►
From YouTube: PWA Studio Community Meeting Oct. 4, 2019
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
Right,
okay,
so
happy
Friday
again
everyone
we're
here
for
yet
another
community
demo.
We
have
some
great
content
from
handy
today.
I,
don't
know
that
we
saw
anything
else
that
was
submitted,
but
obviously
we
have
police
board.
I
know
coming
again
no
coming
up.
We
talked
a
little
bit
about
that
in
terms
of
finalizing
timelines.
There
open
it
up
to
the
rest
of
the
community
on
the
call
for
any
questions
and
then
time
for
their
so
I'm
good,
good,
great,
so
Andy.
B
Here,
with
milestones,
the
rebound
that
I
have
been
working
on,
this
performance
optimization
milestone
specifically
with
the
PW
application,
so
we
went
out
for
a
few
weeks.
I
guess
a
reach
of
research
tasks
and
now
well,
you
can
see
we
created,
you
know
50
tickets
and
now
we're
coming
back
through
and
actually
you
know
tackling
these
issues.
B
So
we
started
I
started
with
the
I'm.
A
quite
a
few
things
to
get
through,
so
I
might
go
a
little
bit
quickly,
but
it's
we
started
with
like
the
low-hanging
fruit,
so
this
should
be
a
low
effort,
kind
of
high
reward
type
stuff.
So
I'll
pause
for
questions
after
each
one,
but
these
first
two
you
should
be
pretty
simple:
the
first
one
and
we're
retrying
apollo
operations
now
sao
paulo.
B
Has
this
concept
of
a
link
and
it's
you
can
kind
of
think
of
it
like
middleware
we're
going
to
insert
a
bunch
of
links
into
the
chain
before
we
like
actually
send
a
graph
QL
request,
and
so
one
of
the
things
they
provide
is
a
retry
link,
and
so
you
know,
PW
A's
are
going
to
be
in
low
bandwidth.
Low
connectivity,
like
that's
kind
of
the
niche,
so
you
know
we
might
not
have
a
network,
and
so
this
this
gives
us
retry
logic
for
making
crafts
QL
call
specifically.
B
So
there's
not
much
to
demo
here
I'll
kind
of
go
through
this,
so
I
I,
basically
just
I
can
make
this
bigger.
I
guess
it
was
like
hey:
let's
go
to
an
endpoint
that
doesn't
actually
exist,
this
just
simulates
like
a
network
error
and
so
before
this
change.
When
we,
when
you
first
initially
load
the
application
we
go
get
like,
we
make
3d
graph,
QL
calls
ones
for
the
store
configuration
ones
for
like
the
navigation
menu
and
I'm
blanking
on
the
third
one.
B
But
anyway
there's
three
requests
get
kicked
off
right
away,
and
so
that
was
before
this
PR
and
if
the
graph
QL
endpoint
was
unreachable
at
that
point
they
would
all
just
fail
and
and
you'd
be
sunk.
So,
but
after
this
PR
you
can
kind
of
see,
there's
now
15
requests
so
that
each
of
each
request
will
get
retried
five
times.
That's
just
the
default
we
went
with
and
so
and
you
can
kind
of
see
this
like
waterfall
of
when
they
get
retried.
B
So
there's
you
know
an
exponential
delay
and
jittering
of
these
things
so
that
when
your
server
does
come
back
online,
it
doesn't
get
slammed
with
a
whole
bunch
of
requests
all
at
once.
So
that's
the
trapping
Apollo
operations.
Are
there
any
questions
on
that
guy?
So
it
kind
of
is
pretty
easy
on
this
one.
B
Okay,
cool
now
questions
moving
on
the
lazy
loading
search,
so
we
had
so
part
of
my
investigation
was
like
the
react,
rendering
the
performance
of
the
application
itself,
and
so
you
don't
really
need
to
know
a
whole
lot
about
these
flame
charts,
but
I
noticed
on
initial
load.
We
have
this
like
initial
flame
chart.
B
That's
like
you
know,
does
a
whole
bunch
of
stuff
the
very
first
time
and
then
right
after
that
there
was
an
update
which
went
and
got
this
search
bar
and
rendered
that
I
was
like
that's
interesting
so
and
then
you
can
actually
see
it
again.
I
don't
know
if
you
can
see
my
mouse
or
anything,
but
I
can
kind
of
zoom
in
on
the
the
search
bar
gets
rendered
every
single
time
the
route
changes,
and
that's
because
it
sits
in
the
header
and
there's
a
very
simple
kind
of
easy
mix
of
mistake.
That
happened
here.
B
So
we
are
conditionally.
This
is
the
code
before
the
PR,
so
we're
conditionally
setting
the
fall
back
on
on
suspense
or
so
a
search
bar
component
is
split
off
into
its
own
webpack
chunk
and
is
lazy
loaded.
It's
not
included
in
the
initial
client
bundle,
but
since
it's
being
used
on
initial
load,
that's
not
really
that
great.
You
do
want
to
include
it
if
you
need
it
right
away,
and
so
the
co
here
was
conditionally
rendering
the
fallback
which
we
don't
actually
want.
B
We
want
to
paste
if
the
search
is
open
or
not
conditionally
rendered
the
entire
thing,
and
so
basically
long
as
shorter.
This
one
is,
the
search
bar
is
still
off
in
its
own
webpack
chunk
and
we
don't
pull
that
in
until
someone
actually
clicks
to
open
the
search.
So
if
they
never
clicked
to
open
the
search,
we
never
download
that
chunk
and
there's
some
savings
there.
So
just
a
quick
fix
on
that
one.
B
So
we
used
to
use
statically
import,
the
create
account
page
and
the
search
route
component
again.
These
were
included
in
the
client
jeaious
bundle,
but
you
can
imagine
like
if
the
client,
if
the
user
never
goes
to
the
create
account
page
or
never
goes
to
the
search
page.
There's
really
no
reason
to
include
this
in
the
initial
bundle,
so
this
splits
these
two
out.
B
This
is
our
render
routes
file
which
may
or
may
not
be
familiar
to
you,
but
basically
here's
where
we
put
in
you
know
the
client
set
reps
search,
HTML
create
account
and
those
are
lazy
loaded
in
so
that
you
know
we
won't
again.
We
won't
download
the
created
camp
page
with
a
search
page
until
the
user
actually
goes
to
those
things.
B
Let's
see,
I
think
that's
all
I
wanted
to
say
about
that
in
the
PR.
You
can
see
that
this
created
cap
page
is
now
in
its
own
chunk
and
where
it
used
to
be
inside
client
touch.
I
guess
so
we
end
up
seeing
savings
on
the
size
of
client
is
used
to
include
the
search
and
create
account
chunks,
101
KB
and
a
793.
B
Ok,
so
that's
it
for
this
one.
Any
questions
on
the
wrap
based
splitting
I
have
the
kind
of
best
practices
here
from
the
react
Docs
the
idea
being
basically
like
users
on
the
way
our
use
to
when
they
go
to
a
new
page
it
taking
some
time
or
whatever.
So
it's
not
a
huge
deal
to
go
and
touch
that
chunking,
because
we're
rear
entering
the
whole
page
anyways.
B
Alright,
then
Jordan's
on
the
call,
pull
requests
or
ma'am
as
he
gets
closer.
It
is
over
launch.
So
this
one
is
the
bug
fix
for
the
heuristic
fragment
matcher.
So
this
is,
if
you've
gone
in
Velma
mode
to
like
a
simple
product.
I
guess
is
the
easiest
way
to
see
this
so
go
to
like
accessories
and
then
a
pair
of
earrings
or
something
you
can
see
in
the
console.
B
B
It's
called
I
should
know
these
things,
but
I
can't
remember
Shane.
Thank
you,
introspection
fragment
majeure,
which,
basically
because
our
back-end
has
these
Union
and
interface
types.
We
need
to
use
this
other
matter
instead,
so
it's
not
too
bad,
but
I
did
want
to
highlight.
Build
pack
now
has
some
utilities
around
making
graph
Kubelik.
Also
at
Build
time
we're
gonna
go
issue
an
introspection
request.
Let's
see
the
graph
QL
schema
and
figure
out
what
are
all
the
types
the.
B
B
So
here's
where
we
go
and
get
it
injected
using
the
defined
plug-in
and
then
later
from
the
app
itself,
here's
Benny
adapter.
We
can
this
this
entire
Union
and
interface.
All
caps
hesitates,
call
it
variable,
but
it
that
will
be
completely
replaced
by
web
pack
with
this
big
long
string.
That
looks
like
put
it
into
mocks.
B
Yeah
I
think
that's
a
I
wanted
to
highlight
yeah
that
build
pack
now
has
some
utilities
around
making
graph
QL
calls,
and
you
can
have
dock
you
graph
QL
files
here.
So
we
we
were
in
lining
these
queries
and
things
as
strings
before
we
can
now
separate
those
out.
It's
just
a
little
bit
cleaner
to
work
inside
build
pack.
Now.
B
B
Basically,
and
these
changes
there's
a
nice
little
message
that
says:
hey,
there's
no
products
in
this
category.
Why
don't
you
try
one
of
these
other
categories
instead?
And
this
list
is
populated
from
the
categories
that
we've
already
retrieved
from
graph
QL
anyway,
it
does
randomize
this
list
and
it
does
not
include
the
category
that
we
know
is
empty.
So
we
won't
give
you
a
link
right
back
to
the
empty
category
that
you're
sitting
in
so
that's
pretty
much
it
yeah
I
was
gonna.
It's
a
story
to
that,
and.
A
B
B
B
View
later
we
are
now
moving
to
the
Apollo
hooks
used.
Query,
use,
lazy,
queries,
mutation,
stuff
like
that,
so
I
guess
the
the
easiest
thing
is
to
kind
of
go
down
this
list,
so
this
PR
is
is
big.
There's
41
files
changed
most
of
that,
so
it
removes
actually
the
Peregrine
used
query
hook
and
use
clear
result
that
kind
of
went
along
with
it.
We've
simplified
a
lot
of
well
Apollo,
wasn't
doing
hooks
exactly
correctly
or
using
context.
Excuse
me
exactly
correctly,
so
we
ended
up
having
two
different
Apollo
providers.
B
We
can
now
get
rid
of
one
of
those
which
is
cool,
so
there's
41
files
changed,
but
it's
a
whole
lot
of
removing
code,
which
is
pretty
awesome,
especially
for
a
minute
maintenance,
so
yeah
so
react.
Apollo
used
to
be
the
thing
that
we
use.
That
gave
us
like
some
higher
order.
Components
like
query
and
I
can't
remember
the
other
ones
that
we
were
using,
but
you've
seen
these
throughout
the
codebase.
B
This
ends
up
being
half
the
size
of
react
Apollo,
so
it's
a
nice
little
bundle
savings
there,
like
I,
said
we
replace
instances
of
this
query
component
with
these
are
very
hooks.
We
removed
Peregrine
to
use
query
that
was
like
our
own
before
they
got
there
together.
That
sounds
bad
but
like
before
they
were
ready.
We
ended
up
rolling
our
own
use,
query
hook,
so
we
just
replaced
that
with
we
removed
the
Peregrine
one
we're
now
using
the
Apollo
one
directly
use
lazy
queries.
B
Interesting
I
should
probably
highlight
that
a
lot
of
times
in
your
functional
component,
if
you're
just
saying
use
query
that
query
will
be
fired
off
right
away,
and
sometimes
you
don't
want
that
you
want.
You
want
a
query
to
fire
off
in
response
to
some
interaction.
They
clicked
a
button
or
they,
you
know,
entered
some
text
in
a
field
or
something.
So
that's
where
you
lazy
query
comes
in
it's.
Basically,
you
can
defer
the
first
call
of
your
of
your
query.
Until
something
else
happens.
B
Do
you
can
see
examples
of
this
in
the
code
and
this
PR?
It's
still
open
right
now,
open
the
feedback
for
sure
number
1827
and
then
yeah.
It
removes
a
couple
other
things
that
I
mentioned
previously,
that
we
needed
a
couple:
different
Apollo
providers
or
so
now,
we've
kind
of
consolidated
that
into
just
one,
which
is
nice.
B
Yeah
and
this
changes
all
of
the
way
we
make
graph
QL
calls
and
render
the
UI,
so
this
looking
at
this
will
get
a
lot
of
tests.
This
will
be.
You
can
also
see
up
here
a
major
version
since
we
removed
some
public
API
from
Peregrine's
so
for
data
I
think
we
mentioned
previously
is
coming
up.
I
think
that's
next
week,
so
this
will
not
be
in
four
out
of
five
will
have
a
lot
of
stuff
in
it,
which
would
be
cool
and
I.
Think
you
know,
should
be
a
pretty
fast
follow
to
for
dunno.
C
Questions
I
mean
I,
I,
guess
I
want
to.
You,
did
a
really
good
job,
I.
Think
of
filling
in
all
the
context,
which
is
normally
what
I
want
to
do
so.
I'm,
not
gonna,
go
on
at
length,
but
I'm
just
gonna
do
my
usual
I'm.
Just
gonna
do
my
usual
like
this
is
really
good,
really
work
and
this
stuff,
like
unlocks
a
lot
of
potential
optimizations
that
were
sort
of
hard
to
trace
in
the
code
before
so.
Yes,
each
and
every
ticket
that
is
labeled
performance
right
now
is
worth
a
second
look.
C
If
you've
looked
at
it
and
been
not
sure
how
to
proceed
once
we
start
using
the
regular
Apollo
provided
hooks
for
use,
query,
etc.
It
is
going
to
be
a
lot
easier
to
introduce
performance
improvements,
and
so
it's
definitely
worth
a
revisit
there.
That's
all
I
got
thank
you
for
doing
all
this
stuff.
It's
a
maker
yeah.
A
D
Yeah
I
got
a
question
if
it's
worth
proceeding
and
so
right
now
when
we
switch
to
either
product
page
or
category
page,
you
have
this
this
loader
and
since
we
use
the
HTML
web
back
living,
then
we
just
make
a
separate
HTML
file
for
the
category
on
the
product
page
for
the
initial
load.
So
we
can
show
skeleton
or
doesn't
it
make
sense.
C
E
E
A
All
right
so
quickly,
I
think
for
other
updates.
We
have
talked
about
and
we
talked
about
before
we
have
the
Peabody
Award.
No,
no,
no,
release
coming
up
we're
targeting
next
week,
sometimes
maybe
7/8
to
release
but,
of
course,
we'll
make
sure
we're
pushing
out
more
communication
as
that's
coming
up
again
and
I
think
we
pointed
to
the
release,
notes
and
what's
contained
within
the
release
previously
so
check
out
the
backlog
and
give
questions
free
to
ask
any
slack.
A
We
have
MLA
you
coming
up
here
in
a
couple
weeks,
a
lot
of
great
content
plan
for
that
from
a
pwace
perspective
with
James
and
Aaron.
We
also
have
a
couple-
or
at
least
one
lab
session
for
sure,
planned
while
we're
walking
through
getting
started
with
peanut,
be
a
studio,
customizer,
creating
components
and
using
Peregrine
and
talents,
so
so
look
forward
to
that.
A
No,
so
you
know,
Jordan
has
been
doing
a
great
job
already,
I
think,
with
basically
fulfilling
the
role
of
the
community
maintainer
without
the
official
in
the
alternative
being
community
maintainer.
So
long
time
come
apart
for
that
and
don't
yeah
there
we
go
I!
Think
that's
it
for
updates
quick
pause
for
anything
for
anyone
around
the
room
on
the
call.