►
From YouTube: SunPy Developer Discussion
Description
There are many PRs and we need a plan of attack for 0.6.
A
B
A
C
E
So
those
well
one
percent
seems
kind
of
high
in
kind
of
bad.
But
it's
certainly
true
that
the
ideal
outfit
only
goes
out
to
a
few
digits
like
not
as
many
as
as
as
Python,
so
they.
C
B
E
A
B
E
E
C
E
B
C
C
C
A
It's
a
good
goal
to
have
right
to
go
for
it.
The
worst
can
happen.
Is
we
fail
miserably
to
be
burning
ball
of
fire?
So
let's
have
a
look
at
what,
under
the
null
point,
six
issue
milestone.
Joey
serious
think
about
this.
B
D
C
A
B
A
E
B
E
F
F
F
A
different
graduation
dependent
of
the
precision
you
want
so
I'm
trying
to
do
the
best
I
can,
but
the
ender
is
the
problem
with
time
and
time
in
all
the
question
that
we
have
are
like
time
in
Julian
dates,
but
actually
it's
not
Julian
dates
is
Julian
dates
in
the
TT
frame,
which
is
that
a
earth
time
frame
know
the
UTC
frame,
no
frame.
Sorry
I'm
much
cool
tail
so
which
there
is
a
correction
to
do
with
within
UTC
and
this
TT.
F
B
A
A
A
B
B
E
I
A
A
The
fact
that
and
there's
some
people
are
going
to
write
a
project
to
the
turn
hyphen
code
snippets
into
a
gallery.
But
people
like
this,
I
clear
image
down
at
the
moment.
Scikit-Learn
doesn't
up
so
someone's
got
some
cash
they're
going
to
hire
a
student
and
try
and
build
a
general
project
based
on
Sphinx
building
code
guns
out
other
our
Python
scripts,
rather
than
app
notebooks,
and
one
of
the
things
they
might
do
is
make
it
so
that
you
can
write
a
Python
script
and
it
will
allow
you
to
create
a
notebook
for
download.
A
Now
that
that's
cool
and
the
psychic
image
people
have
a
late,
remote
execution
environment,
so
you
can
fire
up
their
example,
change
the
parameters
and
be
executed,
yeah,
which
is
very
cool.
So
it's
decided
to
be
possible.
We
might
scrap
our
entire
current
color
implementation
and
go
with
something
else
in
a
couple
of
versions
time,
but
what
we
have
met
up
for
the
moment
works.
Ok,.
B
A
A
B
A
G
A
A
E
A
A
D
A
A
A
A
A
A
A
A
A
A
A
A
D
E
I
F
E
A
E
Yeah
I
mean
so
the
issue
the
issue
is
like
is
just
that.
It's
really
easy
to
code
up
our
own,
and
so
you
know
the
question
is
like
well,
if
it's
so
easy,
why
not
do
it
and
and
Andy
has
done
a
first
cut
at
it
I?
I
agree
I
just.
I
think
we
need
to
talk
about
this.
A
little
bit
more
I
would
bump
it
because
we
don't
need
to
worry
about
this
now.
Volleyball.
A
G
A
F
F
G
A
F
A
A
F
Right
so
they
occupy
time
objects
are
kind
of
rare
quantities
they
design
with
the
time
thing
in
a
strop
I.
It
can't
behave
like
quantities
as
well.
If
they
also
can
extract
the
value
yeah.
F
A
I
mean
shipping
over
trash
my
time
and
have
to
fight
time
delta.
It's
definitely
the
way
forward
and
maybe
you
can
considering
sending
time
range
into
a
strobe
light
once
we've
done
that
and
but
I
would
argue
for
the
moment.
We
stick
to
time
delta
objects
in
day
time,
rather
than
quantities
in
seconds.
Well,.
E
H
E
E
C
D
A
E
A
E
Should
we
just
so
it's
a
little
bit
clearer?
Can
we
add
issues
for
the
big
things
well,
for
what
we're
going
to
actually
what
Google's
approach
things
I
was
thinking
we
do
we
need.
We
need
some
more
issues.
Stephanie
go
so
it's
easier
to
see
them
there
and
so
that
we
know
what
we
actually
have
to
do.
A
We
know
if
I
can
support
vessels
but
yeah
right,
but
to
me
said
we
should
try
and
distinguish
between
pull
requests
and
images
and
pull
regressing
issues
less
right,
because
you
can
actually
attach
commits
to
issues
and
turn
them
into
pull
requests.
And
just
you
have
to
use
some
command
line.
Magic.
E
Okay,
I
didn't
know
that,
but
pull
requests
well
now
that
they've
now
that
they've
broken
out
issues
and
pull
requests,
I,
don't
know,
seems
a
little
bit.
Stranger
like
a
pull
request
should
be
associated
with
an
issue
really
should
be
fixing
something
or
it.
A
A
D
E
A
E
A
E
A
E
Yeah
yeah
I
mean
I'd
rather
as
well,
definitely
and
when
the
problem
is
that
you're
involved
in
both
of
them
right,
you're
the
weakest
link
road
right
now,
oh
yeah,
so
I
guess.
The
question
is:
if
you
have
to
prioritize
your
time
right,
whatever
a
little
amount
of
time
we
can
get
from
you.
Would
we
rather
you
work
on
the
coordinates
or
on
the
light
curve,
factory,
stuff,
I,
think.
A
A
C
C
H
C
C
C
C
H
E
Well,
if
we
think
about
maps
and
map
camp
allure
map-
well,
I
guess
composite
map
is
kind
of
weirder
cuz.
It
mixes
different
things,
but
what
I
would
expect
from
a
map
cube
is
in
that
cube.
We
keep
the
original
maps
right.
So
you
can,
you
can
get
metadata
of
the
original
maps
and
then
we
have
a
higher
level
version
of
the
keywords
which
are
broken
out,
but
like
observatory,
lalala,
look.
A
A
No
metadata
attribute
okay,
well,
lady
apple
picking,
I
didn't
have
a
full
going
off
on
a
little
tangent
here.
I
did
think
we
could
implement
a
bit
of
witchcraft
that
allows
you
to
broadcast
any
operation
which
isn't
a
tribute
or
a
method
of
generic
map
on
a
Matthew,
so
you
could
do
Matthew
rotate
and
if
we
could
turn
you
a
Matthew
where
it
applied
rotate
on
every
map
in
the
Matthew,
oh
and
the
same
goes
with
resample
sub
map.
Everything
is.
E
A
That
would
do
is
I
would
write
a
piece
of
magic
that
I
would
write
again
with
overload
get
a
tribute
with
an
that
class
Matthew
class.
So
when
you
try
to
access
and
attribute
with
the
man
cube
class,
which
is
either
an
attribute
or
a
method,
that's
settling.
So
when
you
try
to
access
it,
if
it
doesn't
exist
in
Matthew,
but
it
does
exist
in
generic
map,
it
will
run
that
for
you
before.
You
then
returns
all.
E
Yeah,
so
what
I
was
thinking
is
essentially,
we
have
some
of
these,
which
currently
for
light
curve.
We
don't
have
is
like
higher
level
broken
out
metadata
right
luck,
instrument
we
have
for
math
right,
and
we
only
update
those
and
updating
them
would
be
fairly
easy
because
you
know
start
time
and
time.
E
D
A
Just
have
a
dictionary,
no
time
range
object
and
the
meta
data
dictionary.
You
know,
maybe
you
can
put
it
and
then
you
could
do.
Then
you
could
have
a
get
metadata
function
where
you
specify
a
time
and
for
every
key
in
the
dictionary.
It
does.
Is
this
time
in
the
time
range?
If
yes
return
it
by
way.
C
A
D
A
D
A
H
A
That
behavior,
the
easy
thing
to
do
would
be
when
you
index
the
light
curve
metal
objects.
You
check
whether
the
time
is
in
between
the
start
time
and
the
end
time
for
the
mathematical
different
and
if
it
isn't,
you
explode,
even
though
the
meta
data
objects
held
in
the
Olmecs
dictionary
I'll
for
a
longer
time
range
than
the
parent
meta
data
object.
E
A
E
E
I
mean
the
way
that
I
kind
of
always
thought
about.
It
is
the
way
we're
kind
of
headed
with
map,
which
is
that
with
map
we've
got
the
fits
header
which
is
saved,
and
then
we
pull
out
some
of
the
key.
You
know
metadata
keywords
into
our
own
keywords:
right
though,
instra
becomes
instrument
and
blah
blah
blah
right
and
then
and
I
always
thought
that
we
would
kind
of
eventually
define
a
unified
meta
header,
which
we
could
then
have
procedures
to
export
to,
for
example,
its
header
right.
E
We
would
know
the
the
translation
between
our
keywords
and
the
standard
fits
keywords,
for
example,
and
we
would
essentially
just
keep
around
the
original
fits
header.
It's
just
you
know
as
and
so
those
those
are
meta
keywords
would
be
the
ones
that
get
updated
and
stuff.
As
we
change
things,
we
don't
touch
the
original
ones
and
then,
when
we
want
to
export
or
something,
then
they
we
have
this
transition
to
go
to.
You
know
header
a
fits
header
from
ours
that
have
been
kept
up
to
date.
D
H
E
E
C
D
E
A
C
A
I
have
an
idea
and
I
will
flesh
it
out
and
I
have
an
idea
for
an
intermediate
fix
that
allows
us
to
do
that,
like
basically
just
a
fancy
way
of
keeping
tabs
on
the
dictionaries
for
each
of
the
maps
when
we
hadn't
have
I
each
of
the
light
goes
on
the
add
them
together.
Okay
in
a
nice
way
presents
a
neat
FBI
for
doing
it,
I'm
the
flush
it
out
and
add
it
to
the
NCP
Julian
be
25
to
27
hours.
Today,.
E
Yeah
and
so
the
so
the
idea
is,
we
want
coordinates
in
0
points
X.
So,
if
coordinates
is
not
ready,
we
would
probably
delay
Oh
point
6
the
light
curve,
factory
and
downloader.
Would
we
delay
Oh
point
six
for
that
I
would
say:
let's
accept
the
downloader
and
then
we
can
wait
push
the
factory.
If
you
know
you
said
it's
easy,
but
I.
A
C
A
There
is
a
question
of
help.
How
can
we
deal
with
the
code
duplication
in
that?
Do
we
just
keep
two
versions
of
the
download
code
for
one
version,
a
version
in
download
and
aversion
in
my
curve,
or
do
we
somehow
make
the
current
like
of
API
patch
into
the
new
downloader
I
know
we
are
too,
I
prefer.
I
would.
E
C
A
A
E
C
D
E
C
G
A
A
conflict
on
it
at
the
moment
up
people
reviewing
1041
would
be
nice.
I
have
a
branch
of
improvement
to
1041,
but
I
want
to
merge
1041
first,
because
I
don't
have
time
to
do
it
currently.
So,
if
you
merge
1041
at
the
moment,
I
will
continue
working
on
my
improvements
or
1041
and
write
the
tests
for
them
and
then
issue
a
separate
for
quest
right.
G
B
E
You
Stuart,
could
you
email
me
where
the
files
are
gonna,
be
URL?
That's
what
you're
gonna
tell
me
right.
Yeah.
G
E
A
A
A
E
F
Yes,
Andy.
We
have
to
look
at
the
at
the
website
the
workers
they
have
to
at
least
the
summary
clocks
of
each
of
the
meetings
yep,
the
head,
like
a
small.
It's
like
a
Peter
have
been
done
over
the
last
few
meetings.
They
have
like
a
final
confrontation
kind
of
a
even
every
information
about
everything
that
has
been
discussed
and
there's
another
point
that
we
should
look
into
that
and
see
how
we
can
tackle
them.
Yes,.
F
F
Don't
know
it
is
that,
isn't
that
the
file,
using
mostly,
for
instance,
okay,
so
but
there's
a
lot
of
instruments
in
situ
there,
so
yeah
I
could
be
something
up
and
we
can
look
into
that
or
we
can
never
die
as
stores
a
very
saying,
sci
fi.
This
ability
to
read
that
and
now
then
yeah
we
have.
We
have
to
look
at
that
we
should.
Maybe
I
will
try
to
look
that
as
well
tomorrow
and
led
to
make
an
awfully
such
things
that
we
should
look
yeah.
D
A
Probable
thing
with
some
heiio
is
the
assumptions
we
make
in
some
pie.
I
oh,
don't
necessarily
fit
the
general
netcdf
file
format
now
specific
netcdf
file.
We
could
make
assumptions
about
the
cdf
format
and
make
it
fit
into
my
IRA,
but
it
would
probably
be
better
to
keep
CDF
freedom
right
outside
some
fire.
We
have
to
cross
that
vision,
yeah.
B
F
C
F
Well,
I've
never
been
personally
in
the
meeting,
but
peter
has
been
trying
to
represent
the
meeting
all
day
and
all
them,
and
the
thing
is
that
at
the
beginning,
the
first,
the
first
meeting
they
have,
they
have
a
discussion
of
whether
the
use
IDL
or
something
else.
That
is
question
last
20
seconds.
Oh.
F
I
dl
yes,
but
that
was
most
mostly
on
them
on
the
tasting
face,
so
what
they
have
to
do
now,
so
how
they
have
to
design
or
threshold
a
the
dots
are
the
demo
data
they
are
producing
and
other
side.
You
know
like
they
have
become
a
cello
players
and
then
they
have
tried
to
detect
it
and
knowing
this
test
data
that
they
are
building
and
and
the
code
that
they
have
to
build
because
anyway,
revel
in
the
code
in
atl,
and
they
have
to
translate
it
into
some
kind
of
see.
F
They'll
be
using
this
at
the
line.
Right
right
is
the
satellite
yeah
yeah.
They
hiring
work
as
far
as
I
know
they
can
work
on
the
on
the
calibration
and
I
see
Sir
in
the
pipeline
that
I'm
going
to
do
once
that
we
get
it
up
on
earth
and
that's
where
we
have
to
show
that
we
can
do
that
right.
So
they
might
there's
still
potential.
D
D
F
F
But
it's
not
just
sulfide,
that's
how
it's
get
up
the
element
and
then
yeah
I
will
take
a
look
at
the
dog
again
and
see
kind
of
label
stuff
and
make
some
kind
of
nonsense.
If
you
see
you
and
the
rest
of
people.