►
Description
Presented by Mariatta Wijaya, Backend Engineer and Python Core Developer at Zapier
Managing a large open source project like CPython is no easy task. Learn how the Python core team automated their GitHub workflow with bots, making it easier for maintainers and contributors to collaborate together. Even if you’re not managing a large project, you can still build your own bot! Hear some ideas on what you can automate and personalize on GitHub to build at your best. Don’t be a robot, build one.
About GitHub Universe:
GitHub Universe is a two-day conference dedicated to the creativity and curiosity of the largest software community in the world. Sessions cover topics from team culture to open source software across industries and technologies.
For more information on GitHub Universe, check the website:
https://githubuniverse.com
A
Thanks
for
coming
so
my
name
is
Mara
and
I
live
in
Vancouver
Canada,
with
my
husband
and
two
children.
I
work
as
back
an
engineer
at
zapier
and
I'm
involved
in
various
communities
like
pi.
Ladies
I
run,
the
PI
Cascades
conference
and
I
contribute
to
open
source
projects
and
when
you
contribute
too
much
to
open
source
projects
one
day,
they
want
to
give
you
the
commit
right
to
it
and
that's
how
I
became
one
of
the
Python
core
developers.
A
So
if
you're
not
familiar
with
the
term
Python
core
developers,
it
means
I
can
merge,
pull
request
to
Python
as
a
Python
code
available.
My
expertise,
area
of
expertise
is
emoji
and
the
core
workflow.
Now,
if
you're
wondering
what
is
an
expert
in
emoji
like
what
is
this
about,
I
don't
know
I,
they
never
really
asked
me
opinion
about
emojis.
So
let
me
tell
you
about
our
core
workflow.
A
A
Well,
the
other
problem
is
that
our
workflow
contributing
to
cpython
is
actually
super
complicated.
We
have
this
big
checklist.
Is
this
a
bug
ordered
feature?
Have
you
signed
the
CLA
that
did
you
write
tests?
Did
the
test
passes?
Did
you
write
the
Nishio?
Has
somebody
requested
this
fix
or
you
just
you
just
want
it
for
yourself?
A
A
A
So
backporting
is
how
you
will
get
three
eight
one
and
three
seven
six
in
a
few
weeks
time,
so
it's
very
important
process
for
us
now
consider
this
scenario:
you're
using
Python.
You
found
back
in
F
strings,
which
is
the
best
strings
there,
which
was
introduced
in
Python
3
6.
So
if
you
have
a
buck,
what
happens
we
will
target,
as
we
will
need
to
fix
this?
Not
just
in
3
6?
In
fact,
we
don't
fix
3
6
anymore.
We
will
first
fake.
A
We
will
fix
it
in
the
newer
versions,
all
of
Master
3
9
3
&
3
7,
but
we
will
first,
you
will
first
make
pull
request
to
the
3
9
the
latest
branch
first
and
then
we
are
going
to
review
it.
We're
going
to
perform
all
those
checks,
make
sure
everything's
good
and
only
after
we
merge
it.
Then
we
will
cherry-pick
and
back
port
to
the
3
8
&
3
7.
A
Let
me
go
one
step
deeper
and
tell
you
how
this
really
works.
This
is
how
we
used
to
do
it.
We
used
to
write
all
of
this
by
hand
back
porting
cherry
picking.
First,
we
need
to
fetch
checkout
a
branch:
do
a
git
cherry-pick
push
it
to
master
right
push
it
to
Arjun,
delete
the
branch,
and
then
we
close
our
terminal
and
go
to
github
and
create
a
pull
request
for
every
commit
that
needs
to
be
back
ported
for
every
branch.
A
Thankfully,
we
figure
out
a
better
way
to
do
this.
We
created
a
command
line,
script
called
cherry
picker.
So
now
it
looks
more
like
this.
Just
a
one-liner
cherry
pick
past
the
commit
hash
tell
it
which
branches
to
back
port
I
thought
we
will
be
happy.
These
cells
are
a
problem,
but
there
are
some
problems
with
it.
First
I
need
to
be
on
a
computer
on
a
laptop
and
that's
that's
a
big
problem
like
if
I
merge.
A
If
I
can
March
a
pull
request,
while
sitting
on
the
beach
on
my
phone,
why
do
I
have
to
wait
until
I'm
back
on
the
computer
to
do
this
right?
It's
a
big
problem
and
because
I
have
to
wait,
sometimes
I
forgot,
so
things
get
missed
and
then
people
start
using
this
command
line
tool
that
I
wrote.
That
was
not
good.
A
And
it's
just
a
boring
thing
to
do
so.
I
thought!
Well,
there
needs
to
be
a
better
way
to
do
this.
We
know
it
is
possible
to
automate
it.
So,
let's
build
the
bot,
so
I
started
figuring
out
something
learning
how
to
build.
Github
BOTS
by
reading
github
developer
guide
and
the
the
exit
architecture
looks
like
this.
Basically,
you
can
subscribe
to
web
hook.
Events
from
github,
it
will
tell
you
when
something
happens
on
github
say
there
is
a
pull
request
that
would
create
a
pull
request.
A
A
A
A
When,
as
you
recall,
but
the
from
previous
screens,
where
I
show
you
all
of
this
git
commands
so
internally,
it
runs
a
set
of
git
git
fetch
and
everything
that
means
in
order
for
the
bot
to
to
work.
It
first
needs
a
copy
of
the
see
Python
it
is
to
clone
it,
and
the
trouble
is
see.
Python
is
really
really
old
is
like
30
years
old.
A
A
So
I
have
to
be
creative
and
figure
out
how
to
solve
this,
but
this
is
actually
supported
in
Heroku.
You
can
run
background
tests
and
in
Python
you
can
use
a
library
like
salary,
so
your
web
service
can
receive
the
web
hook
event
and
then
just
start
a
background
task
to
perform
all
those
of
long
running
long
running
tasks.
A
So
this
is
how
our
github,
what
works?
It's
not
so
complicated,
and
this
is
our
most
prolific
Peter,
but
we
call
her
Miss
Islington
in
case
you're
wondering
miss.
Islington
is
the
name
of
the
witch
from
monty
python's.
The
holy
grail
sketch
so
Miss
Islington
is
three
years
old
2017
we
build
it
before
their
words,
github
actions
or
anything
completely
open-source.
You
can
check
the
codebase
to
study
it.
It's
written
in
Python,
3,
6
and
the
web
service
of
choice.
A
We
use
a
io
HTTP,
which
is
asynchronous
web
server
and
web
file,
but
the
real
magic
that
makes
building
get
her
boss
easy
it's
really
the
gadget
hub
library,
which
is
an
asynchronous
library
written
by
a
fellow
Python
core
developer
from
Canada
Brad
cannon.
So
if
you
you're
interested
in
build
your
own
github
integration
in
Python,
I
really
recommend
checking
out
Gidget
up
here's
a
little
code,
snippet
of
how
the
code
base
of
Miss
Islington
using
github
and
a
io
HTTP
from
the
top
when
it
says
router
register
for
requests
closed.
A
A
Once
we
figure
out
the
commit
hash.
We
knows
we
try
to
figure
out
where
to
back
for
it
to
and
then
just
kicks
off
the
task
to
backward.
So
it's
straightforward.
Is
this
not
complicated?
There's
nothing
magical
to
it.
So
this
was
one
of
the
most
exciting
moments
in
my
life
when
the
bots,
when
this
is
mrs.,
Lincoln's
first
pull
request
as
successfully
backported
at
a
pull
request.
A
This
was
September
2017
immediately
after
she
made
a
pull
request.
She
got
scolded
by
our
other
BOTS
for
her
not
signed
the
CLA,
so
it
was
fun
to
watch
other
problems.
I
used
to
have
with
cherry
picker,
now
gun
right
like
I'm,
not
waiting
I,
don't
I
can
do
this
from
my
come
up
from
my
phone
I.
Don't
need
to
wait
until
I'm
back
on
the
crown
on
the
computer
and
I.
A
A
It
has
something
to
do
with
our
other
parts
of
the
workflow
waiting
for
tests
to
pass.
We
have
lots
of
checks.
We
do
that
needs
to.
There
are
required
to
pass
waiting
for
all
of
these
to
be
done,
takes
at
least
30
minutes
for
one
pull
request:
I,
don't
want
to
wait
right.
So
what?
If?
What?
If
we
let
the
bots
wait-
and
just
tell
me
it's
all
done
like
this,
and
we
totally
have
this
spoiler
and
we
built
this
kind
of
integration.
A
A
A
This
is
not
an
easy
thing
to
do,
because
merging
requires
a
commit
right.
We
don't
give
out,
commits
it's
easily
to
people
out
of
millions
of
Python
users
in
the
world.
Only
hundred
people
have
to
commit
right
like
are
we
really
going
to
want
to
give
commit
right
to
a
code
but
I
guess?
If,
if
we
write
in
Python
with
F
strings,
maybe
it's
okay,
so
we
build
it.
We
wrote
it
and
it's
it's
really
just
few
lines
of
code
just
make
one
call
to
merge.
A
A
The
core
developer
can
now
focus
on
actually
reviewing
it.
This
is
the
point
in
my
life
when
I
realized
like
how
did
we
live
without
this
I
really
thought
we'll
be
happy.
This
would
be
our
happiest
moment.
I
never
thought
will
ever
give
commit
right,
bye,
but
now
after
we
did
it
just
we
can.
We
can
go
back
we'll
spoiled
and
now
that
I
know
just
how
much
things
can
be.
Automated
I
started
asking
myself
what
other
BOTS
can
be
built
and
in
fact,
I
got
grumpy.
A
A
A
They
had
no
idea
like
what
what
do
they
need
to
automate
so
to
help
you
spire,
you
I'm,
going
to
share
even
more
examples
of
bots
and
automations
that
I've
done
that
I
really
haven't
seen
other
people
doing
it
and
I
hope
it
really
inspire
you
to
look
at
your
own.
Workflow
and
automate
I
have
a,
but
this
is
my
own
personal
bot
that
automatically
formats
code
using
black
formatter
when
black
it.
A
Just
a
mere
thought
of
needing
to
download
something
but
like
I,
know
it's
boring,
and
if
you
have
more
than
one
computer,
you
have
to
install
it
again
and
if
you
maintain
projects,
you
then
have
to
tell
your
contributors:
hey,
install
black
on
your
computer
on
you
and
it's
it's
I,
don't
know
it's
a
lot
of
work,
so
I
build
a
bot,
something
that
I
can
easily
install
on
github.
Just
with
a
few
clicks
on
my
phone
and
runs
black
I,
don't
even
need
to
install
black
myself.
So
this
is
my
pride
and
joy.
A
I
like
it
lab
I
use
it
at
work.
I
think
moving
to
another
repository
is,
is
a
pretty
complicated
process.
As
a
comparison,
it
took
two
years
for
cpython
to
move
from
one
repository
to
another,
from
mercurial
to
github
that
took
two
years
building
a
bot
is
so
much
simpler
and
if
it's
really
just
wait
for
the
PR
open
event
and
automatic
quick,
closing
you're
still
thinking,
no,
this
is
really
too
much
work
to
do
it
really
just
four
lines
of
code
like
this
and
I.
A
Have
that
I
have
the
source
code
you
know
is
apache
license
if
you're
interested
you
can
do
this.
You
can
build
a
button
about
once
a
year.
I
take
a
break
from
volunteering
from
any
open
source
activities,
so
I
call
it
my
art
of
open
source
period,
my
hiatus,
but
because
I've
been
active
in
lots
of
open-source
projects.
Sometimes
people
don't
realize
they
still
try
to
ping
me
on
github,
so
to
set
up
boundaries
and
to
remind
them.
A
I
have
my
own
out
of
open-source
auto-reply,
but
it's
accomplished
using
zapier
on
the
app
here
you
can
have.
The
new
github
mention
trigger
I
still
need
to
write
a
little
bit
of
Python
script,
and
you
can
add
that
into
the
appeal
and
it
creates
a
comment
like
this:
you
can
build,
but
that
grades
your
contributors
two
days
ago,
right
here
and
github
universe,
I
let
a
workshop
in
the
maintainer
track.
In
that
workshop,
we
together
we
build
a
github
bots.
A
That
say
thanks
to
the
main,
take
another
thanks
for
installing
the
github
bot
say
thanks
to
first-time
contributors
who
make
pull
requests
to
your
repository.
We
did
this
in
under
one
hour.
It's
it's
not
her.
So
I
think
you
should
have
something
like
this
and
if
you
were
an
add
up
at
the
workshop
I
published
my
materials
online
say
it
with
bots
github
university,
the
docs
io,
so
you
can
follow
along.
A
So
no
more
excuses.
You
don't
know
how
to
build
a
but
I
have
written
the
tutorial
for
you.
It
uses.
Python
uses
lots
of
a
strings,
nothing
magical
to
it.
You
can
do
it
and
if
you're
still
wondering
of
what
kind
of
things
you
can
automate,
really
take
a
look
at
your
own
workflow
like
if
you
find
yourself,
if
you
have
a
checklist,
you
have
to
remind
yourself
about
a
checklist.
A
That's
the
time
you
want
to
write
the,
but
for
it
don't
don't
do
the
checklist.
Don't
don't
do
that!
I'm
I,
if
I
see
you
doing
manual
things
if
I'm
in
your
project,
I
will
tell
you
let's
build
the
word
for
this.
So
don't
be
a
robot.
Don't
do
the
boring
things
build
a
body
Nora
me!
Thank
you.
So
much
for
listening.