►
From YouTube: YUI Open RoundTable 02/21/2013
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
C
A
A
D
B
Such
a
rebel
yeah,
what
we
could
do
things
in
reverse
order,
since
we're
waiting
for
t
lo
and
stuff.
A
E
Discussions
so
we
get
more
people
in
so
okay.
Let
me
share
my
desktop
Eric.
Do
all
right
foot
screen
share
desktop
start
screen
share
so.
E
Take
away
okay,
yeah,
so
so
I
showed
a
demo
of
it.
Last
week,
though,
I
think
probably
what
I'll
do
is
just
kind
of
take
a
few
minutes
when
I
have
new
stuff
for
the
roundtables
and
kind
of
show
off
the
the
latest
things
that
are
going
on
with
it.
So
so,
let's
see
last
week,
I
showed
at
least
just
a
quick
demonstration
of
what
Yui
bench
can
do
so
again.
E
To
recap,
the
goal
of
it
is
essentially
to
provide
a
way
that
developers
as
well
as
the
CI
system
can
consume
performance,
benchmark
results
and
either
visualize
them
or
storm
and
compare
them
for
later
use
across
different
versions
of
the
library.
So
when
we
write
a
performance
benchmark
test,
we're
going
to
know
is
the
performance
of
that
test,
increasing
or
decreasing
or
whatever
other
things
might
impact
it
throughout
development,
throughout
development
of
different
versions.
E
So,
let's
see
so
I
guess,
I'll
show
real
casting
mi
broadcast
yeah
I'm
broadcasting
from
the
screen,
but
it
just
locks
on
that.
Okay,
hey
good
remote
guys!
Can
you
see
my
screen,
or
do
you
see
my
picture?
Oh
I.
D
E
Okay,
interesting,
so
we
have
a
tool
called
web
inch
when
we
start
it,
it's
going
to
ask
us
for
the
Yui
pap.
This
is
basically
the
path
to
the
Yui
repository
locally
on
my
machine.
So
here
it
is
/
Yui
3.
That
is
my
repository.
So
now
it's
asking
for
the
source.
This
is
the
source
of
the
benchmark
file.
So
if
we
happen
to
examples,
what
do
I
have
here?
Benchmark
j/s,
sweet,
okay.
So
now
this
is
actually
going
to
load
in
a
load.
E
Annette
test
I
already
have
a
window
open
to
Firefox.
E
Now,
let's
try
this
again
so
now
what
it's
doing
is
yet
it
starts
up
a
Yeti
server
which,
if
I
open
up
any
browsers
I
want
to
localhost
3000
as
well
as
in
chrome,
do
the
same
thing
now.
It
will
say
two
browsers
connected
and
it's
going
to
run
that
benchmark
along
across
both
of
them.
So
now
you're
going
to
see
chrome
and
firefox
results
for
the
same
test,
and
you
see
there
is
a
little
bit
of
a
difference
there.
This
is
just
a
generic
test.
I
did
on
a
scrollview
rendering.
E
So
if
one
thing
we
can
do
is
compare
it
to
a
different
version
so
right
now
it's
comparing
it
to
the
head
version
of
my
local
repository,
but
if
I
throw
in
37
oh
yeah
370
flag,
as
well
as
a
let's
go
ahead
and
do
380
it's
going
to
run
that
test
across
370
380
as
well
as
that
too.
So
this
will
take
I,
don't
know
just
a
second
to
run
the
equivalent
of
six
tests.
So
can
you
talk
a
little
bit
about
what's
going
on
behind
the
scenes?
E
Yeah
sure,
so
this
is
integrated
with
Yeti.
So,
basically,
it
is
firing
up
its
own
HTTP
server,
which
then
it
passes
it's
an
expressway
server
and
then
pass
that
server
into
Yeti
and
then
Yeti
does
some
modifications
and
there's
well
to
add
in
some
surrouding
and
so
Yeti
is
basically
handing
out
tasks
for
each
of
these
browsers
to
do
which
its
then
executing
and
then
my
client
code,
that's
running
in
the
browser
is
executed.
E
E
It
will
use
basically
whatever
types
of
benchmarking
tools
we
find
most
appropriate
for
specific
tests
in
this
case,
benchmark
j/s
suites
or
what
we
have
throughout
most
of
the
library
at
this
point,
at
least
that
we
do
are
currently
have
benchmark
tests
for
so,
if
we
take
a
look
at
the
results,
then
yeti
already
redirected
us
yeah,
but
I
bet.
If
I
just
go
straight
to
results,
no
I
think
on
this
version
it
is
viewer,
yeah,
ok,
so
now
Yeti
quit
this.
E
E
This
is
the
state
that
was
in
last
week,
but
I
will
go
and
last
week
I
promised
what
at
least
get
integration
to
be
able
to
check
out
any
refs
in
the
library,
not
just
hard
coded
supported
version
flags
that
are
pulled
from
the
CDN.
So,
let's
see,
if
I
do
get
check
out
master
that
will
pull
up
the
latest
code.
Its
most
of
the
commands
are
pretty
similar,
but
so
up
until
that
point,
it's
the
same.
E
One
of
the
problems
with
doing
a
flag
like
dash
dash
be
300
is
that
well
you're
going
to
have
that's
going
to
be
hard
coded
into
the
configuration
options
to
listen
to
look
without
specific
black.
So
by
adding
in
support
to
be
able
to
run
it
run
these
tests
across
any
ref
in
our
in
the
git
repository.
E
E
E
Basically
what
it's
going
to
do
now
when
I
hit
enter,
is
it's
going
to
look
for
a
seed
in
the
yui
seed
file
in
both
head,
which
it
already
exists,
of
course,
but
also
a
370
director
or
in
a
location
where
it
can
find
the
370
seat,
as
well
as
where
it
can
find
this
ref
this
hash,
but
I
sliver
f
as
a
tag.
Yes
in
get
terms
of
ref
is
just
basically
a
reference
to
some
state
of
the
code.
G
E
Reps
can
be
tags
or
just
commits
so
when
I
hit
enter.
It's
already
have
the
370
I've
been
doing
lots
of
testing
with
that
ref.
So
it
already
has
the
370
and
you'll
see
it's
a
it's
detected,
but
it's
going
to
have
to
go
out,
and
so,
when
I
hit
enter
it
says
it
was
able
to
find
the
head
in
the
370
seeds,
but
it
could
not
find
the
861
a
bcc.
So
it's
going
to
create
a
directory
in
a
temporary
location
for
that
clone.
E
E
Also,
one
of
the
other
things
that
I
did
this
last
week
was
at
in
after
talking
abou
Satya,
getting
kind
of
his
work
flow
on
how
he
does
perform
this
type
of
performance
benchmarking.
He
likes
to
do
lots
of
iterations
of
the
same
test
to
try
to
get
a
normalized
value
result.
So
I
added
in
a
iterations
flag,
the
defaults
to
three
so
I,
don't
know
it.
E
E
E
Let's
see
okay!
Well,
let's
see
if
we
go
to
now:
I
change
the
url.
So
if
we
go
to
3,000,
/
results
Eric's
back
okay,
so
this
this
is
basically
a
result
of
the
test
that
I
rent.
So
this
is
lacking
a
little
bit
of
the
visualization
detail
that
we
had
on
the
last
one
on
understanding
the
which
witch
is
which
the
yeah,
which
dot
represents.
What,
but
then
that's
all
just
part
of
some
of
the
charts
stuff
that
I'm
working
with
trip
a
little
bit
on.
E
So
at
least
the
data
is
all
there,
so
data
guess
so
yeah.
All
of
these
results
are
here
and
yeah,
so
the
when
I
had
it
in
the
iterations
I
would
just
went
a
little
crazy
and
just
to
see
how
many
actually
could
support
so
I
threw
in
a
couple
hundred.
I
believe
I
still
haven't
pulled
up
here.
Yeah.
So
here
is
basically.
E
There's
no
way
is
there
any
way
to
get
it,
try
to
bump
up
the
pot,
but
it's
like
it's
a
fixed
type
that
way
yeah,
okay,
well
anyways
looks
like
we're
going
a
little
bit
smaller.
So
one
of
the
things
that
I've
learned
a
lot
is
that
you,
throughout
just
running
lots
and
lots
of
tests,
is
that
you're
going
to
come
up
with
pretty
wide
variances
in
the
in
the
results
between
both
browsers,
as
well
as
the
environment,
so
like
this
little
batch
right
down
here.
E
This
is
this
is
what
I
got
from
running
in
firefox
versus
all
of
this
stuff
up
here
was
running
in
van
MJS,
and
then
it
got.
Let's
see
there
were
some
areas
in
here
where
it
was
actually
where
performance
was
getting
pretty
bad
because
I
was
watching
also.
Has
it
been
video
running
in
firefox
at
the
same
time,
while
I'm
doing
tests
so
then
it's
basically
going
to
make
the
performance
test
for
in
a
little
bit
slower.
So
many
iterations
did
this
room.
F
F
G
F
E
Totally
agreed
so
I
think
now,
I
have
kind
of
the
the
the
basic
architecture
of
this
tool
down
and
it
kind
of
does
what
I
needed
to
do.
I
can
programmatically
tell
it
to
do
reps
and
it
stores
the
data
to
disk,
as
well
as
pulling
the
recent
results
from
it
so
kind
of
where
I'm
at
right
now
is
working.
E
There's
three
things
kind
of
that
are
on
my
plate,
improving
some
of
the
visualization
stuff
to
make
it
a
little
bit
easier
as
we're
developing
to
see
across
browsers
what
exactly
is
going
on
and
what
kind
of
impact
had
I
think
I'm
done
with
the
demo.
So
we'll
go
back
to
the
video
real
quick,
then.
So,
if
I
turn
off,
my
screen
share
yeah
back
on
video.
So
so
that's
one
is
improving.
E
Some
of
the
visit
visualization
but
I
think
it's
a
little
bit
lower
priority
than
the
other
two
things
which
are
integration
with
the
CI
system
and
that's
the
the
main
goal
for
this
tool
is
to
integrate
with
the
CI
system
and
be
able
to
track
the
the
library
performance
statistics
as
a
whole
as
we're
going
through
the
build
process.
The
automated
build
processes.
E
So
then
the
other
part
is
integration
with
yogi,
so
basically
I
think
it
would
be
easiest
just
if
you're
in,
like
a
yogi
module
or
if
you're,
in
a
module-
and
you
just
type
in
I,
don't
know
instead
of
typing
like
yogi,
build
or
yogi
test
selectively,
we
were
on
yogi
text.
This
is
super
super
simple
way
to
give
us
confidence
that
we
didn't
break
something
so
I
would
like
to
have
something
that
gives
us
that's
really
easy
to
use.
E
It
gives
us
the
same
confidence
that
we're
actually
improving
performance
as
opposed
to
decreasing
itself.
You
did
something
like
a
low
you
benchmark,
then
that
would
go
through
using
phantom
jazz,
execute
these
tests
across
the
last
couple
versions,
or
couple
commits,
as
well
as
the
last
couple
versions,
and
give
you
some
sort
of
a
response,
like
it's
same
performance,
slightly
better
way
slower
just
something
like
that
to
give
you
immediate
feedback
and
then,
depending
on
that
result,
you
can
choose
to
step
in
and
investigate
it
a
little
bit
further.
Yes,
I.
A
I
E
Yeah,
when
we're
looking
at
the
when
we're
looking
at
that,
the
the
last
chart
that
I
heard
that
I
displayed
where
fan
MJS
performance
was
way
way
higher
than
firefox.
So
yeah
I
mean
that's
something
to
take
into
consideration.
I
think
throughout
this
the
better
development
in
use
of
this
tool,
we're
going
to
learn
a
lot
of
the
ins
and
outs
of
how
different
environments
and
browsers
react
and
what
types
of
tests
yeah.
F
E
Perf-
and
this
is
kind
of
an
alternative
to
Jasper-
that's
a
little
bit
more
Yui,
specific
and
I.
Think
adding
in
some
of
the
stuff
that
that
uses
I
don't
know
Jasper
flight
behavior
is
something
that
but.
F
That
doesn't
really
jf
perf
doesn't
really
totally
answer.
I
mean
we
don't
want
to
wait
until,
like
some
number
of
people
go
and
take
a
test
and
then
I
don't
does
it
give
like
the
statistical
analysis
of
all
the
runs
of
the
test,
so
that
you
can
determine
that?
Yes,
you
know,
Firefox
is
statistically
significantly,
you
know,
but
based
on
statistical
significance.
E
So
is
basically
what
you're
getting
at
that
these
results
or
not.
You
can't
compare
these
results
across
machines,
but
the
environment
that
these
results
are
determined
and
needs
to
be
consistent
across
all
result.
Sets
not,
and
you
can't
take
result,
sets
from
one
machine
and
compare
them
to
something
else.
So
yeah.
E
A
E
F
Yep,
no
actually
another
like
I,
saw
I
mean
I
know
it's
often
was
saying
about
his
his
problem,
which
is
that
we're
running
these
benchmarks
and
there's
different
results
and
we're
unsure
if
they're
there,
inside
or
outside
of
this
variance.
You
know
the
plus
or
minus
error
rate
that
there
is
to
even
tell
if
we're,
making
an
improvement
or
making
making
things
worse
and
so
being
able
to
have
some
tool.
F
Some
reporting
tool
that
helped
us
make
the
tests
run
enough
times
that
we
can
be
like
feel
confident
that
that
the
the
results
are
an
improvement
or
not,
or
just
that
the
numbers.
You
know
we
can
draw
some
conclusions
from
and
like
right
now
when
we
load
over
benchmarks
and
hit
go
and
we
look
at
the
numbers
and
then
we
switch
the
seed
file
to
like
38
and
then
hit
go
and
we
look
at
the
numbers.
We
can't
really
tell
if
much
confidence.
J
Sodium,
what's
important,
what's
important
to
do
it
as
I've
admission
in
IRC
I
just
decided
to
drop
in
if
you're
going
to
be
doing.
If
you're
gonna
be
storing
benchmarks,
then
you
need
to
store
the
hardware
profile
that
the
benchmarks
were
taken
on
over
time.
So
you
can
so
you
make
sure
that
you're
going
to
continue
to
test
against
the
same
hardware
profiles.
J
To
show
improvement
over
time
and
you'll
want
to
add
new
hardware
profiles.
As
you
know,
what
is
quote
unquote
common
in
any
given
in
a
given
year
is
the
years
progress.
You
know.
Hardware
is
going
to
be
improving,
but
there's
still
going
to
be
some
sense
of
common
hardware,
configurations
or
reasonable
hardware
configurations
that
would
that
should
be
tested
against
to
show
relative
increases
or
decreases
in
performance
yeah.
E
So
one
of
the
when
we
first
started
with
this
task,
it
was
created
something
that
will
get
a
snapshot
of
the
library
at
this
in
its
current
state
right
now
and
then
next
time
we
do
a
build,
we're
going
to
compare
the
new
results
to
the
old
results
and
then
we're
just
gonna
keep
doing
that
process.
But
the
problem
is
tests
can
change
as
well
as
the
current
state
of
the
machine
in
an
environment?
E
J
It
is,
though,
because
you
have
as
long
as
you
have
the
hardware
profile
that
you
can
test.
So
if
you
have
you've
made
some
changes
to
the
code
yeah
and
now
today
you
have
all
of
these
prior
commitments
throughout
history
that
you
can
run
against
that
same
hardware
profile,
to
show
that
there
was
a
performance
improvement
but
you're.
You
have
to
run
all
of
those
historical,
historical
versions
of
the
library
as
well
on
that
same
Hardware
profile
at
this
moment.
In
order
to
confirm
that
right,
there
has
been
a
change.
It's.
K
J
It
is
definitely
a
as
good
as
you
can
get
sort
of
situation
because,
obviously
out
in
the
wild
they're,
there
are
going
to
be
a
lot
of
other
various
variances
into
the
environment
based
on
plugins
and
other
things
that
are
running
in
the
system.
Maybe
somebody
likes
to
keep
500
tabs
open
and
sure
it's
going
to
perform
slow
were
in
that
case
right.
So
so,
but
the
important
point
is
to
run
the
historic
versions
of
the
library
in
the
same
environment.
Yeah.
Yes,.
E
J
E
Why
shorter
I
guess
during
the
development
of
this
so
about
a
week
ago,
is
when
I
added
in
support
for
throwing
in
any
get
breath
that's
on
there.
So
then,
basically,
we
we
don't
have
to
store
these
historical
result
sets
because
we'll
be
I,
guess
ideally
we'll
be
able
to
create
new
result,
sets
on
against
old
versions
of
the
library
pretty
quickly
and
so,
and
that
will
eliminate
the
variances
that
will
get
because
of
differences
in
the
environment
and
the
current
state
of
the
machine
at
that
time.
If.
J
We
can
just
that's
part
of
the
solution
right,
that's
part
of
the
story
there.
The
other
part
of
the
story
is
that
it's
nice,
that
you're
doing
historical
versions
on
the
same
hardware
that
you're
using
right
now
but
you're,
going
to
take
those
findings
and
you're
going
to
post
this
new
new
release
of
the
library
and
say,
look
at
how
the
performance
is
improved
and
we
have
these
these
nice
graphs
that
have
these
numbers
associated
with
these
lines.
Well,
the
next
release.
J
If
you're
testing
on
different
hardware,
then
those
numbers
are
going
to
be
different
for
the
prior
version
that
you
release.
So
it
is
important
to
maintain
some
sort
of
hardware
profiler,
like
keeping
images
of
the
testing
environments,
that
you're
going
to
be
testing
against
and
keep
those
in
perpetuity
so
that
you
have
some
sort
of
consistency
between
version
to
version
in
the
story
that
you're
telling
now
a
common
machine
operates
like
this,
and
this
is
the
trend
on
an
on
a
modern
machine
for
older
machines.
J
E
A
F
E
So
yeah
there's
still
a
lot
of
stuff.
We
kind
of
need
to
figure
out
and
wrap
our
heads
around
just
all
the
discussions.
I
have
a
SAT
TN
quite
frequently
about
because
he
does
a
lot
of
this
performance,
benchmarking
and
figuring
out
ways
that
to
fix
the
pain
points
in
his
work
flow.
When
he's
doing
all
of
this
testing
and
basically
build
a
tool
around
that,
but.
F
F
K
I
think
another
pain
point
is
actually
beyond
beyond
the
reporting
and
the
of
the
beyond
the
results
is
actually
drilling
down
and
finding
what
the
cause
was.
You
know
you've
got
like
one
set
of
evel's
from
fire
fire
bug,
that's
incomplete,
another
set
of
tools
from
chrome,
that's
incomplete
and
maybe
a
way
of
somehow
converging
to
where
you
can
have
a
better
way
of
drilling
down.
In
figuring
out
where
your
performance
bottlenecks
are.
E
So
real
quickly,
I
noticed
on
RC
Luke.
You
mentioned
the
three
to
five
iterations
seemed
low
and
I
agree
that
that
is
low.
This
is
based
the
the
final
number
that
you
get
is
actually
based
off
like
Milo,
whatever
thousands
of
iterations
that
benchmark
jazz
execute
within
a
matter
of
a
couple
seconds.
So
basically,
what
I've
done
is
take
that
benchmark.
Js
result
that
does
eliminate
the
liars,
but
then
also
execute
that
essentially
to
having
benchmark
jas.
E
J
F
E
Noticed
it
I
do
see
it
get
faster.
Yeah
I
have
noticed
that
as
well,
and
so
the
reason,
why
is
it
why
it's
at
three
to
five
right
now
or
you
can
also
specify
it,
but
at
least
defaults
to
three-
is
that
using
Yeti?
It
has
to
execute
it
all
in
parallel
or
sequentially
so
nice
to
execute
it,
while
it
can
do
multiple
versions
of
the
library
at
the
earth.
Sorry,
let
me
think
here
it
can
do
multiple
browsers,
so
it
can
do
Firefox
and
Chrome.
E
E
That's
something
that
reads
working
on
with
with
Yeti
is
to
be
able
to
allow
I,
don't
know,
however
many
Firefox
windows
you
want
to
open
to
execute
all
of
them
at
the
same
time,
which
will
then
help
it
out,
but
we
also
have
to
keep
in
mind
that
you
can
only
have
so
many
Firefox
windows
executing
these
high
or
these
intense
benchmark
jas
tests,
which
will
have
an
effect
on
performance
too.
So
yeah
well
cool
anyways.
So
what's
your
next
steps
for
this?
E
So
next
steps,
yeah
I
think
the
the
three
things
had
talked
about,
which
was
improved,
the
visualizations
integrate
with
CI
and
the
third
one
will
is
I,
don't
know
I'll!
Think
of
it.
When
I
you.
A
E
E
E
An
idea
that
popped
up
I
don't
know
how
reasonable
that
is,
because
there
is
some
human
interaction
that
does
need
to
look
at
the
exact
numbers.
But
I
don't
know
I
mean
basically
seeing
if
that's
seeing
of
something
like
that
once
you.
A
E
Great
little
did
right
now
from
Jasper
I,
don't
know,
I
mean
yeah,
yeah
I,
don't
know
I
guess
this
is
kind
of
a
if
Jasper
really
is
just
a
wrapper
around
benchmark
j/s
that
does
I,
don't
know
some
some
things
on
top
of
it
all
the
the
multiple
browser
stuff,
but
them
yeah.
It
was
cool.
G
Sure,
oh,
this
is
like
a
very
small
devil:
I'll
screenshot!
Yes,.
A
G
Sorry
so
we
talked
about
this
a
little
of
like
a
two
weeks
back
I
think
where
Jeff,
Eric
and
I
have
been
working
on
Yui,
CSS
stuff,
so
thing
yesterday,
I
just
went
in
and
cleaned
up
a
lot
of
the
documentation
and
kind
of
linked
everything
together.
They
were
just
sitting
in
a
lot
of
different
modules,
letting
a
lot
of
different
get
up
repos.
G
So
now,
if
you
go
to
just
go
to
any
one
of
these,
so
if
you
go
to
t
levitra
com
t,
let
me
turn
on
github.com
/
CSS
forms
it'll
bring
you
to
this
page,
and
this
is
like
the
docs
performs
and
then
we're
dogfooding,
our
own
CSS
stuff
that
we've
been
doing
so
form
like
the
menu
at
the
top,
is
using
our
CSS
menu
stuff
and
we're
using
like
CSS
extras
for
like
free,
encode,
styling
and
CSS
button
for
the
button,
styling
and
grids
for
like
responsive
grids
here.
G
So
we
have
forms
and
then
tables.
Now
we
have
lists
which
works
well
with
ryan's
menu
stuff
that
he's
been
working
on.
Jeff
also
made
paginator
'he's
recently,
which
is
down
here,
which
looks
nice.
We
have
CSS
extra
stuff
just
the
same
as
before,
but
the
code
has
been
cleaned
up,
so
I've
been
going
through
it
and
trying
to
one
of
the
things
I
still
have
to
do
is
kind
of
go
through
it
and
make
sure
that
our
coding
conventions
for
the
CSS
class
things
we're
using
are
similar.
G
We
talked
to
this
a
bit
before
just
like
using
Yui
alert
subtle
instead
of
Yui
alert,
Yui
alert,
subtle
things
like
that.
So
there's
still
some
work.
That
needs
to
get
done,
but
I
just
wanted
to
show
that
it's
really
more
discoverable
now
and
I've
added
the
grid
builder
documentation
here
and
actually
improved
this
documentation
quite
a
bit
so
check
it
out
so
so
yeah.
G
A
G
Right
right,
so
that's
been
fixed
now
the
grid
level
up
so
yeah
for
the
grids
just
use,
Yui
3
and
the
grid
up
some
of
these
exist
on
the
CDN.
Some
of
them,
don't
so
I
think
grids,
forms
and
tables
are
all
on
the
CDN
lists
and
X
rupes
list
and
extras
aren't
yet
so
those
will
be
on
the
sphere
and
shortly
for
now,
if
you
can
get
them
from
github
and
if
you're
just
demoing,
you
can
use
rockethub.
G
This
is,
this
is
still
pretty
temporary
in
my
book.
This
is
just
mostly
for
ease
of
use
and
discoverability
of.
In
the
end,
this
will
exist
somewhere
on
YY
library,
but
yeah.
It
probably
won't
look
like
we're.
Library
does,
but
it'll
live
somewhere
there,
but
for
now,
as
long
as
it's
on
my
repo
well,
it's
it's
over
here
very.
G
Up
and
use
cool,
yeah
I
think
it'll
get
easier
once
they
all
cuz
right
now.
You
just
have
to
get
them
individually
and
I
also
wanted
to
say
that,
like
all
the
stuff
is
responsive.
So,
for
example,
for
list
stuff,
you
have
to
get
the
CSS
files
individually
because
there
I
haven't
built
them,
but
once
they're
on
CD
and
you
can
just
grab
a
single
file
and
you'll
be
good
to
go.
You're.
H
G
A
A
All
right,
moving
right
along
I
know
that
Tony
has
a
time
crunch.
So
I
wanted
to
talk
about
the.
Why
I
web
fun
topic?
Maybe
you
can
you
split
background
Tony
about
that?
And
what
do
you
want
to
talk
with
ya.
C
The
one
everyone's
working
on
paginator
you
I
and
decided
that
I
probably
didn't
want
to
use
just
generic
icons
to
go
in
there
for
as
bitmaps,
simply
because
the
way
the
scanner
is
working
and
being
able
to
swap
those
images
out
to
have
multiple
skins.
Remotes
for
colors
for
skins
requires
a
little
bit
more
work
than
just
I
mean
right.
Now
we
have
Sam
and
night
skin
and
those
we
require
a
completely
different
set
of
sprites
for
each
one
and
based
on
the
base
colors
and
the
colors
used
in
the
skin.
C
C
Only
uniform,
Manor
web
fonts
seem
to
be
a
pretty
common
in
a
lot
of
places
for
monochromatic
icons,
for
you
is,
and
I
thought,
starting.
A
small
library
of
icons
or
glyphs
to
be
used
with
these
skins
would
be
a
good
way
to
go,
not
obviously
not
taking
care
of
all
the
use
cases,
but
taking
care
of
a
lot
of
the
use
cases
and,
for
instance,
of
Kent
you
can
do
things
that
have
need
repeating
patterns.
C
So
that
was
where
my
thoughts
started
originally
and
by
using
something
like
font,
awesome
or
some
other
pre-existing
library
would
be
fine
as
long
as
the
licensing
agreement
works
out.
Well,
the
only
issue
is
that
if
we
want
something
change
or
we
want
an
extra
font
to
be
added
in
there,
we
have
to
wait
on
their
development,
their
approval
and
development
cycle
to
get
that
new
icon
in
and
right
now,
I,
don't
know,
I
can't
think
of
anything
that
they
don't
have
that
we
would
like
to
have
in
there.
C
But
you
know
just
as
it
as
an
idea
and
I
walked
into
it.
The
fonts
don't
seem
to
be
terribly
difficult
to
create,
based
on
the
software
that
you
get
not
been
trying
out
a
couple
of
different
software,
or
you
know,
font
creation
utilities
and
like
creating
the
vector,
artwork
and
illustrator
and
trying
to
copy
that
a
copy
and
paste
that
over
so
it's
kind
of
agnostic
that
would
be
ideal
had
merely
been
working
in
most
cases.
C
But
and
that's
that's
where
I'm
at
right
now,
that's
kind
of
the
background
I've
got
a
a
forum
post
or
a
contributor,
the
Yui
contributor
newsletter
rehost
up.
So
that's
that's
been
added
see.
It's
got
a
lot
of
a
lot
of
traction,
a
lot
of
opinions,
lots
of
pros
and
cons,
but
that's
worth
it
right
now
and
I'd
be
happy
for
anyone
to
have
any
more
thoughts
to
provide
them.
You.
A
A
H
H
It
would
be
a
lot
easier,
it's
really
nice
to
be
able
to
change
the
color,
because
you
know
the
background
can
make
all
the
difference,
and
you
know
now
color
and
you
can't
just
have
white
or
black,
because
it
doesn't
work
against
everything
and
Tony
isn't
true,
then,
that
these
fonts
would
also
allow
you
to
set
text
shadow
on
them
as
well.
The
CSS
rules,
yeah
yeah,.
C
A
It
another
thing:
I
see
a
competing
site,
they
have,
for
instance,
they
have
like
github,
they
have
shared
icons,
you
know
it
for
facebook,
twitter
and
you
know
they
have
the
icon,
the
the
fonts
for
those,
and
they
also
have
an
interesting
idea
where,
if
you
add
a
class
icon
spin
to
this
class,
you
make
you
can
use
spinners
for
like
loading
loading
or
indeterminate
progress
as
well.
So
you
can
make
them
do
that
which
is
fairly
nice
I.
E
C
It
would
just
be
more
so
what
we're
using
currently
and
I
looked
at
the
the
Yui
too
bright,
and
most
of
it
was
like
a
bell,
a
no
sign,
the
?
and
stuff
like
that,
and
they
were
multiple
colors,
but
they
could
easily
be
made
into
a
monochrome
icon
without
changing
the
effectiveness
of
the
icon.
And
then,
if
somebody
did
not
want
to
use
the
font
for
the
the
font
icon,
that's
installed
with
the
skin,
they
can
always
use
their
own
sprite,
just
as
they
you
know,
just
as
they
would
now
for.
A
H
That
I
tried
to
get
just
a
left,
pointing
arrow
on
a
right,
pointing
arrow
out
of
that,
a
ski
stuff,
the
extended
ASCII
for
the
paginator
and
for
some
reason,
they've
got
they
don't
really
nap
ones
that
are
the
same,
and
the
only
difference
is
that
they're
left
and
right.
They
seem
to
all
be
slightly
different,
slightly
different
size
and
summer
outline
is
amor
not
so
they
don't
really
have
them
in
logical
pairs
of
left
and
right,
and
when
you
could
find
one
that
was
right,
it
was
correct,
left
and
right
on
firefox.
E
H
C
C
H
H
F
G
F
C
C
I
would
be
interested
to
see
what
a
win
a
screen
reader
tries
to
read
a
not
an
unknown
character
map
as
because
it
in
instead
of
using
the
same
unicode
as
the
care
is
the
a
character.
We
would
use
a
unicode
character,
that's
unmapped
or
is
being
used
as
something
else,
some
other
symbol
and
so
I'd
be
curious
to
know
how
a
screen
reader
would
read
that,
and
so,
what's.
F
C
F
C
F
C
I,
haven't
haven't
done
much
research
into
the
implementation
or
the
performance
aspect
of
it
other
than
just
you
know,
making
sure
that
it
was
something
that
was
possible
and
then,
if
it
was
something
that
the
majority
of
the
group
wanted
or
could
see
a
benefit
out
of,
even
if
it
was
just
a
handful
of
icons
that
were
as
font,
then
go
at
it
from
there.
Yeah.
F
How
would
this
compare
if
we
did
a
sprite
for
like
the
thing
Jeff
was
saying
with
the
up/down
arrows
if
those
are
more
of
a
masking
like
a
light
like
a
a
semi-transparent
white
arrow,
pointing
up
for
the
Sam
skin
and
I
or
for
the
night
skin
and
a
semi-transparent
dark
arrow
for
for
the
night
or
the
Sam
skin,
whatever,
whatever
order
those
would
make
sense,
but
yeah,
so
that
they
essentially
color
the
background,
they
can
use
the
background
color
and
it
is
shade
it
or
lighten
it.
Yeah.
A
F
So
I
mean
I
think
like
if
there's,
if
there's
enough
of
these
complexities
and
or
like
performance
issues
like
oh
now
of
a
sudden,
you
get
for
extra
HTTP
requests
like
you
get
the
CSS
file
plus,
then
for
other
additional
requests.
No
matter
what
browser
your
own
like,
what
what
the
impact
of
that
is
compared
to
the
sprites
right.
C
A
If
anything
from
this
exercise,
if
you
come
the
point
where
you're
like
hey
just
use
fun
awesome
at
least
you've
gotten
you've
come
around
to
that
right
right.
So
that
might
be
the
end
goal.
Is
you
you've
done
all
this
research
and
said:
hey
they've
they've
got
to
this
point.
You
want
to
implement
this
go.
Do
this,
you
know
right
on
awesome,
rotor,.
A
C
Next
step,
it
at
least
for
paginator,
is
I'm
going
to
just
want
to
create
a
stew
for
icons
just
to
use
those
for
right
now,
I
mean
it's
super
easy
to
swap
swap
out
to
a
sprite.
If
we,
if
we
decide
to
go
that
route
or
worry
about
licensing,
if
we
want
to
use
fun,
awesome
and
then
that
way,
I
can
work
on
documentation
and
test
code
and
then
get
paginator
inside
of
data
table
and
give
that
released
as
soon
as
possible.
And
that's
that's
the
route
that
I'm
working
on
so
I'm
out.
A
A
K
A
E
E
E
A
A
E
E
I
F
A
F
A
The
example
just
today
there's
this
gossip
or
request-
that's
been
said
there
for
a
month
that
was
on
some
code
that
is
going
to
be
deprecated.
We
had
to
basically
tell
the
full
requester
sorry,
you
know
we
can't
let
spending
more
time
with
this,
this
coats
going
to
go
away
and
they
spend
all
that
time
to
bring
the
pleura
quest.
We
had
interstate
I
mean
I.
K
Would
I
would
go
one
further,
though,
although
I
think
someone
disagreed
lot
when
we
talked
about,
I
meant
probably
maybe
jenny.
I
don't
want
to
make
her
out
to
be
the
bad
guy,
that's
not
the
case,
but
if
it,
but
if
it's
something
that
we
safer,
we
got
where
we're
thinking
about
popping
out
deprecating.
There
should
be
social
level
of
warning
there
to
a
high
risk
or
something
that's
deputy
of
falling
out
of
the
library
that
should
be.
In
my
mind,
it
should
be
tagged
a
little
differently
because
I
wouldn't
worry
Mike.
E
On
that
either
so
should
to
that
point
it
seems
like,
instead
of
unassigned
on
the
sign
user,
which
we
currently
have,
which
these
bugs
are
assigned
to,
that
doesn't
really
convey
anything
aside
from
just
fits
on
the
side.
If
we
had
like
two
different
users,
which
would
be
up
for
grab
like
the
user
name,
would
be
for
grabs
meeting,
anybody
could
grab
that,
but
then
also
another
one
that
the
user
name
would
be
deprecated,
and
then
we
reassign
all
of
these
bugs
to
deprecated.
Why.
F
E
K
K
F
Not-
and
I
think
the
good
thing
is
like
the
the
bugs
will,
when
we
close
them
out,
they
still
well.
We
don't
have
good
search
for
bugs
on
the
website,
but
at
least
github
has
a
better
search
that
they
still
maintain
a
history
in
the
sense
that
they
still
exist,
even
though
they're
closed,
so
I
feel
like
if
somebody
else
ran
into
that
problem.
F
F
A
In
track
too
great
yeah
definite
process
for
deprecation,
there's
like
blog
posts
and
then
and
we're
supposed
to
convey
that
so
that's
supposed
to
be
what
how
people
discover
that
part
I,
don't
think
we
have
to
go
into
every
bug
and
go.
You
know
when
it
does
actually
get
deprecated,
though
we
do
need
to
like
mention
in
the
bug.
That's
why
it's
been
closed,
yeah,
yeah,
I
think
that's
important,
so
for
these
right
now
that
are
unassigned,
should
we
just
kill
them
or
I?
Would.
A
A
F
Fix
me,
I
should,
I
should
add
stuff
to
these.
You
like
to
this
list
keeping
do
right.
E
F
A
B
F
F
F
A
F
Teyla
would
know
teyla
he's
probably
getting
a
sandwich
right
now.
I
think
I
think
the
plan
is
to
deprecate
it
from
what
I
remember.
Yeah.
F
A
A
H
G
Saying
I
was
eating
cake
yeah,
the
the
plan
is
deprecated,
but
I
haven't
actually
like
switch
the
module
name
to
css-based
deprecated.
Well,
you
don't
have
to
do.
F
G
A
A
E
E
H
F
G
F
Released
with
it
had
its
initial
release,
which
is
when
Yui
30,
the
first
version
came
out
and
then
I
had
one
bug
fix
in
341,
but
every
any
there
hasn't
been
touched.
Besides
that
so
yeah,
it
seems
like
I
know
nobody
really
files
bugs
again
CSS
base
and
does
anything
about
them,
because.
A
Okay,
let
me
just
run
through
the
last
couple
topics
for
quick.
We
have
a
stale
pull
request
from
you.
I'm
number
look
through
those,
the
oldest
one
that
looks
like
it.
We
just
jump
through
your
some
of
these
and
see
if
they
are
early
in
need
of
attention
or,
if
they're,
just
kind
of
waiting.
There's
one
that's
375
from
ilion
and
that's
really
in
his
court
to
update.
So
we'd
only
need
to
talk
about
that
one,
the
next
one
it
was
from
Ryan
Voss,
and
that
was
165.
F
I
think
I
replied
to
it
saying:
let's
see
it's
276,
that
this
would
be
a
good
candidate
for
data
binding,
so
okay,
so
when
we,
when
we
get
further
along
with
the
data
binding
feature,
then
I
think
that
that's
a
perfect
one
to
revisit,
because
it
also
will
push
on
the
data
binding
implementation
as
well.
So
it's
kind
of
a
good,
a
good
12
to
work
through
then,
okay,.
E
A
minutes
and
then
yeah
that
one
I'm,
let's
see
it's
reading,
I'm
just
going
to
close
that
one
out,
because
it's
not
I
mean
I
created
the
new
ticket,
for
it
all
put
a
description
down
there
that
why
I'm
closing
it
out
but
yeah
I'm,
just
going
to
close
it
10,
okay
and.
A
A
G
A
Requests
that
are
two
weeks
old,
the
only
one
that
that
might
be
of
interest
to
talk
about
in
the
future
would
be
like
the
data
binding
stuff
and
we're
gonna
we're
going
to
visit
that
sort
of
the
next
time
around
I
guess
leastwise
from
Luke
the
outside
that
all
the
rest
of
our
requests
are
not
out
of
date
yet
and
then
in
terms
of
unassigned
bugs
we've
already
talked
about
that
at
length
so
that
pretty
much
covers
everything
or
this
roundtable
must
anybody
has
anything
else
they
want
to
bring
up.
Yeah
I
want
to
talk.
F
Last
night,
the
category
access
to
seattle
in
reverse
order,
yeah.
K
A
F
K
Oh
oh
yeah,
because
I'm
actually
going
to
go
through
it,
and-
and
I
have
I
have
something
pedantic
about
that-
I
want
to
discuss
with
it
next
week
and
and
then
I'm
going
to
after
that,
I'm
going
to
go
ahead
and
change
it
to
triple
equals
like
that,
but
I
just
want
to
bring
it
up
and
kind
of
get
a
little
bit
of
record
I'll
put
it
in
a
pull
request.
Actually,
so,
in
short,
it's
like
actually
in
a
string
string
comparison
I.
Arguably
it
may
be
better
to
do
the
double
equals.
K
If
you
know
that
they're
both
string
values
in
the
because
of
the
name
goddess.
This
is
why
it's
so
ridiculous
with
pedantic
is
that
in
case,
someone
actually
for
some
reason
passes
a
new
string
of
whatever
the
value
is.
It
would
still
work
so
there's
like
a
millionth
of
a
percent
chance
that
it
would
college
a
regression
to
change
all
these
two
triple
equals,
and
and
that's
it
is
that's
all
I
wanted
to
bring
out
before
I,
can't
imagine
that
there's
anyone
doing
that,
but
at.
A
F
I
mean
it's
just
going
to
be
more.
Forgiving
is
essentially
what
so,
what
what
is
the
the
main
issue
here
that
so
its
just
oh
Egyptians,
with
the
bug
yeah.
K
Yeah
yeah
I've
got
the
King
dwell
basically
for
a
kid
categorical
data.
If
it's
a
vertical
axis,
the
dad
would
need
to
actually
go
backwards,
like
from
top
to
bottom
from
the
from
the
range
because
I,
yeah
and
I
just
I
got
the
I
got
the
direction
backwards
when
I
was
doing
factoring
out
unit
tests
at
the
same
time
and
got
it
backward
so
and
it's
something
that
unless
it's
actually
your
chart
that
you're
using
for
something
that's
hard
to
really
catch
with
just
manual
visual
test,
because
everything
looks
fine.
K
F
I
mean
I'm
it,
it
makes
sense.
I
can
do
the
which
gift
should
I
use
this
one
magic
wand,
all.
B
F
K
F
So
I
pinged
a
rape
ango
about
the
whole
I
6pm
thing,
yeah
I've
been
able
to
scroll
up
on
that
Twitter.
For
some
reason.
Why
did
he
say?
Oh,
he
basically
said
he
would
look
into
it
and
I
don't
know
if
he
has
or
not,
but
I
I
would
just
maybe
check
back.
Hopefully,
tomorrow
yeah
I'll
trick
her.
Lastly,.