►
From YouTube: Weekly Sync 2020-06-12
Description
Meeting Minutes: https://docs.google.com/document/d/16u9Tev3O0CcUDe2nfikHmrO3Xnd4ASJ45myFgQLpvzM/edit#heading=h.69876fr9qbik
A
C
B
B
A
D
E
A
A
A
Hashim
and
I
ran
into
this
a
while
ago
what
warehouse,
James
Ben.
We
got
a
ping
him
and
see
where
he's
been
so
we
went
into
this
a
while
ago,
where
it
was
like
throwing
some
weird
errors
like
it
didn't
have
the
multi
processing
module
installed
and
I
think
it
has
to
do
with
when
there's
connection
issues
when
it's
doing
updating
it,
just
sort
of
drops
files,
and
so
then
you
end
up
with
a
not
entirely
working
Python
install
with
in
mini
condo,
which
is
the
one
that
we
end
up
using.
A
So
the
solution
of
this
I'm
thinking
is
to
build
a
massive
docker
container,
so
history,
so
I'm
just
saying
it's
gonna,
be
huge.
I
have
four
point:
seven
gigabytes,
at
least
so,
let's
see
what
happens.
Doctor
push
Tucker
tag.
A
Access
to
requested
resources
denied
its
privates,
oh
I
haven't
created
the
repo
it's
been
so
long
since
I've
actually
pushed
something.
So
basically
what
I'm
thinking
we're
going
to
do
here
is
make
it
so
that
we
build
everything
from
a
docker
container
Wow,
who
the
hell
knows
what
my
docker
hub
has
to
be.
A
Okay,
well:
okay,
that's
that's
also
bad,
but
so
I'm
thinking
what
the
solution
here
is
going
to
be.
Is
we
build?
We
basically
take
all
the
dependencies
and
we
take
all
the
dependencies.
So
let's
look
at
the
file.
So
we
have
this
depth,
stop
py
file
or
depth
dot
Sh
under
CI
right,
and
it
knows
how
to
install
sort
of
like
specific
things
like
things
that
need
to
be
installed
from
Conda.
A
A
Putting
all
that
stuff
in
here,
and
so
this
is
basically
install
all
the
dependencies
that
you're
not
going
to
install
with
pip
and
so
then,
basically
in
an
effort
to
get
everything
in
one
place,
we're
gonna
have
here
is
like
so
this
is
install
the
dependencies
that
can't
be
installed
with
pip
and
then
for
each
project.
You
know
the
specific
ones
and
then
the
and
then,
and
then
we
also
go
through
and
just
install
every
single
dependency
that
there
is
so
so.
A
Basically,
we
read
another
little
script
in
here
that
basically
goes
through
and
reads
all
the
set
of
py
files
and
grabs
all
the
install
requires,
and
we
dump
them
into
this
dot.
Ci
requirement
txt
and
it's
gonna
be
kind
of
like
we
did
before,
where
we
had
to
update
the
docs
folder,
but
gonna
have
to
update
this.
We're
gonna
have
another
CI
job
that
basically
just
checks
and
said:
hey
you
didn't
update
this,
and
that
way,
this
docker
container
has
everything
that
we
would
need.
A
A
A
Workflows,
testing
IMO
so
basically,
because
now
we're
gonna
be
running
talker,
you
can't
pull
so
one
of
those
steps
was
pulled,
a
Maria
DV
image
and
that's
just
not
gonna
fly
anymore
because
you
can't
there
were
at
docker
demon,
it's
not
running.
We
can
do
darker
and
darker,
but
you
can't
do
Ducker
within
docker
when
you're
doing
docker
build
so
basically
we'll
just
set
up
everything
and
then
we'll
pull
it
within
the
github
testing
yamo.
And
that
way,
it's
all
it's
all
good.
So
basically
it's
it's
we're
done
working
on
it.
A
So
solution
move
to
docker,
and
hopefully
that
will
mean
that
you
know
when
we're.
When
we're
running
CI
locally,
we
can
just
pull
down
the
image
and
not
have
to
deal
with
running
the
install
of
everything
locally.
So
let's
go
with.
Let's
go
reverse
order,
so
yes,
so
we're
still
facing
those
sequel,
Lite
error
issues.
A
G
H
A
E
A
Alright,
ok,
so
it
looks
like
we've
still
got
my
HTTP
test
bugs
here.
It
looks
like
this
one's
a
new
one
that
I
think
it's
a
new
test
on
master
branch.
That's
now
failing,
ok
and
now
we're
talking
about
the
unopen
to
symbolic
link,
privilege.
Okay,
that's
skill
stuff,
and
this
is
the
main
one,
we're
concerned
with
here
right,
yeah,.
F
F
D
John
there
was
one
more
issue
with
the
model
and
l4p
white
sari
service
Jen.
Yes,
so
I
was
saying
problem
with
them
del
4py
model.
So
when
I
run
the
D
flat
formal
service,
they've
installed
user,
it
says
he
said.
Del
4
py
does
not
know
such
package
found
p
IP,
so
I
have
to
manually
install
using
the
condor
commanders
there
in
the.
H
H
A
Let's
see
this
is
weird
yeah
cuz
I
haven't
I
hadn't,
seen
this
until
a
couple
days
ago,
either
so
I'm.
Just
you
know
they
haven't
released
a
new
version
of
three-point-seven
anytime
recently
so
like.
Why
is
this
happening
all
of
a
sudden
and
argue?
Is
it
all
the
tree
specs
have
to
do
with
mini
Conda
like
do
they
save
mini
condo
in
all
of
your,
in
both
your
cases
trace
backs.
H
A
Let's
see,
okay,
okay,
that's
good
to
know
it's
very
important!
Now,
okay,
so
well,
maybe
all
the
stalker
workers
for
nothing.
Then
hopefully,
I
mean
it's
a
pain
in
the
ass
to
figure
out
Sam,
alright,
so
CI
air,
so
issue
is
not
localized
to
mini
Conda,
so
I
see
I
experiencing
on
local
all
machines.
Okay,
all
right.
A
A
This
is
weird.
This
is
very
weird,
okay,
so,
okay,
let's
see
so
all
right.
So
the
thing
is,
we
only
have
I
only
have
thirty
minutes
here.
So
I
think
this
is
an
issue
that
we're
probably
not
gonna
solve
right
now,
just
I
have
a
hunch,
it
seems
like
I
mean
I
was
working
on
at
all
last
night
and
I
couldn't
see
what
was
up.
A
So
my
guess
is
we'll
probably
need
to
skip
that
one
right
now
so
and,
and
it's
not
really
affecting
us,
we
know
that
we
know
that
it's
just
a
it's
just
a
bug
right,
so
what
we
can
probably
do
is
as
a
stopgap
measure.
Let's
just
skip
that
test
so
because
I'm
I
hesitated
for
us
to
spend
too
much
time
on
this
thing,
especially
if
we
know
that
the
test-
actually
it's
mostly
working,
there's
just
something
weird
about
it.
So
let
us
just
okay
so
which
test
was
it?
A
No,
so
it's
passing
on
the
master
branch
of
the
CIA,
so
this
test
was
introduced
like
a
few
commits
ago
now,
but
it
probably
hasn't
had
that
many
CI
runs
on
the
master
branch
between
now
and
then
yeah.
It
was
introduced
11
days
ago,
and
it's
only
had
one
two
three
CI
runs
on
the
master
branch
since
then,
so
it
might
just
be
like
a
non-deterministic
thing
that
didn't
show
up
until
now,
which
is
probably
what
it
is.
But
then
it's
got
to
be
an
error
for
some
reason
right.
A
A
A
A
A
A
A
A
F
C
A
A
I
A
A
A
A
A
We
create
a
new
context,
object
and
then
we're
actually
entering
the
context
on
this
DB
context,
which
doesn't
actually
do
anything
so
we
have
to
like,
for
example,
if
we
took
this
stuff
out
of
here,
then
there
would
actually
be
no
more
context
because
we'd
be
outside
the
orchestrator,
which
was
what
was
doing.
You
know,
that's
what
was
actually
making
it
making
it
work
right.
Let's
see
where.
Why
is
that
not.
G
A
A
A
H
A
C
A
A
A
A
F
F
A
A
A
It
doesn't
really
matter
because
so,
if
the
only
I
guess,
if
the
only
we
figured
that
there's
this
weird
bug
happening
in
the
CI
with
the
model,
SLR
HTTP
test
for
the
tutorial
that
I
added
recently-
and
it's
like
not
failing
it's
failing,
sometimes
not
feeling
other
times,
it
sounds
like
most
the
time
it's
now
failing,
and
so
basically,
we've
just
added
a
commit
that
that
skips
that
test,
and
so,
if
you're,
seeing
this
fail
and
your
PR
don't
worry
about
it.
Actually,
we
I
mean
it's
always
good
to
merge
master.
A
If
you
merge
master
you'll
pick
up
the
skip
the
test
commit.
If
not,
then
you
know
once
we
merge
your
commit,
if
that's
all
merging
you
commits,
if
that's
the
only
CI
test
failing
and
then
from
there,
and
so
it's
not
incredibly
necessary
to
merge
master.
So
I'm
gonna
actually
review
this
later,
because
we're
running
low
on
time
here
so
I'll
review
that
and
thank
you
for
putting
that
up.
Nice
nice
work.
Let's
see
so,
let's
see
so,
let's
talk
about,
let's
talk
about
so
so
the
the
dal
for
pie
thing
so.
A
We
need
to
make
a
clear
message
about
needed
to
install
from
Conda
I
feel
like
we
had
some
kind
of
discussion
about
this
some
recently
with
a
similar
thing.
Basically,
oh
yeah,
this
is
actually
exactly
what
soutane
she
asked
the
other
day
is
so
we
need
to
add
these
instructions
to
so
need
to
leave
that
up
to
use
at
this
one.
A
It's
all
polls
from
here
right,
so
anything
you
write
in
the
NIP
UI
top-level
in
it
up,
he
Y
goes
right
after
the
install
command,
and
so
it
would
be
helpful
to
say.
Okay,
you
also
need
to
run
blank
and
then
put
the
Conda
install
come
in
there
and
that's
basically
what
we
need
to
do
for
anything
that
has
to
be
installed
from
Conda,
and
so
so.
The
reason
why
I,
we
should
probably
not
go
and
install
things
for
people
is
because
people's
environments
are
always
going
to
be
different.
A
In
the
moment
we
start
trying
to
install
things
for
them.
We
get
into
the
mess
of
we're
trying
to
be
the
installer
in
some
senses,
and
that's
not
really
something
that
we
should
do
so
because
you
know
people
might
have
other
ways
they
want
to
do
things
or
it
just
might
blow
up
in
a
weird
or
a
weirder
way
than
it's
already
exploding
right.
If
they
get
the
error
message
that
they
can't
import.
This
thing
that
they're
gonna,
google,
it
and
the
first
thing
they're
gonna
see
is
probably
Conda
install.
A
A
F
A
I
noticed
some
stuff
insured
is
failing,
so
the
thing
about
should,
I
also
is
that
it
uses
the
it's
doing
is
running
a
few
commands
that
are
network
based
and
rely
on
external
services,
and
so
those
services
do
not
have
Karen.
You
have
time
so
like
safety
and
or
let's
see
ya,
analyze
JavaScript.
This
is
just
so
the
NPM.
This
is
one
that
happens
pretty
commonly
it's
just
it's
just
that.
The
this
end
point
here
is
not
reliable,
so
don't
worry
about
that.
A
G
A
A
A
A
A
A
I
A
A
A
D
D
A
A
Bot
that
you
know
you
maybe
has
like
a
paragraph
of
text
and
you
ask
it
a
question
about
the
paragraph
and
it
responds
so
this
might
be.
You
know
what
you
could
do
here
is,
let's
see,
how
could
we
take
that
and
spin
it?
Why
don't
we
do
something
that
maybe
grabs
we
could
grab
like
the
first
section
of
a
Wikipedia
article
or
something
you
could
say
like
you
know
that
error,
let's
see.
D
A
A
A
With
random
predictions,
does
that
make
sense,
noggin
so
basically
just
create
a
I
think
there
should
be
okay,
great
yeah,
so
just
just
just
make
it
right
and
then
use
mock
and
everything,
because
that's
probably
what
we
should
do,
anyways
in
the
testing
and
then
when
they
actually
go
run
it.
We
don't
do
that
alright
cool!
So
that's
a
good.
Do
you
feel,
like
you,
have
a
path
forward?
There
then
yeah
great.
A
D
A
A
A
A
Well,
basically
anytime,
you
import
your
it's
like
if
you're
looking
at
your
dick
get
directory
and
you
do
get
clean,
I
pretty
much
can
guarantee
you
that
a
bunch
of
UIC
files
will
go
away
so
and
and
sometimes
when
it
implies
those
I'm,
not
sure
what
the
default
is,
but
sometimes
it
decides
to
remove
those
asserts
as
well,
and
so
we
need
to
keep
it
in
this
syntax.
So,
let's
add
these
back
in
so.
I
C
I
A
Thank
you
for
doing
that.
Okay
and
then
you
know
how
to
rebase
back
and
edit
just
that
commit
right,
yeah,
okay
and
then
the
rest
of
this
looks
good,
oh
yeah,
that
new
syntax.
Have
you
guys
seen
this
before
I
said
I
saw
this
recently,
it's
pretty
nice.
You
can
just
put
the
underscore
that
way.
You
can
write
longer
numbers
that
no
I
know
and
know
how
many
zeros
there
are.
A
A
A
Was
gone,
what
they
I
think
the
cache
got
wiped
out
or
something
I
started
in
slotting.
The
new
one
and
I
was
gone.
Why
is
it
feeling
like?
Oh
it's,
the
previous
release?
It's
not
our
current
really
excited
to
go
update
the
previous
release,
tech,
all
right,
okay,
great
and
then,
let's
see
so,
and
then
we
want
the
new
classification
models
pure.
But
you
said
it
depends
on
that.
So,
let's
see,
let's
just
go
check
it
out.
Right
now.
Maybe
we'll
just
reveal
oh,
no
I'm
getting
kind
of
okay
I'm
we're
gonna!
A
G
D
B
A
A
A
B
B
A
B
A
The
STR
method
should
do
this,
okay,
but
so
now,
I'm
not
sure
what
is
this
actually
clear
or
is
the
other
one
clear?
It
might
be
clear
or
the
other
way
yeah.
That
might
be
clear.
The
other
way
this
isn't
actually
clear,
yeah
yeah,
keep
it
keep
it
the
other
way.
What
do
you
guys
think
top
way
or
bottom
way.
C
A
A
A
Great
yeah,
then,
basically,
if
you
do
pretty
and
no
print
you
know
image,
then
it
would
pretty
print
it,
but
now
with
the
image
and
then,
if
you
just
said
no
print
image,
then
you
would,
you
know,
take
out
the
damage
from
the
final
data.
Now
the
thing
with
this
that
you're
gonna
find
is
well
I,
mean
you're,
just
working
on
the
CLI
command
here
right,
so
you
can
mess
with
it.
How
you
want.
B
B
B
A
A
B
G
A
A
G
A
Yeah,
oh
and
I
had
did
I
put
I
might
have
put
up
another
issue,
but
there's
one:
that's
not
it's
not
actually
verifying.
Okay,
and
this
is
actually
tensorflow
or
way.
Tensorflow.
There's
an
error.
Psych
it
okay,
I
see
what
you're
saying
don't
support
it
raises
features
awesome.
Okay,
so
we
need
to
figure
these
things
out.
A
B
A
A
G
A
A
A
A
A
H
A
All
right,
great
and
then
I
think
that
flow
command
is
in
there.
So
Sood,
not
sure
if
you
wanted
to
reference
I
think
well,
actually
I
think
that
was
actually
emerged.
The
other
day
right.
So,
let's
put
in
the
the
PO
request
to
reference
here,
update
existing
instead
of
rewriting
flow.
Oh
wait,
no
where's,
the
one
changed
add
Victor
create
this
is
the
one
where
he
also
add
didn't
you
change
the
flow
command
somewhere.
A
A
Well,
that
is
yeah.
We
should
not
grow
up
or
flow
in
this
project.
That's
not
going
to
be
helpful.
Let's
see
yeah,
okay
and
sorry
tom,
you,
maybe
you
did
you
use
it
in
or
maybe
it
was.
You
were
going
to
use
it
in
the
open
poll
request.
Maybe
we
don't
have
an
example,
yet
we
have
an
example
somewhere
just
in
case
you
wanted
to
go
update
that
and
revalidate
it.
Oh.