►
From YouTube: Offline Web Apps on GitHub Pages - GitHub Universe 2015
Description
Myk Melez lets us in on how Offline Web Apps hosted on GitHub Pages can use Service Workers to cache themselves consistently and persistently, including automatic, silent, and incremental updates, with graceful degradation on browsers that don't support them.
Learn what Service Workers are, find out how to integrate them into Grunt and Gulp toolchains, and see demos of Offline Web Apps on GitHub Pages surviving page reloads and browser restarts on disconnected devices.
About GitHub Universe:
Great software is more than code. GitHub Universe serves as a showcase for how people work together to solve the hard problems of developing software.
For more information on GitHub Universe, check the website:
http://githubuniverse.com
A
Hello,
everybody
thank
you
for
sticking
around
late
in
the
afternoon
and
our
friday.
I
know
it's
been
a
long
week.
My
name
is
McMullin
and
I'm
going
to
talk
to
you
about
offline
web
apps
on
github
pages.
First
things.
First,
this
is
my
octo
cat.
She
helps
me
to
code.
She
pair
programs
with
me
by
sitting
on
my
lap
and
staring
at
me
very
helpful,
but,
more
importantly,
I
work
for
Mozilla.
You
may
have
heard
of
us.
We
make
firefox,
and
my
team
at
Mozilla
has
lately
been
thinking
about.
A
How
do
we
demonstrate
the
promise
of
emerging
web
technologies
to
close
the
gap
between
the
web
platform
and
native
platforms?
Not
only
mobile
but
also
desktop,
and
so
the
the
idea
that
we've
been
kind
of
riffing
on
lately
as
well
offline
web
apps
are
great.
It
would
be
great
to
have
a
really
simple
way
to
publish
them
and
deploy
them.
Github
pages
is
a
great
way
to
publish
and
deploy
websites.
What,
if
we
mix
those
two
things
together
so
get
a
pages.
A
How
many
people
were
here
at
Ben,
Baltar's
presentation
yesterday
this
this
room
same
exact
time?
Okay,
so
been
talked
about
10
ways
that
people
are
using
or
misusing
github
pages,
I'm
I'm
going
to
demonstrate
an
11th
way
to
misuse
them.
Definitely
a
misuse,
and
so
github
pages
are
really
simple
static,
hosting
service
for
static
websites,
and
it's
really
easy
to
publish
to
them.
You
just
use
a
repository.
You
create
a
branch
in
your
repository,
so,
like
I
said
at
Mozilla,
we've
been
thinking
about
how
do
we?
A
A
It
definitely
is
offline
though,
and
it's
got
a
lot
of
juice,
so
I
figured
it
was
a
reasonable
metaphor
for
offline
web
app.
So
what
are
they
offline?
Web
apps
are
very
simply
web
apps
that
run
and
that
work
when
your
network
connection
doesn't
and
that's
not
necessarily
to
say
that
you're
in
airplane
mode
or
you
have
no
network
connection
at
all.
A
Sometimes
you
may
be
theoretically
connected
to
a
Wi-Fi
network,
but
what
you
really
have
is
something
that
my
colleague
Bill
Walker
calls
life
I,
because
your
network
status
indicator
says:
hey,
you've
got
a
network
and
then
you
try
to
load
something,
and
it
just
fails
so
with
offline
web
apps
that
doesn't
happen.
Offline
web
apps
just
work.
Let
me
show
you
an
example
of
a
web
app
that
doesn't
work.
This
is
the
thing
that
people
often
see
today.
You
know
you.
You
try
to
load
some
website
or
some
web
app.
A
This
particular
example
is
from
Android
and
it's
Firefox
on
Android,
but
this
happens
on
all
browsers
on
all
platforms.
You
don't
work
connections,
so
you
don't
have
a
website.
You
can't
load
the
web,
it's
particularly
frustrating
because
in
many
cases
the
browser
has
actually
got
a
lot
of
the
assets
if
you've
ever
loaded
this
site
before
the
browser
has
a
bunch
of
these
assets
in
its
cache
in
some
cash
somewhere.
A
But
if
it
can't
connect
to
the
website
it
doesn't
know,
it
doesn't
know
that
those
assets
are
up
to
date,
and
so
this
is
what
you
get
the
web.
Is
broken
by
contrast,
native
apps
just
work
by
default
if
you've
installed
a
native
app,
that
means
you've
installed
all
of
the
assets
that
are
required
for
that
native
app
to
function
and
most
native
apps
work,
whether
or
not
you
have
a
network
connection.
There
are
a
few
that
you
know
will
obviously
show
you.
A
You
know
more
useful
information
when,
when
you
have
a
network
connection
and
they
can
synchronize
data-
and
there
are
even
a
couple
that
just
don't
work
at
all
and
say:
hey
I'm,
sorry,
but
for
the
most
part
native
apps
just
work,
and
we
want
web
apps
to
work
just
as
well
as
native
apps
do
so
for
the
last
few
years.
The
the
answer
to
this
question:
how
do
you
offline
a
web
app?
How
do
you
make
it?
Just
work
has
been
application.
Cache.
A
An
app
cache
is
a
standard
that
was
designed
a
few
years
back
to
be
very
simple,
very
focused
on
off
lining
a
declarative
API.
You
point
your
web
app
at
a
manifest
the
manifest
lists,
all
of
the
assets
that
your
app
needs,
the
browser
downloads
them
and
then
your
app
works
offline
sounds
great
right.
A
Unfortunately,
after
browser
implementers
implemented,
app
cache
and
web
developer
started
using
it.
We
realized
that
there
were
all
sorts
of
edge
cases
that
we
hadn't
considered
and
problems
with
the
specification
and
foot
guns
like
if
you
happen
to
mistakenly
cash,
your
manifest
for
a
long
time.
You
could
no
longer
update
your
web
app.
You
were
just
stuck
until
your
manifests
expired
and
if
you
set
a
really
long
time
to
live,
then
you're
really
screwed.
A
In
fact,
it's
so
bad
that
Jake
Archibald,
who
works
on
the
google
chrome
team
at
Google,
published
a
really
long
laundry
list
of
everything
that
was
wrong
with
appcache
and
said.
Basically,
you
know
application
cache
is
a
douche
bag,
so
web
platform
developers
essentially
started
over
from
scratch
and
what
we
did
is
we
worked
on
something,
and
this
is
not
just
at
Mozilla.
It's
at
Mozilla,
google
chrome
developers
are
working
on
this
opera
as
well
as
we
implemented
something
called
service
workers
and
service
workers.
A
Is
basically
the
ante
app
cache
everything
that
app
cache
is
or
isn't
service
workers
is
so
app.
Cache
is
a
declarative
interface
service
workers
is
an
imperative
interface.
App
cache
focuses
solely
on
the
offline
encase
and
service
workers
is
a
more
generic
solution.
It
supports
off
lining,
but
not
only
off
lining.
It
supports
a
variety
of
other
use.
Cases
and
I'll
talk
only
a
very
little
bit
about
those
a
little
while
later
in
the
presentation,
but
suffice
to
say
the
two
api's
could
not
be
more
different.
A
So
service
workers
are
a
very
powerful
abstraction.
How
many
people
here
know
what
a
web
worker
is?
Okay,
so
that's
a
good,
maybe
half
of
you
how
many
people
of
you?
Sorry,
how
many
of
you
know
what
a
javascript
is.
Okay,
that's
almost
everybody.
So
the
a
web
worker
is
a
javascript
file
that
runs
in
a
separate
context
from
your
web
page.
A
It
runs
in
its
own
context,
also
in
its
own
thread,
and
it's
useful
for
spinning
up,
basically
a
background
process
to
do
some
kind
of
heavy
heavy
duty
lifting
without
making
your
web
page
unresponsive
and
a
service
worker
is
similar
to
a
web
worker
in
that
it
also
runs
in
a
separate
context
in
a
separate
thread,
but,
unlike
a
web
worker,
it
has
a
very
defined
cycle,
a
life
cycle
for
installation
activation
and
update,
so
that
you
can
your
web
app
can
specify.
This
is
my
service
worker.
A
The
browser
can
launch
your
serviceworker
know
that
it's
downloaded
all
of
the
resources.
It
needs
to
consider
itself
to
be
installed,
activate
your
serviceworker,
and
then
your
service
worker
is
now
serving
all
of
the
files
that
your
app
needs
out
of
a
local
cache.
So
what
a
serviceworker
actually
does
once
it's
insulted,
it
can
actually
intercept
all
Network
requests
that
your
application
makes,
and
it
does
this
trip.
A
Can
you
all
hear
me?
Okay,
great,
it
sounded
like
my
mic
might
have
gone
out
there.
It
does
this
transparently
so
that
your
app
can
progressively
degrade.
If
you
add
a
serviceworker
to
your
app
to
make
it
work
offline,
and
then
somebody
runs
you're,
a
panda
browser
that
doesn't
support
service
workers.
That's
okay!
Your
app
will
just
work
the
way
you
know
it
works
when
it's
online.
A
So
so
the
thing
that
my
team
has
been
working
on
at
Mozilla
is
a
project
called
off
liner
and
yes,
it's
pronounced
offliner
like
trough
or
Gough
Street
here
in
San,
Francisco
or
fish
according
to
English
writer
a
long
time
ago,
and
so
what
offliner
does
is
a
few
different
things.
One
thing
that
it
does
is
it
creates
a
service
worker
for
a
nap,
so
you
tell
it
offline
my
app
and
it
actually
writes
the
service
worker
script.
Obviously
you
can
do
this
yourself.
A
The
point
of
this
tool
is
to
try
to
make
it
super
simple
for
you
to
offline
your
app
and
deploy
it
to
github
pages
so
step,
one
offline,
your
app,
you
point
it
at
your
assets.
It
generates
a
list
of
your
assets
and
writes
a
serviceworker
script
that
will
offline
all
of
them
when
a
user
loads,
your
app.
A
So
that's
not
number
two
deployment,
it's
not
that
hard
to
deploy
get
up
pages,
but
we
want
to
make
it
as
simple
as
possible,
and
so
what
we've
done
with
offliner
is
added
a
command
called
deploy
which
will
check
out
a
copy
of
your
repository
push
the
files,
including
the
serviceworker,
that
that
the
offline
command
created
to
the
gh-pages
branch
commit
the
changes
and
then
push
them
up
to
github.
Github
then
takes
over
and
actually
automatically
deploys
your
website
to
github
pages
and
then
the
last
command
we
created.
A
It
might
be
the
github
flow
or
it
might
not
be,
but
for
a
lot
of
people
it
is,
and
that
means
that
you're
submitting
pull
requests
and
other
people
are
reviewing
them
and
then
they
might
merge
them
into
the
master
branch,
and
it
would
be
really
great
if
every
time
somebody
merged
the
change
into
the
master
branch,
you
could
automatically
deploy
that
change
to
your
github
pages
site
and
so
what
what
the
configuration
tool
an
offliner
does
is.
It
goes
through
this
sequence
of
steps
to
configure
travis
CI
to
auto
deploy
changes
to
your
app.
A
Whenever
you
push
changes
to
the
master
branch
and
for
those
of
you
who
don't
know,
I
guess
a
lot
of
people
do
travis,
CI
anyone
everyone,
it's
a
continuous
integration
tool.
What
that
means
is
that
not
just
for
web
apps,
but
for
any
any
project
you
can
hook
it
up
to
github
so
that
any
time
you
check
in
changes
to
a
branch
Travis
will
build
your
project
it'll
run
tests
and
it
can
also
deploy
to
a
variety
of
hosting
services,
notably
it
can't
deploy
to
github
pages.
A
But
what
it
can
do
is
it
can
run
any
arbitrary
script
after
successfully
building
and
testing
your
changes.
So
we
can
write
a
script
that
makes
Travis
auto,
deploy
to
get
hub
pages,
and
that's
what
we've
done
and
if
you
Google
online,
you
can
find
people
who
have
documented
the
process
of
setting
up
auto
deploy
manually,
and
it's
straightforward.
It's
not
that
complicated,
but
it
is
a
long
process.
You'll
find
that
there's
you
know
10
or
12
steps.
A
So
the
point
here
is
to
to
make
it
as
simple
as
possible
to
make
it
a
single
step
to
essentially
say:
okay
configure
my
repository
configure,
github
and
Travis
so
that
when
I
push
to
master
Travis
will
auto
deploy
after
building
and
testing
directly
to
github
pages
I'm,
not
going
to
demonstrate
that
one
today,
because
it
still
has
a
few
quirks
and
kinks
to
work
out.
But
I
will
demo
a
few
other
things
and
I'm
going
to
juggle
this
microphone
while
I'm
doing
it
so
that'll
be
interesting.
A
So
the
first
thing
I'm
going
to
demonstrate
I
have
I'm
I'm
in
a
directory
containing
a
project
I've
already
configured
this
to
work
with
offliner.
The
text
you'll
see
is
going
to
be
bigger,
I
see
a
guy
saying,
bigger,
bigger
I
can't
make
the
tabs
bigger.
Apple
has
decided
what
size
tabs
I
should
have
so
I
can't
do
that,
but
I
did
make
the
text
bigger.
So
hopefully,
that'll
show
up
for
you
back
there
anyway.
A
So
I've
already
configured
this
to
work
with
offliner
I've
already
installed
offliner,
it's
a
node
module
in
this
case
I'm,
going
to
show
you
how
to
use
it
as
a
command
line
tool.
So
I've
it
globally.
Npm
install
dash
G
offliner,
but
you
can
also
integrate
it
into
a
gulp
or
a
grunt
task.
Runner,
if
you
want,
we've
actually
used
gulp
to
to
build
off
liner
itself
so,
first
to
offline
the
set
of
assets
which
I've
already
built
into
the
dist
directory
of
this
app.
A
And
so
what
it
does?
It
says
it's!
Caching,
that's
not
it's
not
completely
accurate.
It's
not
actually,
caching,
it's
more
that
it's
writing
a
script
that
will
do
the
caching
on
the
client,
after
a
user
loads,
the
app,
but
in
any
case
you
can
see
that
it's
cached
all
of
the
files
in
that
disk
directory.
There
are
options,
if
you
want
to
specify
that
you
only
want
it
to
cash,
a
subset
of
files.
You
can
do
that
as
well.
A
It's
always
a
terrible
idea.
Something
will
go
wrong.
You
should
always
just
take
a
video
of
yourself
doing
it,
but
so
far
that
worked
okay.
But
what
does
this
actually
mean?
This
app
is
just
a
kind
of
a
throwaway
app
that
I
bootstrap
for
the
purpose
of
showing
this
process,
but
over
here
I
have
this
other
app
egg
timer
and
it's
a
very
simple,
very
simple,
app
that
counts
down
from
some
time
you
specify
20.
A
A
And
for
those
of
you
in
the
front
row,
you
can
probably
see
this.
Those
of
you
in
the
back
may
have
to
take
it
on
faith.
This
is
a
power
cord.
This
is
not
an
Ethernet
connection,
that's
DVI!
That
is
also
not
an
Ethernet
connection.
I'm
not
going
to
pull
that
or
you
won't
see
anything
oh
and
then
this
little
thing
right
here
that
just
that's
just
a
USB
dongle
that
powers
my
clicker.
So
that
is
also
not
a
network
connection.
I
promise
now
I
will
press
the
reload
button.
A
Did
you
catch
that
you
probably
didn't
for
a
couple
of
reasons,
one
of
which
is?
This
is
a
very
small
app,
so
it
reloads
pretty
quickly,
even
when
I'm
reloading
it
over
the
network,
it
reloads
pretty
quickly,
but
in
this
case
I'm
not
I'm
reloading
it
from
a
service
worker
that
is
running
on
my
machine.
My
laptop
right
here
serving
files
out
of
its
own
cash
and
it's
not
just
reload.
By
the
way
I
can
close
this
tab
and
I
can
reopen
it.
A
Re
open
tab
reopen
browser.
The
app
comes
back
in
a
way,
you
think
well,
okay,
sure,
but
this
doesn't
happen
on
the
web
today
today.
What
you
get
is
web
page
broken
cannot
load
no
network
connection
most
web
page
most
web
apps
should
work
like
this
most
web
apps
should
just
come
back
just
like
all
native
apps
just
start
and
run
when
you
start
them
by
the
way.
This
also
works
in
chrome
and
Chrome
developer
tools.
A
This
also
might
be
a
little
bit
difficult
for
folks
in
the
back
to
see,
but
the
developer
tools
say
from
service
worker
to
indicate
that
these
files
were
actually
served
out
of
the
service
worker
cash.
So
that's
another
way
in
chrome.
You
can
see
that
they're
really
doing
what
you
think
they're
doing
and
we'll
get
that
in
Firefox,
soon,
I
hope
as
well.
A
By
the
way
you
can
also
work
around
the
service
worker
cash.
If
you
really
don't
want
service
worker
to
serve
your
files
for
a
particular
load,
litter
you're
developing-
and
you
want
to
make
sure
that
the
files
that
you
push
to
the
server
are
really
there
force
refresh
force,
reload
shift
command
or
ctrl
R
will
bypass
the
service
worker
so
that
you
can
see
what's
actually
so
if
I
do
it
right
now,
remember
I
have
no
network,
so
it
can't
connect
to
the
internet.
So
I
get
the
you
know:
Phil
Dino,
I,
guess.
A
A
Sync
and
another
is
we're
also
working
on
building
web
push
on
top
of
service
workers
so
that
web
sites
can
let
users
know
when
something
has
happened,
that
they
think
they
want
to
know
about,
like
on
facebook,
that
somebody's
dog
looks
cute
so
because
service
workers
can
run
in
the
background,
that's
a
very
powerful
privilege,
and
we
want
to
make
sure
that
it
is
as
hard
as
possible
for
a
bad
actor
to
exploit
it
via
a
man
in
the
middle
of
it.
Men
in
the
middle
attack
so
HTTPS
required
it's
not
required.
A
A
A
It's
a
way
for
a
server
to
say
it's
okay
for
web
apps
from
other
domains
to
access
the
services
that
I'm
providing
the
resources
that
I'm
serving
or
the
API
endpoints
that
expose
and
in
the
same
way
that
xmlhttprequest
is
limited
to
two.
The
two
same
origin
requests.
Unless
you,
you
know
jump
through
a
few
hoops
or
two
or
if
a
server
publishes
course
headers
service
workers
has
that
same
kind
of
great
power
and
great
responsibility.
So
there
are
limitations.
A
It's
not
that
you
absolutely
can't
load
a
resource
and
a
service
worker
from
another
domain,
but
you're
limited
in
your
ability
to
do
so.
Github
API,
for
example,
all
publishes
course,
headers,
I,
think
Travis's
API,
all
publishes
course:
headers,
enlightened
web
servers.
What
services
will
do
this,
so
you
can
write
an
offline
web
app,
for
example,
that
is
in
a
github
client
or
a
Travis
client
for
that
matter.
A
It's
not
quite
done
yet
can
I
use
service
workers
tells
you
some
information
that
is
mostly
up
to
date,
except
for
in
the
case
of
Firefox,
which
I'm
going
to
work
to
fix,
because,
as
you
can
see
from
this
screenshot
that
I
took
yesterday
can
I
use
thinks
that
service
workers
are
not
available
or
won't
be
available
in
the
upcoming
42
43
and
44
releases,
and
we're
actually
planning
to
enable
them
in
either
43
or
44.
They
already
work
in
nightly,
builds
in
developer.
A
A
They
already
work
in
chrome
and
opera
Chrome
for
Android
and
by
the
way
Firefox
for
Android
same
as
Firefox
for
desktop
43
or
44,
so
we're
targeting
service
workers
for
both,
for
basically
all
of
the
platforms
that
Firefox
for
desktop
or
android
supports
in
the
43
or
44
or
time
frame.
Actually,
if
you
want
to
find
out
the
status
of
service
workers,
an
even
better
website
is
Jake.
A
Archibald
is
a
service
worker
ready
which
is
kind
of
like,
can
I
use
but
specific
to
service
workers
rather
than
being
kind
of
about
the
web
in
general,
and
he
goes
through
in
a
great
bit
of
detail
the
status
of
service
workers
on
all
of
the
various
browsers
and
is
completely
up
to
date
on
that.
So
that's
what
I
actually
recommend
if
you
want
to
find
out
about
status-
and
the
last
thing
I
just
want
to
say
is
stay
tuned.
A
Obviously,
like
I
said
we're
at
Mozilla
we're
planning
to
ship
service
workers
in
43
or
44.
We're
looking
at
web
push.
Maybe
445
there's
a
lot
of
things
you
can
do
with
service
workers
that
are
really
cool
besides
off
lining
I.
Think
off
lining
itself
is
like
hugely
important,
and
it's
going
to
make
web
apps
work
a
hell
of
a
lot
better,
but
there's
a
lot
more.
You
can
do
with
service
workers,
so
stay
tuned
for
that
and
by
the
way,
this
presentation
is
that
come
off
line
web
app
hosted
on
github
pages
thanks.
A
A
Yeah,
so
the
question
is:
does
this
work
for
mobile,
safari,
chrome,
iOS,
Android,
site,
safari,
I?
Think
the
answer
is
no
Chrome
for
Android.
Yes,
as
well
as
the
the
non
Chrome
browser
for
Android,
it
will
work
on
Firefox
for
Android,
so
I,
don't
I,
don't
have
a
good
sense
of
the
story
for
Safari
and
for
iOS
generally
I,
don't
actually
know
Bill,
my
colleague
and
Brendan.
My
other
colleague
are
both
here
and
they
may
be
able
to.
They
may
happen
to
know
something
but
they're
not
looking
positive
about
that.
A
Yeah,
so
that's
a
great
question,
so
the
question
is:
how
do
service
workers
solve
the
problem
of
stale
data
which
confronted
us
with
appcache,
so
the
service
worker
specification
specifies
that
service
workers
have
have
this
basically
an
update.
They
have
a
life
cycle
and
the
browser
has
to
check
the
status
of
a
service
worker
on
the
server
periodically.
A
In
fact,
I
think
if
you
say
that
a
service
workers
like
time
to
live
is
greater
than
a
day,
then
the
browser
will
just
ignore
that
and
check
it
at
least
once
per
day
anyway,
and
it
does
a
byte
by
byte
comparison,
so
it
ignores
whatever
the
server
says
and
actually
downloads
the
service
worker
itself.
If
it's
any
different
from
the
service
worker
that
it
has,
then
it
goes
through
a
process
whereby
it
installs
the
service
worker.
A
It
lets
the
service
worker
get
to
a
point
where
it
says:
okay,
I'm
ready
to
activate
I've
done
any
say.
Caching,
I
need
to
do
if
the
page
that
is
being
controlled
by
the
service
worker
is
actually
currently
loaded.
Then
the
browser
will
wait
until
the
user
unloads
that
page
for
to
activate
the
new
service
worker
to
deactivate
the
old
service
worker
or
the
new
service
worker
can
say
screw
that.
A
Let
me
take
control
I'm
ready
to
go
so
service
worker
supports
the
kind
of
silent
update
model
where,
like
a
user,
never
knows
that
their
application
is
periodically
being
updated.
It
also
supports
a
more
exit
model
where
the
service
worker
can
say:
hey,
I'm,
ready,
I'm,
ready
to
update
this
app
and
send
a
message
back
to
the
app
the
app
can
then
say
to
the
user:
hey.
You
know,
reload
me
to
get
the
latest
stuff
or
it
can
just
do
it
automatically
I,
don't
recommend
it,
but
that's
possible
to
do
so.
A
A
Can
you
so?
Can
it
cash
the
state
of
the
application,
so
a
service
worker
can
cash
arbitrary
data,
and
so
so,
yes,
I'm,
not
sure
that
it
necessarily
makes
sense
you
know,
could
could
be
more.
It
could
make
more
sense
for
the
for
the
app
to
do
that
itself.
For
example,
the
that
egg
timer
app
uses
local
storage,
because
it
has
very
little
I
mean
it
has
very
little
state
if
it
had
a.
A
But
if
you
had
a
bunch
of
data
you
say
want
to
want
to
use
index
DB
and
I
think
the
service
worker
has
access
to
index
DB,
as
well
as
its
cache
API
that
it
uses
to
cash
static
assets.
So
you
could
certainly
cache
data
in
the
service
worker,
but
whether
or
not
that
makes
more
sense
than
having
the
application
cache.
It
itself
that's
a
good
question
and
then
you
had
a
question.
A
A
Yeah
sure,
so
the
question
is
a
little
while
ago
I
showed
you
well:
I
can
bypass
the
service
worker
cash
by
doing
a
hard
refresh.
I
did
that
in
chrome
and
that
obviously
broke
my
egg
timer
app.
So
what?
If
I
now
do
a
soft
refresh?
What
will
happen
there
and
what
will
happen
is
that
the
app
will
come
right
back,
because
when
I
do
a
soft
refresh
the
browser
says:
okay,
hey,
there's
a
service
worker
active
for
this.
A
A
Yeah,
so
the
question
is:
is
the
specification
flexible
for
handling
multiple
workers
and
making
sure
that
they
don't
step
on
each
other's
toes?
Essentially,
that's
a
great
question.
I
don't
have
a
really
good
answer
for
that.
To
be
honest
with
you,
the
the
specification
defines
the
life
cycle
by
which
a
service
worker,
a
new
service
worker,
can
kind
of
take
over
for
an
old
one.