►
From YouTube: SunPy Developer Discussion
Description
0.6 release!
A
A
C
A
A
C
D
E
B
B
F
E
B
E
E
E
G
D
E
A
A
B
A
E
A
B
B
E
E
B
E
A
B
A
A
C
A
A
B
A
A
C
C
B
E
A
B
B
B
B
So
it's
like
a
it's
like
a
standard
web
thing
and
what's
really
nice
is
that
since
it's
a
vector
format,
it
doesn't
depend
on
resolution
at
all,
so
you
can
essentially
zoom
in
it's
super
sharp
and
everything
almost
all
browsers
now
can
render
it.
So
you
don't
need
to
like
a
PNG
or
something
just
include
the
SVG
so
but
it
does
just
like
a
web
web
page
like
you
can
say
you
know
this
text
here
and
this
text
is
what
made
with
this
font
and
just
like
any
font
on
the
web.
B
If
you
don't
have
it
installed
locally,
then
it
has
to
put
in
something
else
there,
and
so
that
would
mean
that
the
logo
could
kind
of
change.
If
you
didn't
have
the
font
available,
it
would
like
some
pie
would
get
rendered
differently,
which
is
weird
not
not
really
what
you
want
from
a
logo,
so
I
changed
it.
So,
basically,
now
I
removed
the
reference
of
the
font,
and
so
it's
basically
like
a
think
of
it,
like
a
rendered
shape
now
rendered
it's
just
a
shape
of
the
letters.
B
They
all
know
propriety
is
not
the
right
word.
It
is
coakley
reasons
yeah
they're
copyrighted,
but
with
a
very
complete
permissive
license
yeah,
you
know
there
I
think
they're
as
copyrighted
is
like
Times,
New,
Roman
and
aerial,
like
all
the
other
fonts
that
you
use
throughout
the
web,
that
you
don't
think
about
because
you
know
the
licensing
is
so
easy.
C
B
B
B
C
A
A
B
Well,
it's
only,
I
reduced
the
royalty
to
the
smallest
possible
value
that
they
would
allow
me
to
do,
which
is
five
percent,
and
yes,
it
technically
does
go
into
my
paypal
account,
but
you
know
I'm
not
going
to
keep
it.
We
can
use
it
for
whatever
we
want.
On
the
other
hand,
I've
I've
been
through
this
before
and
we
had
I
build
some.
I
sold
some
stuff
like
this
before
for
another
project
and
I
saved
up
as
high
as
26
dollars.
Oh,
but
what
we'll
do
is
next
time
we
meet
together.
B
A
F
H
B
F
B
E
D
A
B
A
E
C
B
C
C
A
F
F
A
A
B
A
F
A
Point
6
says:
integrate
after
pie,
coordinates
didn't
happen,
functions
to
exceed
except
test
reply.
Quantities
did
happen,
include
how
to
quality
so
for
a
fairer
eyes
didn't
happen,
replace
our
picks
with
low
transformation
functions.
Bastrop
I
WCS
did
happen.
It
enables
storing
light
curve
data
in
the
database.
I
think
that
happen.
Oh.
F
F
B
F
A
A
Want
to
release
them
is
no
point,
seven
with
one
there's
no
point.
Seven,
and
one
is
no
point.
Eight,
that's
fine.
We
can
do
lost
with
small
releases,
but
they're
not
point
releases.
Well,
what
releases
are
stable,
API
bug
fixes,
as
he
should
break
between
no
point.
Six
ignore
point,
six
point:
one:
that's
fine!
No
code
will
break
or.
A
A
B
A
C
F
F
A
A
F
A
B
E
F
F
G
F
B
G
A
A
And
mighty
way,
cuz
he's
got
someone
who's
working
on
that
at
the
moment,
no
parse
it
that's
the
other
thing
like
this
whole
hot.
The
whole
cubes
plan
is
very
fragmented.
At
the
moment,
we've
got
people
scratching
specific
issues
for
specific
instruments
and
there's
no
coordinated
plan
right
and
I'm,
suggesting.
B
The
whole
fragmented
thing
well,
what
I
suggest
is,
we
add
an
iris.
We
add
iris
support
in
the
instrument
package
in
instrument
side
and
then
to
do
like
basic
things
like
being
able
to
view
you
know
the
SGI
images
we
can
do
some
of
that
already,
but
to
be
like
usually
the
way
you
want
to
work
with
iris
data,
you
you
want
everything
linked
together,
so
you
want.
You
want
the
SGI
images
at
the
same
time
as
you
can
look
at
the
spectrum
that
comes
from
the
actual
slit.
A
A
B
A
E
A
B
E
B
A
Time
you
need
in
it
then
see
I
agree.
So
we
should
keep
in
mind
that
we
roughly
want
to
go
six
months,
which
is
why
I'm
saying
we
probably
don't
want
to
say
we
want
to
get
a
whole
cube
module
into
north
point
seven,
because
you
ain't
ever
going
to
happen,
and
it's
not
good
wrapping
in
the
next
six
years.
Unless
we
get
some
more
programming
time
on
it,
let
alone
the
next
six
months,
yeah.
A
C
B
B
If
we
can
say
that,
oh
no,
no,
no,
you
can
use.
In
fact
you
can
use
some
pie
free
and
everything
that
would
possibly
bring
in
a
lot
of
people
over
there
and
then
also
one
of
us.
Well
somebody
from
Goddard,
because
as
Goddard
money
could
be
funded,
to
go
to
go
to
India
and
actually
give
a
tutorial
on
some
pie.
Support
for
iris.
C
A
Or
broccoli
look
mmmmmm,
you
won't
start
using
Python
3
until
we
support
Python
3,
so
we
can't
go.
You
should
be
using
Python
3
because
we
don't
support.
We
have
almost
every
other
major
pipe
of
atra
at
Python
project
in
the
scientific
Python
ecosystem
is
now
supporting
Python
3
and
there
are
a
lot
of
features
coming,
especially
in
Python
3.5,
which
are
really
good
for
scientific
computing.
Like
a
matrix
multiplication
operator,
that's
a
first
off
type
hints
lot
of
all.
The
new
development
is
happening
by
3.
We
gotta
go
there
shortly.
C
C
See
it
unless
and
the
wiki,
then
I
would
like
to
see
those
reasons
for
each
of
that
from
each
thing
that's
under
the
the
in
the
wiki.
Let
me
just
rattled
off
a
whole
bunch
of
important
reasons
why
we
should
be
supporting
three
point
X
same
for
everything
else.
That
will
give
us
an
idea
of
what
actually,
what
what
we
want,
what
the
communities
are
asking
for.
What
you
know
give
us
some
idea
how
important
it
is
and
the
feasibility.
C
A
H
A
F
A
Okay,
yeah
sure
we
can
put
stuff
in
the
wiki.
That's
good.
I
don't
discreet
it's
good
idea,
but
we
can
also
have
that
discussion
in
open
with
the
power
of
Audio
right
now
and
write
up
note
layer.
So
the
current
things
squabbling
four
times
in
4.7,
our
astra
x
coordinates
Fido,
iris
support
and
python
3
ephemerides,
but
that's
really
just
david
getting
it
fixed.
So
I
would
argue
that
we
don't
hold
no
points7
on
the
ephemerides.
That
is
a
when
it's
done.
It's
done
job.
C
B
A
In
saying
that
we
don't
release
no
point,
seven
until
we've
done
like
urban
53,
I
don't
think
I
can
freeze
of
ever
push
actually
I.
Quite
all
Python
3
is
not
hard.
It
just
needs
time.
It
just
needs
a
group
so
that
that's
a
problem.
Most
we've
come
back
to
five
and
three
in
a
minute.
That's
discussed
like
a
fit
so
light
curve.
I
mean
the
like
so
like
curve
when
we
say
improve,
like
of
what
we
mean
is
create
a
factory
interface
like
map
right.
That's.
C
B
A
In
theory,
but
in
practice
that's
just
going
to
ugly
like
it's
good,
that's
going
to
be
the
API
will
be
a
mess
like
this.
You'll
have
you'll,
have
concatenate
flags
for,
like
a
the
cube
and
composite
flag
from
that.
You'll
have
to
worry
about
mega,
be
an
option
for
manually
overriding
the
data
source.
You'll
have
to
worry
about
having
one
factory
registration
for
everything
that
make
sure
you
don't
have
two
things
that
return
valid
for
every
source.
It
would
be
messy,
I
mean
come
on
when
you
load
the
data
type.
A
A
B
A
A
A
A
A
A
The
make
one
of
the
reasons
why
it's
quite
easy
is
because
there's
one
that
already
exists.
It's
called
the
map
factory
in
to
a
large
extent,
you're,
basically
just
got
a
copy
and
paste
it
and
then
add
in
the
next
few
extra
bits.
Yeah
a
couple
of
the
edge
cases
will
probably
get
tricky,
but
it's
not
something.
That's
particularly
complicated.
A
E
B
One
other
thing
that
I
think
we
should
add
which
actually
I
just
want
to
call
out
my
name
because
I
I
think
it's
included
in
this,
so
improved
support
for
the
data
type
will
be
we'll
have
to
we'll
have
to
fix
meta
right.
So
meta
in
like
her
right
now
is
kind
of
crap
and
that's
not.
Our
fault
is
because
meta
doesn't
really
come
down
easily,
but
we
need
to
create
what
I
think
would
be
great
is
if
we
created
a
metal
object
that
was
the
same
across
like
curve
and
map
I.
A
B
A
About
with
that,
metal
object
is
actually
to
a
large
extent
just
what
map
does
like
the
vast
majority
of
code
in
generic
map,
especially,
is
handling
and
is
abstracting
metadata
yeah,
so
I'm
not
disagreeing
that
it
may
be
something
we
should
look
at
doing.
I
just
think
that
again
is
another
design
project,
because
I
mean
you
started
working
on
an
SCP
for
it
and
that's
what
it
needs
it
needs.
It
needs
a
design
effort
that
does
because
it's
not.
B
A
We
need
to
abstract
that
away.
That's
the
other
thing
that
needs
to
that.
That
needs
to
be
abstracted
away.
I,
don't
like
how
dependent
map
is
on
that
at
the
moment,
because
things
like
generalized
WCS
and
the
cubes
shouldn't
have
to
follow
that
standard.
That's
another
problem,
I'm,
not
sure,
I,
understand
what
your
a
foul
again.
B
A
Not
okay,
hang
on
wait.
A
second
before
I
go
off
on
one
I
need
to
actually
make
sure
that
I'm
not
going
to
be
saying
something.
That's
not
what
you're
saying
so.
Are
you
suggesting
that
the
metal
object
provides
a
unified
interface
to
the
dictionary
or
it
just
holds
the
raw
dictionary
and
whatever
keys
that
may
have.
A
A
B
A
A
A
The
basic
solution
is
tita
when
you
concatenate
to
like
us
together.
You
have
two
sets
of
metadata
and
you
only
combine
the
date
page
one
but
I
mean
that's
the
most
basic
solution
and
we
can
release
a
factory
in
that
site.
Okay,
that
can
be
done.
You
don't
have
to
have.
We
don't
have
to
solve
this
problem
too.
Lisa,
like
a
factory.
F
A
A
C
C
B
B
C
B
C
A
I
think
that
that
is
like
I
agree
that
that
is
obviously
a
really
nice
feature
to
have
in
my
coke.
My
point
is:
don't
know
whether
we
want
a
halt
nor
point,
seven
on
that.
That's
in
what
sounds
more
like
a
like
a
factory
to
point
o
problem
for
me.
As
long
as
we
have
a
functional
version
of
it,
we
should
release
nor
point
seven
I.
A
E
C
Yeah,
so
we
shouldn't
let
the
perfect
be
the
enemy
of
the
good
range
is
we
gave
something
down,
but
that
starts
the.
This
is
going
to
be
version
2
or
see
a
lake
over
enemy
by
the
end
of
0.7
yeah.
When
you
think
about
the
questions
that
we've
had
so
and
writing
the
light
curve
data
object,
it
will
find
out
what
metadata
we
can.
No
iron
from
from
all
types
of
white
cotton
and
then
from
man
will
be
from.
G
G
Even
when
you're
creating
my
curves,
like
some
some
like
objects,
take
the
date
Alex
kind
of
system
like
silly
but
like
some
Kate
the
date
and
some
take
a
full
start
time
strain
or
you
have
to
truncate.
It's
not
consistent
and
could
do
it.
You
have
to
just
kind
of
play
around
with
it
until
it
works.
Yeah.
A
So
now,
yes,
that's
was
sitting
home
is
something
that
is
are
potentially
fixed
by
the
light
directory
yeah,
because
the
reason
for
that
at
the
moment
is
that
the
way
like
I've
is
implemented.
If
the
provider
only
provides
date
in
whole
days,
be
a
little
too
if
it
provides
arbitrary
time
frames.
Oh
yeah.
E
C
A
B
Yeah
I
do
think
we
should
talk
about
some
general
generally.
We
need
to
talk
with
the
board
about
you
know
is:
does
everybody
agree
that
light
curve
should
be
the
thing
40
point,
seven
and
and
agree
to
have
people
agree
on
the
six-month
type
of
timeline
and-
and
some
general
discussion
of
like
where
we
are
headed,
would
be
also
really
good.