►
From YouTube: SunPy Developer Discussion
Description
Stuart has written his Thesis, and is officially lead developer again.
It is time to make some progress on 0.7.
A
A
C
A
B
B
B
B
C
A
A
A
B
A
Begin
with
because
we
try
and
keep
the
metadata
but
like
we
I,
don't
know
so
to
some
extent
we
keep
the
metadata
the
same
extent
we
done
and
threads,
but
it
seems
well.
My
argument
with
this
is
that
this
is
working
around
specifically
working
around
a
bug
in
after
pie
or
a
not
bug
so
much
as
a
eaten
incompleteness
for
our
purpose
in
a
drop
I,
in
that
it
just
doesn't
support
their
particular
description
to
coordinate
system,
because
it's
not
standard
compliant
to
the
way
that,
as
provides
me
code,
so
it's.
C
B
C
B
B
So
if,
if
you
read
an
EIT
file
and
the
pitch
header
is
garbage
because
it's
not
up
to
modern
standards
well,
it
will
be
easier
for
us
to
just
keep
our
code
up
to
modern
standards
and
then
sanitize
on
the
way
in
so
that
we
don't
have
to.
We
can
have
a
separate
sanitization
layer,
and
then
we
don't
have
to
worry
about
having
code
all
over
the
place
like
this,
where
we're
trying
to
fix
things
Steven.
If.
C
D
A
A
C
A
A
A
My
put
my
argument
for
a
low
point,
6
patch
on
this-
is
that
it
doesn't
actually
change
our
public
API
because
it
doesn't
change
what
coordinate
type
of
terms.
It's
still
just
returns
the
header,
but
it
just
makes
after
by
WCS
that
we're
now
using
to
do
the
coordinate
transformations
to
do
the
right
thing.
B
A
C
A
C
B
I
mean
the
metal
object,
won't
be
holding
things
as
a
dictionary
with
the
original
keys,
it'll
be
holding
things
differently
and
there
will
be
some
export
utility.
You
know
that
will
pop
it
out
in
maybe
a
fit
there
or
some
other
standards,
and
that
fits
header,
I,
don't
think,
should
a
visitor
needs
to
be
standard.
Yeah.
A
B
A
B
A
C
A
A
A
B
A
A
A
A
A
A
E
A
That
is
a
very
simple
patch
that
just
makes
so
this
is
a
this,
so
the
way
that
they
read
the
docs
and
thing
is
set
up
is
that
the
home
directory
is
not
configured
properly
inside
the
docker
container
that
the
documentation
is
built
in.
So
you
wear,
Sun
pi
by
default,
tries
to
write
its
config
file,
isn't
writable
by
the
user
that
the
documentation
is
being
built
as
which
means
that
you
get
a
sump
I
configure
it
because
it
can't
write
the
config
file.
A
A
This
is
just
changing
the
directories,
because
I
was
trying
to
build
Alex's
magnetic
field
extrapolation
package
on
read
the
docs
and
because
Sun
pi
is
hard-coded
to
use
the
Sun
pipe
clone
directory,
which
wasn't
didn't
exist
inside
the
daughter
image
when
you're
not
building
some
pie,
it
means
that
you
can't
build
the
sub
that
any
documentation
that
relies
on
some
pie
on
read
the
docs.
So
this
just
like
takes
it
back
to
the
writable
directory.
Basically,
it
was
a
mistake.
Last
one.
A
A
A
B
A
B
A
A
A
B
A
A
Start
awful
someone
started
testing
here,
but,
like
just
silent
pages,
is
that
because
someone's
building
one
on
by
side
by
a
map
or
live
before
they
tell
us
a
it
works,
because
if
you've
got
that
far,
it
already
works
well,
I
haven't
I'm,
not
doing
right
now,
anyway,
with
what
Bobby's.
What
was
your
WCS
issues?
Okay,
are.
B
Voice,
okay,
this
has
to
do
with
this
issue.
I
put
up
about
adding
a
shift.
This
is
issue
1537,
so
I
started
working
on
that
and
then
I
sort
of
got
into
all
kind
of
worms,
but
ok.
The
first
issue
is
the
following:
we
want
the
first
question,
for
you,
guys
is
how
we
want
to
handle
this.
So,
as
you
may
have
seen
my
issue
that
I
emailed
to
the
list
earlier
on.
Oh
right,
yeah
I
would
that
was
interesting.
Yeah
I
didn't
it
was
kind
of
silly
of
me,
but
am
ok.
A
B
So,
okay,
so
what
I
want
to
do
is
add
a
shift
variable
right
that
people
can
change
so
that
you
can
shift
your
offense
essentially
in
the
background
you're
shifting
around
the
reference
pixel
right,
so
you
can
move
the
map
around.
However,
you
want
and
the
reference
that
shift
would
be
zero
by
default.
When
you
create
the
map
and
then
you
can
change
it
to
whatever
you
want.
B
Okay,
so
the
first
problem
is
shift
should
likely
be.
It
needs
to
be
a
property,
so
you
need
to
be
able
to
inspect
it,
and
you
want
to
be
able
to
change
it.
So
I
thought
at
first
well,
we
just
create
a
property
and
create
a
setter,
but
the
problem
is
that
we
currently
hold
things
in
the
background
as
pairs.
This
pair
object,
yeah
and
so
to
be
able
to
look
at
the
shift.
It
needs
to
return
a
pair,
so
you
can
ship.
You
can
see
the
shift
as
exit
dot.
Why?
B
B
A
B
A
Amitav
Lee
line
other
places
in
that,
but
that's
horrible,
yeah
I
still
don't
like
the
fact
that
we
have
various
matic
methods
and
take
out
some
wires
separate
argument.
I
much
rather
just
be
providing
like
two
bulls
or
quantities
of
tuples
pond
sorry
to
positive
quantities.
The
so
you
don't
like
the
name.
A
No,
the
name
Thomas
is
good
for
the
way
that
we're
representing
the
data
internally,
as
you
say
me,
it's
just
as
user
info
it's
better
to
take
in.
If
a
third
XY
position,
my
I,
would
have
said
the
best
way
to
do.
It
would
be
to
take
in
a
length
to
quantitative
way
or
something
that
is
bull
comfortable
to
the
length
to
quantity,
right
right.
B
A
B
B
A
B
A
Yeah
he's
to
something
a
great
sluggish,
but
that
is
kind
of
the
point
of
setters
and
getters
in
then
I
love
you
to
do
funky
things
with
attributes
like
that,
whereas
if
we
weren't
playing
around
with
the
input
and
output
like
that,
we
could
just
haven't
a
tribute
on
the
top
yeah
standard
contribute.
Ok,.
B
B
Where
so,
okay,
so
I
have
a
shift
and
I
need
to
apply
it
to
when
you
plot
or
do
anything
you
know,
I
need
to
apply
it.
So
what
I
currently
do?
The
easiest
thing
to
do
is
I
go
in
and
I
change
the
reference
reference
coordinate,
so
reference
coordinate
is
another
getter
and
it
goes
out
and
gets
er
Val,
1
and
0
Bell
to
and
returns
them
so
currently
I
just
add
the
shift
to
those
values
and
that
propagates
everywhere
else,
because
the
WCS
object
that
we
use
uses
that
value.
B
The
weird
thing
now
is
that
if
somebody
looks
and
checks
whats,
ER
Val
one
is
and
then
compares
it
to
reference
pixel,
they
may
not
be
the
same
if
the
shift
is
nonzero
I
agree,
but
that
is
a
good
thing
in
some
sense.
But
now
we've
broken.
We've
broken
this
translation.
What
we
kind
of
changed
to
me
is
like
we
provide
this
transition
layer
right
reference,
pixel
and
reference
ordinate
and
scale.
B
A
With
also
kind
of
broken
it
in
rotation
already,
because
the
rotation
properties
don't
return
necessarily
what's
in
the
header,
because
unless
it's
stored
in
the
header
as
a
PCIJ
matrix,
it
returns
something
different
okay.
So
it's
not
the
first
time
we've
done
one
okay
with
that
easy.
That
is
also
kind
of
the
point
of
the
meta
data
handling
in
map,
in
the
way
that
it's
currently
written
is
that
is
supposed
to
return
the
right
thing,
rather
than
just
the
thing
right,
because
that's
the
whole
point
of
standardizing
it
over
all
the
instruments
right
right.
B
Right
well
so
then
the
other
thing
I
noticed
and
there's
now
I'm
kind
of
coming
out
of
this
particular
discussion
and
broadening
it
a
bit,
but
it
does
seem
that
we
are
currently
doing
a
number
of
very
strange
things
when
WCS
is
involved
so
much
because
WCS
is
very
strange
but
yeah
cool.
Well,
look.
The
strangest
thing
that
we
do,
which
has
nothing
to
do
with
the
strangeness
of
WCS,
is
that
but
WCS
object
is
created
every
time
you
ask
for
it
which
get
starts
to
live
room.
E
B
The
library,
the
object,
the
the
dictionary
header
to
be
kind
of
our
defaults,
and
so
you
don't
want
to
break
you
know
the
translation
between
the
two.
That's
fine,
but
I
also
found
a
number
of
other
places
where
we
use
the
WCS
object
to
calculate
derived
quantities,
but
in
other
places
we
do
it
in
my
hand.
So
like
deta,
deta
to
pixel
and
pixel
to
data
we
use
WCS
but
then
get
center.
We
don't
use
WCS
or
I
still
maintain
what
the
hell
needs
get
center.
A
Go
to
the
center
of
the
map;
no,
it
doesn't
not.
According
to
the
comments
in
the
WCS
library
dozen,
according
to
the
WCS
needs
of
some
pho
WCS
function,
it
returns
the
offset
from
the
center
of
the
map
to
the
center
of
the
image
or
something
of
the
center
of
the
Sun
to
the
center
of
the
image
I.
Don't.
B
B
X
range
is
the
left
edge
of
the
map,
the
right
edge
of
the
map
and
then
divide
that
by
two
and
that's
one
of
the
one
of
the
values
in
center,
and
currently
we
do
that
both
of
those
things,
regardless
of
whether
you
agree
with
or
don't
understand
what
center
is.
We
do
those
all
by
hand
and
we
do
not
use
the
WCS
object
at
all,
because
extra.
A
A
A
B
B
A
B
A
A
A
Which
I
don't
think
is
correct,
I
suppose
that
does
make
sense
when
you
explain
it
like
that.
The
offset
between
the
center
of
the
Sun
in
the
center
of
the
map
is
the
center
of
the
image
in
nice.
Yellow
around
Center.
The
image
and
WCS
in
world
coordinates
is
the
offset
between
the
image
and
the
Sun.
Not
the
Son
and
the
image
all
right
well
I
knew
I
coordinates.
Make
my
brain
hurt
physically
make
me
ache.
E
A
B
A
B
B
F
A
A
B
Different,
so
does
that
suggest
that
we
should
not
spend
any
time
fixing
our
code
to
use
the
VCS
more
correctly
then,
because
well,
we're
just
gonna
have
to
change
all
that
in
such
things,
bah,
it's
a
blowj
pins.
What
you
mean
I
guess
we're
going
to
go
over
to
generalize
WCS
as
well.
So
that's
the
plan.
Well,
I
didn't
have
that
something
to
happen
for
a
while,
like.
A
B
A
Because
I
actually
I
think
that's
probably
wrong
to
the
point
where
that
probably
classifies
as
an
all-points
it
bug,
because
I
think
get
center
is
actually
incorrect.
If
the
maps
not
rotated
is
rotated
yeah,
it
probably
it
yeah.
So
we
should
probably
fix
that
sort
of
bug-fix
release
as
well
yeah
issue
time.
B
A
Yeah
I
mean
yes,
dupuy
is
again
kind
of
the
point
of
generalized
WCS
is
because
a
stroke,
I,
WCS
and
the
WCS
live,
see
library
that
it
wraps
have
been
specifically
coded
for
Fitz
WCS
yeah,
whereas
generalized
WCS
has
been
explicitly
coded
for
not
Fitz
WCS,
because
Fitz
WCS
was
not
up
to
the
job.
I
wanted
to
do
for
james
webb
space,
which
is
why
the
recode,
why
they've
created
a
whole
new
file,
format
and
WCS
implementation?
Oh.
A
Community
wall,
so
they
are
still
going
to
be
so.
The
point
of
the
as
diff
final
format
and
generalized
WCS
is
that
that
allows
them
to
store
the
complete
pipeline.
I,
don't
know
it
allows
I,
don't
know
too
much
body,
but
it
allows
them
to
do
the
transformations
and
store
the
meta
data.
They
need
to
store
to
be
able
to
think
it's
to
do
like
the
full
calibration
pipeline
on
the
data.
A
So
if
you've
got
one
of
those
eyes
do
files,
you
can
run
it
through
generalize
WCS
and
go
from
like
I'm
collaborated
coverage,
or
maybe
it's
only
part
of
that
transformation.
Vein,
so
they're
still
going
to
be
putting
outfits
files,
but
if
you
want
like
the
full
data
with
the
full
transformation,
then
you're
going
to
have
to
get
it
in
as
diff,
because
fits
literally
cannot
store
the
information
required.
Yeah.
A
Agree:
I'm
WCS
has
been
WCS
in
the
coordinate
stuff,
which
is
all
kind
of
bundled
together
at
the
moment,
is
obviously
being
a
top
of
my
head
of
this
for
a
while
cuz
they
can't
stop
I've
been
working
on
so
I
do
agree.
There
is
some
major
issues
there
and
the
job
is
far
from
done.
Yeah
and,
of
course,
supporting
generalized
WCS
properly
is
the
answer
to
all
our
Q
blows.
C
A
E
A
A
A
B
A
I,
like
their
earlier
for
models,
models
he's
just
amazing.
It
is
so
powerful.
Now
they've
gotten
composite
model
fitting
as
well.
That's
the
big
thing
came
out
in
one
point,
know
that
turned
it
from
autumn,
like
quite
nice
to
absolutely
awesome,
because
you
can
like
say
you
wanted
to
do
a
multiple
gaussian
fit.
You
create
four
instances
of
gaussian
and
then
add
them
together
and
then
fit
it.
It's
very
nice.
B
B
A
B
B
A
E
A
A
A
E
B
E
A
And
then
so,
once
that's
merged
in
so
he's
currently
the
summer
student
pull
request
is
to
the
uni
down
branch,
which
adds
a
load
of
extra
sources
and
stuff
to
light
curve
in
uni
down,
and
then
it
is
a
case
of
merging
knew
if
I
downloader
in
so
which
so,
where
are
we
going
to
do
so?
You
need
down,
doesn't
currently
break
any
existing
AP
I.
A
Don't
think
I
mean
it
change
you
some
stuff,
but
think
all
of
the
current
like
of
this
code
and
think
of
the
current
like
I've
gotten
v
and
everything
is
still
there
still
is
a
copy.
Some
stuff
replicate
some
functionality
and,
according
to
my
to-do
list
on
issue
1300,
there
is
currently
the
API
documentation
needs
doing
verify
the
you
need
an
sources
or
today,
with
light
curve
sources,
test
ability
to
download
different
levels
of
Laura
and
Eve
data
and
testing
to
make
sure
it
plays
nice
with
Eve
where
the
clients
conflict.
A
A
B
A
A
Yeah
well
in
five
minutes,
you
could
do
one
of
the
check
boxes,
which
is
here,
you'd
verify
the
ability
to
download
different
levels
of
Lara
and
you
data.
That's
just
a
case
of
running
like
for
queries
through
the
pipeline
or
might
take
more
than
five
minutes.
If
you
got
explode
the
branch
and
read
the
documentation
to
remember
how
it
works,
but
it
works
not
big
job.
Well,
where
is
issue
of
everything
hundred
I?
Don't
see
it
smile
stone,
no
point:
seven
is
pull
request.
1300,
porpoise.
D
B
A
A
As
I
say,
we
should
get
this
if
we're
going
to
get
no
point
7l
in
a
sensible
amount
of
time.
This
has
to
be
merged
into
master
soon,
yeah,
because,
obviously,
if
there's
a
lot
more
work
than
he's
doing
on
top
of
it,
and
that
work
can
progress
easier
and
in
parallel
as
soon
as
this
is
in
master,
rather
than
constantly
waiting
for
stuff
to
be
caught.
On
top
of
this
yep
agreed
and
to
some
extent
the
same
goes
for
coordinates.
We
should
get
the
coordinates
PR
merged
in
as
well.