►
From YouTube: Kubernetes SIG Apps 20170828
Description
No description was provided for this meeting.
If this is YOUR meeting, an easy way to fix this is to add a description to your video, wherever mtngs.io found it (probably YouTube).
A
Today,
we've
got
some
stand-ups,
a
helm,
QA
and
some
other
things
to
talk
about,
and
so
I
will
start
off
by
sharing
the
agenda
for
anybody
who
doesn't
have
it.
I
just
dropped
the
agenda
into
chat
if
it
doesn't
come
up
right
away,
scroll
down
to
the
August,
28th
and
you'll
see
what
we've
got
going
on
today.
A
My
name
is
Matt
Farina.
For
those
of
you
who
know
me,
there
is
one
new
thing:
I
am
now
working
at
Samsung
SDS
on
kubernetes
and
you'll,
see
that
I
am
focusing
primarily
on
upstream
things
now
and
so,
where
I
didn't
have
as
much
time
before.
You'll
see
that
I'll
have
a
lot
more
time
and
be
a
little
bit
more
responsive
than
I
have
been
in
the
past.
A
The
other
announcement
I
want
to
make
is
just
a
reminder.
September
1st,
which
is
coming
up
here
later
this
week,
is
the
version
1.8
code
freeze
and
then
the
beta
release
is
slated
for
September
6th,
which
is
early
next
week,
and
so,
if
you're
working
on
getting
things
in
be
aware
of
these
upcoming
milestones
41.8
before
we
get
into
stand-ups
I
also
wanted
to
point
out.
If
anybody
is
on
dealing
with
stateful
sets.
I
just
wanted
to
point
out
that
we
have
some
issues
of
them
being
flaky.
A
In
fact,
for
tests
related
to
cig
apps
stateful
set
tests
are
the
ones
considered
the
most
flaky
at
the
moment
in
the
agenda
today,
there's
links
to
a
number
of
things.
If
anybody
wants
to
try
to
sort
some
of
that
stuff
out,
they
can
get
to
some
more
details
on
it.
There's
a
little
bit
of
work
being
done
right
now,
but
the
one
thing
is
is
for
the
1.9
time
frame.
That's
after
1.8
comes
out.
A
We
would
like
to
take
stateful
sets
and
some
others
and
try
to
drive
them
towards
GA,
so
either
fixing
tests
or
removing
tests
that
shouldn't
be.
There
would
help
us
drive
in
that
direction
of
stability
and
that's
what
I
have
for
the
the
intro
stuff.
Does
anybody
have
any
other
announcements
or
related
stuff
before
we
get
started.
A
B
I
think
so
for
staples
I've
even
said
deployment
of
replica
said
we've
at
this
point
gotten
a
version
of
the
V
1
beta
2
API,
which
is
the
one
that
we're
going
to
try
to
promote
ga
bring
the
Reese's
together
enabled
by
default
and
testing.
There
are
a
few
items
that
we
want
laughing.
You
want
to
try
to
finish
up
on
this
week,
but
increase
it
or
B
1
beta
2
release
and
that's
about
it.
C
Go
ahead
on
the
defeating
speaking
about
cron
jobs,
the
cron
jobs,
API
types
got
merged
and
our
part
of
the
beta
already
and
I'm
just
finishing
my
final
PR,
which
will
enable
the
the
batch
p1
beta
1,
where
contracts
will
then
by
default
in
and
kubernetes
I
had
some
issues
initially
so
I
just
decided
to
split
the
PR
about
it.
It
is
working
pretty
good
from
my
local
environment,
I'm
gonna
open.
If
you
are,
and
in
a
couple
minutes
so
we
should
have
the
condoms
in
beta.
Finally,.
A
A
E
D
D
My
name
is
Chameli
Conrad
I
am
mr.
Jack
computers,
resident
of
the
University
of
booyah
in
Cameron,
and
currently
I'm,
currently
Google
Summer
of
Code
participant
with
a
cloud
native
computing
foundation
where
I
work
on
develop,
interpreter
notebooks
for
equivalences
fights
on
clients,
with
my
mentor,
Sebastian
balls
Ben.
So
the
Cuban
ideas,
Python
client
is
actually
the
official
Python
client
for
kubernetes.
It
can
easily
be
used
to
develop
scripts
python
scripts
to
run
to
manage
kubernetes.
It's
a
what
almost
all
Cuban
Estes
endpoint.
D
It
is
generated
using
the
suave
occlusion
in
soccer
coach
and
so
that
it
could
easily
be
updated
to
meet
up
with
latest
versions
of
kubernetes
and
it
functions
by
helping
by
after
soonduk
Vanessa's
REST
API.
This
project
can
actually
be
found
as
part
of
the
Cuban
Asus,
with
a
bit
of
projects
on
github,
so
Jupiter
is
actually
a
web
application
which
is
used
for
creating
and
creating
sharing
of
life
code,
equations,
visualization
and
images
and
documentation
text.
It's
a
both
languages
such
as
Python,
our
Dropbox.
A
F
A
E
D
D
D
Can
us
in
the
slides?
Yes,
hello,
okay,
so
as
part
of
the
google
Summer
of
Code
I've
been
working
on
developing
Jupiter
notebooks
for
the
cadets
Python
client,
and
this
can
actually
be
used
to
test
the
functioning
of
the
Python
client
interactively
online.
So
here
are
some
of
the
notebooks
we
have.
We
have
an
introductory
notebook
for
cabinets
this
resource
on
it
beneath
the
resource
on
api's.
We
have
a
notebook
for
ports,
deployment,
services,
config,
Maps
secrets,
watch
third
party
resources
and
custom
resource
definitions
so
to
get
started.
D
You
have
to
install
Jupiter
by
running
the
command
pipe
installed.
Jupiter.
We
install
a
Python
client
by
running
pip,
install
key
beneath
this.
You
clean
the
Python
client
project
from
github,
using
the
github,
using
the
link
from
the
kubernetes
of
Gepetto
organization
and
give
up,
and
then
you
change
directly
to
the
notebooks
and
you
run
Jupiter
notebook.
So
we'll
move
on
to
a
demo
of
how
these
notebooks
work.
D
So
when
you
run
Jupiter
notebook
within
the
notebooks
folder,
you
actually
see
an
interface
like
this
with
all
showing
you
all
the
current
existing
notebooks
and
then
you
can
actually
open
a
notebook.
Such
as
the
create
a
notebook,
so
this
is
a
notebook
which
simply
showed
us
how
to
spark
a
port,
and
we
begin
by
you
begin
by
importing
all
the
necessary
kubernetes
modules
and
you
can
run
itself
by
using
the
run.
The
run
so
important
are
both.
D
So
we
run
this
cell
to
import
the
modules
and
then
next
we'll
load
configuration
from
the
default
locations,
and
then
we
run
we
run
the
next.
We
run
the
next
cell
in
order
to
create
in
she's
not
pretty
instances.
Sorry
call
v1
API
class,
and
the
next
cell
enables
us
create
models
for
the
view
for
v1
port
and
we
hoped
and
also
create
a
model
for
the
v1
pot
spec
by
running
the
next
cell
and.
D
D
D
G
D
So
we
now
move
on
to
the
next
cell
to
get
the
list
of
existing
pots,
so
we
now
see
our
visible
busybox
orders
that
are
not
increasing,
and
then
we
can
run
the
next,
the
next
cell
actual
in
order
to
delete
our
port.
So
when
you
run
it,
we
see
it's
run
successfully
and
now,
according
the
latest,
so
this
is
actually
functioning
is
actually
a
function
of
how
the
jupiter
notebooks
work.
D
A
C
D
A
A
D
A
D
A
H
They
can
github
actually
statically
renders
if
it
doesn't
actually
need
to
run
the
note,
there's
a
way
to
just
embed
the
notebook
and
have
it
show
the
contents
of
the
notebook,
so
you
can
use
it
as
part
of
documentation
and
then
link
to
using
it
like
with
the
notebook.
You
know
that
you
have
locally
or
something
okay,.
A
I
I
Swift
is
a
proxy
for
helm,
tailored
service,
so
tiller
is
a
RPC
service
and
yeah.
Should
your
business
services
that
you
cannot
directly
access
that
from
a
web
page,
because
we
pages
talk
ejects,
so
we
have
built
like
a
proxy
server
that
reads
the
data
from
killer
and
then
exposes
that
as
a
HTTP
1.0
service.
So
for
this
we
use
this
G
RPC
Gateway
project,
it's
a
gr,
PC
community
project.
It
implements
Google's
cloud
endpoints
specification,
so
so
this
is
actually
a
standardized
format.
How
you
can
convert
a
PC
service
into
an
HTTP
service?
I
It
should
be
one
services,
so
the
this.gc
get
a
project
implement.
That
is
specification.
So
what
we
did
is
essentially
take
the
profiles,
the
request
and
response
photos
that
was
in
the
tiller,
serviced
and
annotated
them
with
these
annotations
for
the
Google,
API
and
regenerated
this
files.
So
once
we
generated
this
files
using
the
GC
gate,
a
project's
own
photos
a
plug-in.
It
generated
like
a
to
a
file
that
automatically
handles
the
HTTP
requests
and
proxies
to
the
underlying
GRDC
service.
I
I
So
you
first,
you
do
helm
installs,
just
a
standard
way.
You
install
your
home
service
and
then,
if
you
want
to
around
sorry,
you
run
your
tiller
service
and
then,
if
you
want
to
run
sweep
in
the
same
cluster,
you
can
use
this
ENL
file.
You
just
say
tiller
deployment
and,
and
it
automatically
knows,
as
long
as
you
run
in
the
same
namespace
with
the
tiller
service,
it
will
connect
to
it.
I
It
finds
out
using
the
labels
for
the
tiller
service
and
and
then
uses
the
tiller
jpc
service
to
communicate
to
it
with
it.
We
also
have
a
mechanism
to
initially
run
it
from
an
external
cluster
using
AQ
config
file.
So
it
kind
of
follows
the
same
pattern
that
helps
CLI
follows
which
it
will
open
a
tunnel
and
then
use
the
tunnel
to
connect
to
the
service
so
I
it
running
locally.
On
my
mini
cube,
so
I
have
both
then
running.
You
can
see
the
tiller
service
running
and
then
also
the
honey.
I
So
if
I
go
here,
for
example
like
if
you're
on
the
list
API,
you
can
see
that
it's
just
the
JSON
object
dumped
there
for
bad,
and
then
you
know
it
has
a
bunch
of
so
this
is
one
of
the
other
releases
the
same
release,
and
then
this
is
the
version
API.
So
essentially
you
can
use
all
these
API
is
now
directly
from
a
web
page.
So
this
project
also
comes
with
a
like
a
simple
wrap:
our
client
J's
client.
I
So
the
you
don't
really
need
it.
You
can
just
do
your
standard
Ajax
calls.
The
benefit
of
this
is
I,
did
if
you,
so,
if
you
give
it
a
like
a
JSON
object,
man
object.
It
will
see,
realize
that,
like
the
query
parameters
or
the
message
post
request,
it
will
encode
those
because
body
is,
according
to
the
GRDC
HTTP
api
specification.
A
I
A
A
E
E
J
J
Six
point:
one
release:
I,
don't
think
we
have
a
firm
date
that
we're
talking
about
for
that
one
yet
and
then
development
is
underway
for
two
point:
seven
there
have
already
been
several
cool
new
features
merged
in
for
that
and
there's
a
pretty
big
number
of
them
sitting
on
the
PR
queue
that
could
use
a
little
bit
of
triaging
and
and
testing
out.
So
if
you
get
some
spare
cycles
and
want
to
contribute,
that's
a
great
way
to
get
going,
I
think
that's
the
the
news
for
this
week.
E
Cool,
so
just
to
give
some
context
like
there
was
some
questions,
the
last
retro
about
helm.
A
lot
of
them
have
already
been
addressed,
but
I
just
wanted
to
throw
this
home
QA
on
the
schedule
just
to
address
any
other
questions.
We
have
I
just
looked
at
the
form
that
we
sent
around
for
the
home,
Q&A
and
I,
don't
see
any
questions
were
submitted.
So
at
this
time,
if
anybody
has
any
questions,
we
can
take
those
just
anything.
That's
been
on
your
mind.
J
E
J
We
have
been
Tomales,
actually
the
originator
of
the
of
our
hatches,
attempting
to
add
authentication
pass
through
to
to
to
tiller.
So
the
tiller
could
act
on
behalf
of
the
end
user,
but
kubernetes
itself
doesn't
have
an
internal
model
to
support
this
yet
short
of
basically
rewriting
all
of
tiller
as
a
as
an
aggregated
api
service,
which
is
an
option,
but
not
one
for
the
home
2.0
roadmap,
because
it
would
be
a
serious
backward
compatibility
break.
So
we
investigated
several
other
options
and
tamal
did
a
version
of
a
patch
I
did
a
version.
J
J
So
for
the
time
being,
we've
actually
removed
that
feature
from
the
roadmap,
not
from
the
yeah
I
would
say
from
the
road
map
from
the
enumerated
list
of
items
that
we've
committed
to,
but
it's
still
kind
of
at
the
top
of
our
minds
collectively,
as
we
work
with
that
working
group
to
try
and
figure
out
what
we're
going
to
do.
So,
that's
why
it
disappeared
off
of
the
2.7
roadmap,
and
there
was
that
funky
little
note
in
the
release
notes.
That
said,
we
dropped
it
off.
So
that's
the
reasoning
behind
that
one.
J
H
J
E
J
K
A
No,
no,
maybe
let
me
try,
so
it's
was
convened
to
solve
a
specific
problem
that
affects
a
gaps
and
other
SIG's
and
once
they've
solved
the
problem.
The
working
group
will
most
likely
disband
and
the
solution
that
they
come
up
with
will
be
supported
by
whatever
SIG's
are
needed
to
support
that
at
the
end.
In
this
case,
it
affects,
say,
gaps
because
things
like
tiller
and
applications
that
deploy
and
manage
pods
inside
of
kubernetes
have
to
deal.
G
E
J
Fisher's
on
to
all
the
Matt's
simultaneously
unmute
your
mics
and
give
answers,
so
we
are
planning
on
kicking
off
the
helm,
three
roadmap
discussion,
so
in
January.
So
for
those
of
you
not
familiar
with
the
way
the
helm
has
that
helm
has
handled
stability
of
requirements,
we've
followed
cember
very
to
the
very
best
of
our
ability.
J
So
we
have
tried
to
not
introduce
any
major
breaking
changes
since
home
2.0
final
came
out,
and
that
has
served
the
community
really
well,
but
at
some
point
we
want
to
be
able
to
take
advantage
of
the
new
features
that
have
made
their
way
into
kubernetes.
In
the
last
you
know
year
and
a
half
or
so
in
particular.
J
We
want
to
be
able
to
take
a
chance
to
review
of
CRTs
and
aggregated
api's
and
many
of
the
newer
types
and
resources
and
see
if
there
are
major
architectural
things
we
should
change
with
helm.
So
we
are
talking
about
kicking
off
that
discussion,
which
is
the
helm,
3
discussion,
some
at
a
particular
summit,
the
helm
summit
that
we
will
convene
in
currently
we're
talking
about
January
right,
Michelle,
yeah,
so
that'll
be
an
opportunity
for
people
in
the
community
to
get
involved
and
say.
J
J
We've
been,
we've
been
looking
at
doing
it
in
somewhere
along
the
u.s.
west
coast
right.
So
we
don't
have
a
venue
quite
specified
yet
so
that's
the
kind
of
thing
that
as
we
get
closer
to
it,
and
once
we
have
finalized
some
dates
and
places,
we
would
love
to
open
up
and
let
people
who
are
you
know
a
long
time
or
opinionated
helm
users
have
an
opportunity
to
come
out
there
and
hang
out
with
us
and
do
some
design.
A
I
do
I'm,
actually
gonna
relay
somebody
else's
question
that
I've
been
asked
and
I've
been
asked
this
question
a
few
times
and
I've
gotten
the
answer,
but
it's
probably
useful
for
folks
to
hear
the
answer
from
this.
So
helm
uses
four
templates.
Users
go
for
templates
and
I
know
goes,
templates
is
a
little
bit
opinionated.
Some
of
the
things
are
verbose
and
some
other
languages
have
some
nuances
like.
A
If,
if
something
is
missing
from
the
data
object,
it
just
treats
it
as
false
zero
blank
that
kind
of
thing
in
the
appropriate
situation
and
there's
a
bunch
of
templating
systems
out
there
what's
Helm's
status
on
templating
filling
in
these
kinds
of
features
versus
having
multiple
templating
systems.
What's
the
thought
around
templating
in
helm,
going
forward.
J
Well,
since
since
home
to
since,
before
helm,
two
was
released,
there
has
been
a
facility
to
declare
other
template
languages
and
rendering
engines
inside
of
the
helm
core.
Originally
you
had
to
compile
in
your
own,
but
nobody
has
ever
used
it
as
far
as
I
know
recently,
ma
chafe,
who
is
now
one
of
our
core
maintainer
zhh,
had
contributed
the
the
entire
tiller
extension
mechanism
that
we
call
sort
of
colloquially
the
rudders.
J
We
would
only
support
them
as
extensions
much
like
we
support
various
features
through
plugins.
Of
course,
during
the
3.0
development
cycle,
when
we
can
break
our
backward
compatibility
constraints
and
all
kinds
of
stuff
like
that,
we
can
entertain
adding
in
other
ones,
but
we
because
of
the
legacy
of
helm.
We
probably
won't
at
any
point
deprecated
the
goatee
PL
once.
A
Chat
here
so
I'll
read
it
repeating
my
questions
on
the
charts
back:
an
option
to
be
able
to
set
on
charts
DML
chart
requirements
as
nodes
count,
nodes
CPU
our
back
is
needed
or
not
and
etc.
Sorry,
my
mic
is
not
working.
Sorry
I'm,
just
not
sure,
I
understand
this
an
option
to
be
able
to
be
set
on
the
charts,
ie
ml
chart
requirements
as
node
count.
Node
CPU
are
back
yeah.
J
I
think
what
Remus
is
asking
about
is
in
the
past.
We
have
talked
about
making
it
specific
in
one
file
or
another.
What
the
minimum
requirements
are
for
running
a
particular
chart
is
that
what
you're
getting
at
Remus
and
minimum
requirements
might
include
a
number
of
nodes
that
would
be
needed
or
number
of
CPUs
or
our
backs,
or
things
like
that
and
our
our
sort
of
stopgap
since
that
is
not
part
of
the
existing
chart.
Daemul
has
been
to
suggest
that
people
enforce
that
via
the
template
mechanism.
J
I
don't
know
if
anybody
has
done
any
additional
work
on
seeing
how
that
could
be
implemented
in
the
chart.
Yeah
Mille,
but
as
projects
like
monocular
have
risen
to
prominence
monoculars
the
package,
the
website
front-end
for
viewing
and
understanding
packing
helm
charts.
It
might
actually
be
a
useful
idea
to
consider
adding
something
like
that
in
the
future.
J
Even
though
we've
tried
not
to
make
any
operational
constraints
make
their
way
into
the
chart,
amyl
the
the
high
visibility
of
that
might
be
a
nice
thing
to
be
able
to
allow
people
to
say
what
do
I
need
to
have
resource
wise
in
order
to
run
this
chart.
So
again,
that's
the
sort
of
thing
that
PRS
are
welcome,
probably
wouldn't
make
it
into
the
2.7
cycle,
which
is
relatively
buttoned-down,
feature
wise
at
this
point.
But
it's
the
kind
of
thing
that
we
would
consider
in
the
future.
I
think.
A
A
If
you
have
questions
and
as
always
one
of
the
best
ways
to
ask
questions
or
get
more
details,
because
sometimes
people
miss
things
in
slack,
which
is
a
synchronous
file,
an
issue
because
then
they've
got
to
deal
with
it
all
right,
then
I
guess
we
can
continue
on
with
the
kubernetes,
tooling,
stand-ups
and
so
for
the
next
one
that
I
have
listed
here
we
have
composed.
Is
anybody
on
to
talk
about
compose
and
where
that's
at,
let's
compose
with
Kay.
A
A
L
J
L
A
F
Each
has
like
a
focus
point
on
different
topics
so,
like
the
last
release
was
getting
a
UI
in
there
and
it
was
also
a
major
refactoring
of
the
entire
infrastructure
where
we
were
using
websites
before
for
communication,
we
were
finding
it's
a
lot
better
integrations
with
other
things,
I'm
using
G
RPC
so,
and
we
were
also
just
like
it's
a
lot.
We
should
for
us
to
push
things
up
that
way
in
the
bi-directional
communication.
F
What
not
so
er
proceed
was
an
awesome
fit
for
draft,
so
that
was
Virgin
ozox,
so,
along
with
the
plugin
installer,
so
the
o7o,
which
is
what's
on
the
roadmap
right
now
we
have
merged
a
new
draft
user
experience
like
a
UI,
which
makes
a
lot
more
pretty
and
then
we're
gonna
be
looking
at
more
in-depth
logging
for
applications.
We're
looking
at
other
different
things
as
well
on
the
whole
application
development
lifecycle
for
communities.
F
A
Are-
and
we
have
some
time
we
can
have
an
open
discussion.
Does
anybody
have
anything
they
want
to
bring
up
or
talk
about?
What
is
qej,
I
actually
don't
know.
I
must
have
noticed
that
CAD,
it's
the
rename
of
something
else
and
I'm
looking
it
up,
because
I
was
just
wondering
that
in
my
head
and
I
knew
what
it
was.
M
New
to
this
group,
but
I've
been
hanging
on,
but
I
will
get
that
Red
Hat
on
an
open
ship,
so
kids
is
defied
team
actually
based
in
India
upper
deep,
though
who's
worked
about
he's
to
be
called
open,
compose
and
then
it
transitioned
to
a
temporary
name
briefly,
and
then
it
moved
to
catch
one
way
to
think
of
it.
It's
just
like
a
it's
a
DSL
kind
of
idea,
exploration
for
what
kuba
news
application
might
be
so
sort
of
simplifying.
You
know
some
of
the
aspects
of
having
to
define
a
bunch
a.m.
M
A
A
G
Hey
Matt,
one
quick
thing
was
his
damn
peak
I.
Think
one
of
my
our
newer
product
managers
at
Google,
palak,
Dell
ow.
What
was
on
the
call
I'm,
not
sure
she's
here
right
now,
but
I'm
actually
gonna
transition.
Some
a
lot
of
the
cig
p.m.
duties
over
to
her
on
the
next
week
or
two.
So
I'm
just
wanted
to
give
a
heads
up
for
that
name:
palak
that
she'll
probably
be
more
active
in
the
meeting
that
I
am
going
forward.
G
G
Okay,
yeah
so
Pollock
if
you're
on
I,
think
yeah
I,
think
I,
say
CEO
name
on
there
now,
so
I
just
I
just
give
them
one
quick
update
that
you'll
take
over
most
of
the
sig
PM
stuff,
as
well
as
representing
the
team
in
CPF,
so
yeah
feel
free
to
introduce
yourself.
If
you're
able
to
hear
this,
oh
but
yeah.
A
Okay,
so
I
had
a
question
here
because
so
in
here
we've
talked
about
plugins
a
couple
of
times
once
with
a
prejudice
tree.
We
talked
about
it
with
in
terms
of
extensions
for
templates.
What
is
the
kind
of
rest
support
with
the
REST
API
gateway
involving
plugins?
So
if
I
wanted
to
say,
have
the
app
registry
plug-in
and
consume
it
through
a
REST
API?
Is
that
something
that
would
somehow
fit
in
here
or
is
it
just
the
core
tiller
features?
Oh.
I
L
Can
provide
some
background
into
this,
because
the
app
registry
code
actually
like
that
plug-in
works
almost
entirely
external
to
helm
and
completely
external
to
tiller,
and
if
there
was
some
way
to
actually
register
like
custom
api's
with
till
her
home.
That
would
be
like
a
different
way
to
do
that,
where
you
could
to
expose
it
through
the
rest
gateway.
But
currently
everything
happens,
external
in
a
separate
process.
I
L
A
A
F
Right,
you
can
all
see
it
terminal.
Let
me
just
make
this
a
little
bit
bigger
for
everyone.
There
we
go
a
little
big
and
expensive,
so
I
just
wanted
to
show
off
the
latest
versions
from
drafter
when
I
was
talking
about
the
version
o
6o
release
in
the
o7o
release,
I'm
just
kind
of
like
what
the
UI
looks
like
for
what
you're
building
an
application.
Now
so
probably
multiple
of
you
have
seen
the
demo
before
of
what
Draft
is.
Essentially,
it
is
a
server
that
is
running
up
on
a
cluster.
F
It
runs
alongside
with
tiller.
It
communicates
directly
with
tiller,
so
it's
kind
of
like
an
in
cluster
server
that
we
communicate
with
to
with
the
draft
client
and
what
it
does
essential.
Is
it
bootstraps
an
application,
but
there
was
a
couple
of
changes
that
happened
in
the
last
release
and
in
master
recently
that
really
make
this
a
lot
different
than
what
it
was
before.
F
So
before
what
we
were
doing
for
language
detection,
you'll
notice
that
I'm
in
an
examples
Python
application,
so
in
here
there
will
be
an
app
dot
pi
that
will
have
like
a
basic
class
Kappa
requirements
that
just
shows
that
it
needs
to
install
class
very
basic
web
api
in
python
in
version.
Oh
five,
oh
and
previously,
no
black,
one
and
previous
releases
of
draft.
It
was
a
very
dumb,
go
look
inside
the
application
directory
and
go
and
find
a
certain
files
and
it
fits
in
there.
F
Then
it's
considered
a
Python
now
and
then
you
would
do
this
for
each
and
every
single
pack.
In
the
latest
release
of
draft.
There
is
a
smarter
language
detection
that
actually
is
being
used
by
get
up.
If
you
go
on
to
like
github
calm
and
if
you
go
search
up
on
there
on
a
project,
you
notice
this
little
bar
here
that
shows
like
go
is
91.3%,
ensure
five
shell,
five
four
three.
This
is
actually
an
open
source
library
called
linguist
and
github
calm.
F
F
But
now
we
actually
do
is
we
do
a
draft
create
and
you'll
notice
that
it
checks
the
entire
application
directory
recursively
checks
through
it
and
sees
all
the
application
source
code,
and
it
makes
a
certainty
check
that
it
is
eighty
five
point:
nine,
nine
percent
of
shirt
that
is
Python
that
bootstraps
with
a
now.
If
it's
wrong
you're,
more
than
welcome
to
do
a
draft
create
with
a
certain
pack,
so
you
can
do
like
a.
Let
me
remove
the
doctor
file
the
chart
and
the
draft
that
creates
I.
F
Do
a
draft
create
and
I
say
I?
Don't
it's
not
go
over
Python
or
it's
actually
go
or
that
should
attack
these
go.
It
will
create
it
with
go
and
you
can
see
in
the
doctor
file
excuse
and
go
like
so
it.
This
was
just
something:
I
wanted
to
show
up,
not
just
to
what
a
draft
is,
but
also
this
is
a
really
cool
project
for
anyone
else
who
is
doing
language
detection
inside
their
projects.
F
The
source
code
available
is
on
like
yet
general
tso,
linguist
and
then
we've
used
inside
of
giraffe.
So
you
need
an
example
and
do
you
need
any
help
of
a
copy
to
give
any
demonstrations,
but
that's
basically
it
and
then,
if
you
do
a
what
was
the
other
thing,
the
with
the
prettified
UI
that
I
wanted
to
show
off.
F
F
Docker,
doesn't
necessarily
know
kubernetes
too
bad
they're
just
wanting
to
build
an
application
and
push
it
off
on
the
communities
for
their
ministry
or
for
whatever
else.
So
we
made
and
hidden
that
behind
inside
of
UI,
and
what
we're
doing
is
we're
just
showing
off
a
nice
UI
to
say
we're
building
the
docker
image
if
the
status
fails.
F
For
that,
then
you're
more
than
welcome
you're,
given
a
build
ID
at
the
end
of
the
build
that
says
if
they-
and
it
will
tell
you
that
just
do
draft
Logs
with
the
build
ID
to
get
more
in-depth
information
like
younger,
build
fail
your
logs
and
it
will
tell
you
what
happened
in
there,
but
for
the
most
part,
it's
not
really
relevant
to
the
end
user.
So
that's
about
it.
This
takes
about
a
minute,
so
I'll
monitor
you,
but
at
the
end
of
it
it
shows
a
nice
little
anchor
emoji.