
runs
that
image
and
then
consequently
we
can
also
run
that
locally.
A
Now,
because
then
those
docker
images
locally,
and
then
you
fix
that
issue
in
the
process
or
you
could
sort
of
you
could
just
try
to
install
anaconda
within
or
find
some
kind
of
good
I
think
there
are
other
github
actions
out
there.
There
might
be
you
enough,
an
official
one
that
would
be
in
akanda,
actually
I
think
there
was
not
I.
D
D
I'm
not
sure
that
I
have
a
I
saw
fields
they're
using
it.
Okay,
cool
cool.
A
D
C
A
D
D
B
A
A
A
A
A
A
C
A
A
Let's
see
did
we
come
up
with
something
less?
Oh
yeah,
you
had
the
file
service
tutorial.
Okay,
let's
see,
let's
see,
but
else
if
we
got
here
and
I
just
saw
a
bunch
of
things
and
I
should
have
created
issues.
Let's
see
what
were
they,
what
were
they?
What
were
they.