►
From YouTube: SunPy Coordination Meeting 2022 - Tuesday
Description
Participate in the chat and the call here: https://openastronomy.element.io/#/room/#sunpycoordinationmeeting:openastronomy.org
A
A
B
A
A
I
don't
know
if
you
can
hear
me,
but
you're
broken
up
or
maybe
it's
whatever
way.
I
have
jc
set
up,
but
I
can
barely.
Oh,
my
god.
It's.
A
A
A
F
Cool,
I
guess,
maybe
just
in
the
interest
of
time,
we'll
get
started.
Actually
danny
also
wants
to
talk
a
little
bit
about
some
stuff,
so
we
changed
the
schedule
for
those
online
if
you
didn't
see
before,
as
this
was
shortened
down
until
quarter
two
but
we'll
see
how
we
get
on.
So
I
guess
just
the
first
morning.
I
think
we'll
just
start
with
kind
of
talking
about
kind
of
some
point-
outreach
kind
of
retrospectively
like
what
we've
done.
F
What
we
can
do-
and
we
kind
of
touched
on
a
lot
of
this
last
week-
and
I
think
outreach
is
maybe
we
want
to
kind
of
talk
more
about
engagement
rather
than
outreach,
because
not
just
most
outreaching.
It's
like
we
want
feedback
or
whatever
use.
So
I
guess
in
terms
of
who
we
want
to
engage
with
it's
not
just
users,
but
maybe
users
as
maybe
who
we
would
think
of
straight
away,
and
also
this
is
really
informal
anybody
jump
in.
If
I
really
miss
something
here,
so
I
guess
users
will
be
researchers.
F
You
know
phd
students,
even
undergrad
students,
master
students
and
pis
things
like
that,
and
we
also
want
to
engage
with
like
future
developers.
So
how
do
we
go
from
a
user
to
somebody
that
wants
to
contribute
code
and
then,
in
the
future
become
you
know,
a
contributor
to
senpai
and
maybe
a
kind
of
a
significant
developer?
And
this
is
kind
of
needed,
I
guess
for
sustainability.
F
We
also
really
want
to
talk
to
instrument
teams,
and
this
is
something
that's
becoming
more
prevalent
with
more
instruments
coming
online,
more
like
cubesat
ideas
and
things
like
that.
We
want
to
kind
of
start
off
early
when
they're
writing
software,
so
that
it
can
be
written
in
such
a
way
that
kind
of
builds
in
nicely
with
the
senpai
ecosystem
and
then,
similarly,
you
want
to
talk
to
data
providers
and
people
like
bso.
F
We
talk
to
people
like
they
that
are
running
the
solar,
orbiter
archives
and
how
that
can
kind
of
again
fit
in
within
the
senpai
ecosystem
and
be
useful
for
these
users
and
and
the
community
in
general,
and
I
guess
why
you
wouldn't
engage
with
them.
Is
I
mean
this
list
goes
on
and
on?
But
you
know
it
is
a
community
that
simply
is
and
we're
trying
to
grow
this
community.
F
So
it's
all
well
and
good
to
write
your
software
that
people
use,
but
you
wanted
to
do
more
than
software
and
to
be
a
community
of
kind
of
a
feedback
loop
and
make
sure
we're
kind
of
providing.
I
guess
the
software
we're
not.
We
like
that
the
community
has
the
software
available
and
that
everyone
can
contribute
to
it.
F
F
So
it's
not
like
this
bus
problem
that
if
one
person
goes
that
the
thing
doesn't
fall
apart
and
we
also
want
to
engage
so
that
we
find
out
about
issues
in
bugs,
and
you
know
I
feel,
like
we've
all
kind
of
heard
stories
of
people
saying
that
oh
there's
a
big
issue,
so
I
fixed
it
myself
or
I
didn't
work.
So
I
moved
to
something
else.
F
So
we
want
to
make
sure
that
we're
engaging
with
people
that
have
those
bugs
and
that
they
tell
us
that
they're
there
and
I
also
want
to
find
out
what's
needed
in
the
community.
You
know
I
mean
it
is
biased
by
what
people
who
contribute
code
views.
So
I
mean
maybe
that's
why
there's
a
lot
more
remote
sensing
than
situ
data
and-
and
I
guess,
work,
communication
teams
from
the
ground
up
et
cetera,
there's
lots
of
reasons
why
we're
going
to
engage
with
the
community.
F
I
don't
think
we
need
convincing
there
and
I
guess
how
we've
done
this
so
far
I
mean
we
have.
We
talked
about
this
a
lot
yesterday.
We
have
lots
of
forums,
and
last
year
we
are
looking
back
in
the
notes
we
were
like.
Is
this
too
many
forums
like
what
is
the
go-to
thing?
If
we
ask
if
someone
says
who
do
I
contact?
F
What
should
be
our
first
thing?
Should
it
be?
The
chats
should
be
the
main
list,
github
discourse.
Maybe
we
should
push
the
discourse
a
little
bit
more
and
maybe
it's
good
having
lots
of
different
options,
because
there
are
lots
of
different
ways.
People
might
want
to
contact
your
issues
that
they
have
like
if
you
have
community
meetings
and
which
again
are
usually
just
people
in
this
room
that
are
on
that
call,
and
I've
done
the
past.
F
You
know
doing
tutorials
at
meetings
and
meetups
at
that
kind
of
workshop,
whereas
at
big
conferences
like
tess
or
agu,
or
something
like
that
do
a
twitter
which
is
quite
active.
So
if
he
was
really
good
at
kind
of
pushing
that
and
I
think
one
of
the
biggest
ways
I
think
we
do
engage
with
the
community-
and
maybe
this
is
maybe
bias
from
my
own
experience,
but
it
really
is
like
personal
connections.
F
If
you
know
someone,
that's
working
on
sunfire
or
the
developments
on
fire
knows
how
to
use
some
pi
you're
more
likely
to
engage
and
tell
them
about
feedback
and
get
involved.
I
mean
that's
my
own
experience
of
getting
involved.
If
I
was,
you
know,
actually
talking
big
to
you,
danny
and
people
and
then
steve
and
jack.
F
G
C
At
this
point,
but
it
could
and
whether
it
should
or
not
is
something
we
can
discuss
like.
We
have
two
mailing
lists.
At
the
moment
we
have
senpai
and
some
paid
dev
senpai
dev
gets
approximately
zero
emails
per
any
length
of
time.
So
I
think
we
can
pretty
safely
murder
that
one
and
replace
it
with
the
discourse
which
to
be
fair,
actually
isn't
the
case,
because
any
dev
stuff
happens
on
github
and
people
who
are
interested.
You
know.
E
C
So
I
think
we
can
probably
shutter
that
one
a
discussion
about
whether
we
want
to
shutter
the
actual
main
senpai
google
groups,
making
this
or
not
and
replace
it
with
discourse
is,
I
don't
know
I
mean
we
should
have.
We
should
talk
about
that.
I
think
what
are
the
two
I
think
roles
that
they've
before
so
you
can
use
discourse
like
a
mailing
list.
You
can
configure
discourse
to
send
you
emails
and
reply
by
email.
You
don't
actually
have
to
log
into
the
website
after
you've
set
up
an
account
at
all.
C
If
you
want
to
use
discourse
like
a
mailing
list,
you
can-
and
there
are
even
scripts
that
will
migrate.
Google
groups
stuff
to
discourse
like
encourage
people
to
set
up
a
discourse
account
and
stuff.
So
it's
definitely
something
we
could
do.
My
opinion
is
that
we
probably
need
to
put
a
bit
more
effort
into
discourse
in
its
current
state
before
we
do
that
and
like
get
all
of
us
happier
moderating
it
and
running
it,
and
then
we
can
have
that
like
decide
when
it's
worth
doing
that
or
if
it
is
at
all
all
right.
G
C
G
F
E
F
F
I
don't
know
because
I
think
the
discord
could
also
be
informed
for
discussion.
That's
not
necessarily
made
in
this
thing,
or
it
could
be
a
showcase.
It
could
be.
Like
oh
cool,
look.
I
made
this
little
plot.
Okay,
right,
where
you
wouldn't
maybe
mail
a
whole
mailing
list
that
type
of
thing
it
could
be
a
bit
more
informal.
I
guess,
though,.
G
B
A
C
Like
one,
we
should
be
posting
release,
announcements
on
discourse
as
well,
and
we
haven't
been
two.
I
think
you
can
configure
discourse
to
email
you
if
there's
a
post
in
the
sample
announcements
channel,
for
instance,
now
that
is,
that
requires
clicks
and
stuff
that
people
might
not
necessarily
do
so,
while
in
theory,
discourse
can
do
that
whether
practically
actually
will
fulfill
the
same
role
or
not.
I
don't
know.
H
H
D
H
D
H
E
E
A
E
C
G
C
C
I
think
the
trick
is
you
need
to
know
what
the
email
address
is
you
need
to
email?
I
don't
know
I
yeah,
so
I
set
up
discourse,
but
I
am
far
from
an
expert
in
how
to
use
it
or
like
what
to
do
with
it.
So
I
know
that
it
has
these
features,
but
I
do
not
know
how
easy
they
are
to
use
and
how
practical
it
is
to
replace
the
mailing
list
with
them.
I'm
just
saying
that,
like
discourse
has
the
op
ability
to
be
used
entirely
by
email
now,
then
that's
that's
all.
C
F
F
Exactly
right,
yeah,
so
I
guess
yeah
I'll
just
run
through
this
that
we're
going
to
open
so
like
I
guess
what
have
we
done
so
far,
and
I've
done
here
like
the
orange
kind
of
text
of
maybe
issues
that
what
we've
done
or
things
you
can
improve.
So
you
know
what
many
main
properties
over
the
last
last
number
of
years,
we've
done
kind
of
like
some
point
tutorial
on
whether
this
is
like
introduction
to
some
pi.
You
know
what
is
senpai
data
containers
data
downloads
for
it's.
What
can
you
do
so?
F
The
issues?
I
guess
is
that
like
it's,
not
that
we're
at
every
conference
and
it's
somewhat
hot,
it
can
be
a
last-minute
thing,
oh
god,
I'm
also
going
to
this
conference.
Maybe
we
should
do
some
type
thing
and
then
it's
in
the
evening
and
we're
not
directly,
you
know,
haven't
planned
it
straight
ahead
and
that's
something
I
think
we
can
definitely
build
on
in
the
future
and
kind
of
throw
the
momentum
and-
and
I
think
what
we've
done
in
the
past
again
again
is
working
directly
with
people
a
minute
ago.
It's
either.
F
You
know,
if
you're,
in
a
lab,
that
I'll
use
a
senpai
and
you're
sitting
in
an
office
you're
more
likely
to
use
it,
and-
and
this
is
issues
with
again-
accessibility
like
if
you
don't
have
someone
in
your
group
using
senpai
or
you're-
not,
I
don't
know
you're
working
in
a
different
country
where
nobody
is
a
developer.
It
might
be
kind
of
hard.
So
this
is
maybe
an
issue,
and
how
can
we
kind
of
use
the
fact
that
people,
it
works
well
working
with
people,
and
how
can
we
grow
that
and
that's
really
visibility?
F
F
Is
this
something
that
we
need
to
put
forward
on
the
website
like
go
and
ask
on
the
mailing
list,
and-
and
I
think
something
we
did
this
year,
which
was
at
the
test
meeting
where
it
wasn't
just
the
sun
tutorial
again,
it
was
like
a
community
showcasing
forum
and
it
was
in
the
evening
and
we
had
people
come
and
like
put
a
slide
up
of
like
what
they
had
done
with
something.
I
think
that
was
really
cool,
because
it
really
shows
all
the
weird
wonderful
things.
F
People
have
used
that
weird,
fantastic
and
wonderful
things
that
and
people
have
used
input
for
it
and
I
think,
going
forward.
I
think
that
would
be
really
nice
to
do,
and
also
have
a
community
forum
like
what
just
a
very
informal
job
of
like
what
work
stuff
doesn't
work
and,
and
things
like
that,
just
to
get
a
feel
for
the
room,
and
I
think
we
did
get
some
kind
of
good
stuff
out
of
there
and
I
think
to
improve
this
and
it
was
late
on
the
day
after
an
afternoon
off.
F
F
Again,
it
goes
back
to
the
first
point
of
like
organize
it
sooner
and
I
think
there
was
a
huge
disconnect
in
the
room
of
like
there
was
people
there
who
never
used
some
fight.
Who
were
there
interested
to
learn,
and
then
there
were
people
who
were
advanced
users
who
wanted
to
learn
this
specific
thing
about
the
coordinate
stuff.
So
maybe
we
should
have
an
idea
of
having
like
a
break
it
up
if
you're,
a
new
user.
F
Somebody
come
here
or
go
there
or
have
kind
of
time
today
with
different
people,
and
this
goes
on
to
like
perspective
things
right.
So
I
think
going
forward.
I
think,
would
be
really
good
to
have
sorry
again,
I'm
just
talking
anyway,
but
these
are
just
some
ideas.
Maybe
we
can
talk
a
bit,
and
this
is
like
the
last
slide
one
more
after
this
and
they
kind
of
go
forward
with
this
forum
style
at
conferences.
So
like
have
structured
feedback
discussion,
I
think
we
didn't
really
have
enough
time.
F
I
think
at
the
end,
for
the
amount
of
feedback
we
wanted,
and
maybe
we
should
have
more
specific
things.
We
want
feedback
for
and
and
again
like
plan
ahead
for
the
calendar
year
like
we
know
next
year.
What
meetings
are
going
to
be
on
or
that's
starting
to
trickle
in
now
plan
ahead
who's
going
to
what
again,
this
visibility
thing
make
sure
there's
somebody
at
each
gets
onto
organizers.
F
Now
then
like
can
we
have
an
afternoon
or
a
morning
or
a
whole
session
of
the
time
and
and
again
have
dedicated
showcase
time
so
have
a
time
where
the
community
to
show
what
they've
done,
whether
it
be
for
a
paper
or
just
difficult
plot
their
ways,
and
I
think
that's
really
kind
of
a
nice
idea
or
something
they
want
to
do
as
well
like.
I
would
love
to
do
this.
F
How
do
you
do
it
and
kind
of
see
what
community
needs
are,
and
I
think
some
things
we
did
this
year,
which
was
kind
of
cool,
was
go
to
some
of
the
summer
schools.
So
it's
a
pipe
c
and
the
christmas
skill
also
at
the
solar
orbiter
summit.
So
we
had
a
a
tutorial.
So,
maybe
again,
we
should
identify
the
summer
skills
for
next
year,
like
the
sdfc
one
is
a
great
one
for
first-year
phd
students.
I
make
sure
that
there's
a
specific
supply
tutorial
to
to
again
visibility.
F
A
A
G
F
F
F
G
So
then,
how?
How
do
we
do
that?
Because
normally
we've
been
focused
on,
like
oh
we'll
get
sort
of
users
and
like
lower
level
users,
but
now
we're
talking
about
people
who
sway
in
the
community
so
like?
How
do
we
sort
of
access
them
and
make
them
think
of
us
in
the
same
kind
of
moment
as
they
think
about
the
conference.
B
F
Know,
like
you
know,
it's
like
it'll
be
kind
of
more
common
than,
and
people
will
expect
like
the
sunflower
thing.
When
is
that,
honestly,
you
know
it.
A
F
Yeah
yeah
exactly
or
people
will
know
then
or
maybe
go
off
the
show
or
yeah.
I
think
that'd
be
really
great
and
maybe
I'll
just
look
like.
I
see
a
few
things
here,
just
twitter.
I
think
there
is
a
really
good
twitter
community
online
for
like.
A
F
A
A
F
F
Yeah
yeah
absolutely
well.
F
E
A
E
A
F
E
G
F
F
F
F
For
sure,
so
I
was
thinking
like
like
one
like
I'm
happy
to
do
that
as
like
like
to
to
reach
out
and
then
like,
be
the
point
of
contact
and
then
reject
everyone
in
the
community
to
be
like.
I
mean
this
community
or
developers
like
who's
going
to
which
fill
it
in.
If
there's
nobody
going
to
find
ways
to
fund
somebody
to
go,
you
know
things
like
that.
I
think
we're
really
good
and
his
last
slide
is
kind
of
like
action
items
as
well,
and
this
is
kind
of
something
that
we're
saying
like
branding.
F
I
think
this
is
when
he
said
yesterday
when
something
has
been
talking
about
today.
I
guess
it's
like
just
visibility,
because
it
really
is
personal
connection,
which
I
think
making
a
huge
difference
in
this,
and
I
think
again,
going
back
to
discourse-
and
I
think,
usually
yesterday
were
like
it
kind
of-
is
like
a
safe
space
to
ask
questions
where
it's
like
less
less
formal
than
an
email
thread.
Yeah.
F
D
Yeah
I'll
say
that
just
about
the
conferences,
something
that
would
be
it's
very
obvious
to
everyone.
I
assume,
but
just
I'll
just
say
it
just
in
case
it
isn't,
is
that
yeah
instrument,
specific
conferences.
This
is
something
I've
seen
before
at
them,
but
the
analysis
of
this
instrument
with
some
pack.
G
G
No,
I
should
say
with
the
solar
over
the
workshop
right.
The
last
day
is
slated
not
specifically
sunpi,
but
like
yeah
software
for
all
the
different
like
instruments,
and
I
think
they're
lively.
F
A
A
Contact
us-
and
I
I
do-
I
do
think
that
if
you
go
to
the
the
openness
or
if
you
follow
the
link
from
our
from
the
docs
to
get
the
open
astronomy
page,
maybe
just
change
that
link.
So
it
goes
to
the
senpai
section
just
so
that
it's
created,
because
at
the
moment
I
think
there
is
a
bit
of
you
click
on.
Oh,
if
you
click
on
the
link,
but
what's
open
to
show
me,
I
want
to
get
somewhere.
A
A
A
F
G
C
D
F
A
I
don't
know
what
chat
is
and
what
form
where
the
form
is,
and
so
I
think,
just
removing
all
the
inertia
like
the
resistance
that
you
any
kind
of
problem
that
you
have
to
go
from
one
side
to
another
site,
yeah,
make
it
as
transparent
as
possible,
because
once
you
have
the
community
like
aware
that
you
have
all
these
tools
available
and
you're
doing
all
these
things
in
the
summer,
schools.
F
A
They
need
a
way
to
like
asynchronously,
then
or
offline.
Go
from
that
knowledge.
G
A
A
A
F
F
D
We
moved
on
I'm
sure
this
has
been
discussed
all
in
the
past,
but
what
about
the
senpai
github
landing
page
being
a
github
pages
so
that
you
can
make
it
so?
The
github
pages,
like
the
github
io,
the
senpai,
you
know
github.com
some
pie
could
be
on
github
pages,
because
when
you,
google
senpai
the
sample
github's
one
of
the
early
like
one
of
the
top
search
engine
results
and
but
when
you
click
on
the
github,
it's
a
little
bit
like
you
know,
github
is
an
intimidating
website.
D
Let's
be
honest,
whereas
then
there's
some,
the
github
io
pages
is
like
just
a
web
page
would
be
the
opening
page
of
your
of
your
repo.
D
C
B
B
I
D
You
can
always
talk
about
that
specifically
like
if
you're
talking
about
these
sessions
at
conferences,
you
can
always
specifically
have
a
talk
about
swapping
from
ssw
to
somebody
from
someone
that
knows
both
to
be
like
hey.
You
know,
this
is
how,
if
you
want
to
do
this
in
ssw,
this
is
how
you
can
do
it
in
some
pi
and
it's
easier
or,
and
it's
better
or
whatever.
Well,
we
would.
E
Yeah
exactly,
I
think
we
often
present
ourselves
as
like
some
all
in
on
senpai
or
nothing
but
like
the
reality
is
that
like?
Yet
there
is
a
lot
of
really
really
useful
stuff
in
ssw,
and
there
are.
There
are
particular
instruments
with
the
the
majority
of
their
analysis.
Codes
like
still
lives
in
ssw
and
like
if
you're,
a
researcher
in
that
area.
Anything
about
like
a
lot
of
the
resin
stuff
right,
like
you
still
need
ssw
to
do
or
or
chianti
is
great.
E
Especially
if
you're
like
well,
I
was
gonna
say
especially
for
ph,
but
no
you
have
lots
of
time
to
just
rewrite
everything
if
you're
a
or
like
early
faculty.
Something
like
you
need
to
get
something
done.
You
need
to
write
a
paper
like.
Maybe
I
want
to
use
the
coordinate
section.
Sometimes
I
need
to
like
prep
data
or
something
like
that
and
in
you
know,
ssw,
and
I
don't
like.
We
should
not
present-
that
as
like
being
or
yeah
yeah.
It
should
be
at
both
and
like
how
do
we
bring
the
two
together
yeah.
F
E
G
E
G
F
I
guess
that
maybe
the
point
is
more
like
if
you're
an
ssw
user,
you
don't
know
about
some
fight,
you
don't
know
like
what
does
somebody
do
like
how
do
they
fit
together,
like
what
can
I
you
know
like?
Can
I
read
a
fitz
file
with
somebody
like
people?
Don't
if
they're
coming
from
sw
completely
naive
to
fighting
their
senpai?
It's
like
you
almost
need
a
page
of
like
some
type
of
ssw
users
and
they
have
a
description
of
sw
like
what
what's
the
kind
of
we
haven't.
B
G
E
I
think
I'm
making
less
less
a
much
less
technological
technological
argument
than
that.
It's
more
just
like
a
cultural
one
like
to
to
make
use
of
the
ecosystem
you
don't
have
to
like
buy
into
it
completely
like
it's,
not
it's
not
like
a
one
day.
It's
like
oh
I've,
chucked
ssw
in
the
trash,
and
I
can
only
use
python
now.
I'm.
A
G
G
G
F
F
I
noticed
a
code
quite
simple,
but
she
said
she
was
really
confused
of
like
how
does
the
arc
of
supply
work
compared
to
my
assistant
and
like
do
all
those
things
and
she
was
like
you
just
wanted
a
visual
or
some
description
of
how
these
things
work,
which
would
be
something
really
tailored
for
three
specific
need
of
ssw
cell
users
and
how
it
works
out.
I.
C
F
F
A
A
F
E
F
A
Maybe
if,
by
going
to
these
summer
schools
we
start
off
early,
then
the
supervisor
might
say
yeah.
You
can
do
this
in
blab,
but
the
students
might
be
aware
of
kaizen
and
somebody
I
might
come
and
ask
a
question
be
like.
Can
I
do
this?
I
need
to
say
actually
no,
no
or
yes,
you
can
here's
what
you
do.
Yeah.
D
Anyone's
decision
here,
but
things
like
the
I'm,
I'm
sure
there
was
something
this
is
going
back
a
long
time
now
to
when
I
did
summer
schools,
I'm
sure
there
must
have
been
something
to
do
with
some
pie,
but
I'm
not
actually
it
almost
feels
like
there
should
be
right
like
there
should
be
a
dedicated
one
hour.
You
know
that
you
know
you
got
an
hour
on
this
an
hour
on
that
an
hour
on.
You
know
there
should
be
an
hour
on
senpai,
but
that's
not
anyone's
decision
here
to
make.
I
A
F
D
G
Like
you
know,
senpai
ai
pi,
you
know
potentially
nd
cube,
astrophy,
numpy,
lib
and
you're.
Like
saying
this
is
the
ecosystem
from
that
level,
you
say:
if
you
want
to
do
scientific
analysis,
you
don't
install
something.
That's
it.
You
know
this
is
effectively
your
ssw
equivalent,
whether
you're,
depending
on
the
audience
you
may
make
that
connection
or
not
but
effectively.
That's
what
you're
saying
like
the
python
ecosystem
right
exactly
right.
F
Yeah,
absolutely,
and
even
things
like
the
concept
of
like
senpai,
builds
properly
prep
data
like
it
takes
the
data,
and
maybe
you
know
you
do
your
data
prep
in
a
specific
affiliated
package
yeah
and
it's
supposed
to
work
with
higher
level
data.
Would
you
all
kind
of
know,
but
it's
not
really?
Okay,.
F
A
A
F
A
I
think
the
huge
benefit
of
having
a
survey
that
has
at
least
a
set
of
questions
that
are
always
asked
is
that
you
get
to
see
if
the
things
you're
doing
I
have
are
having
an
effect,
maybe
not
see
if
they
might
be
having
an
effect.
Obviously,
at
least
you
get
some
feedback
on.
What's
going
on
other
than
just
worry
about
amazing,
you
can
naturally
simplify
it
and
then
just
have
the
same
questions
here.
F
A
E
G
A
F
E
And
this
is
just
it
goes
back
to
this
branding
seems
to
be
the
theme
of
the
last
two
days,
they're
like
yeah.
You
know
if
we
have
a
picture
that
can
explain
what
this
is.
I
think
this
clarifies
all
of
these
words.
I
think
that
probably
don't
like
it's
core
versus
project
versus
package.
I
think
that
just
you
know
flies
over
a
lot
of
people,
because
it's
like
not
that
it's
like
above
them,
but
it's
just
like
what
does
this
mean?
Why
do
I.
A
F
F
Know,
yeah
yeah,
yeah
good,
you
think,
as
you
said
like,
I,
don't
think
people
go
to
a
website
and
read
a
block
text
where
they're
kind
of
just
trying
to
get
an
idea
of
something,
but
a
visual
would
be
much
better
and
I
was
thinking
so
very
last
thing
like
something
like
not.
This
is
just
a
slider
pulled
but
like
on
the
page
of
how
to
get
involved,
not
a
block
of
text
have
like
feedback
issues
with
pictures.
It's
like
just
to
make
it
a
bit
more
visually
and
friendly.
D
F
F
E
The
one
thing
I
wanted
to
bring
up
is
that
next
year
at
spd,
they
are
doing
a
like.
I
think
it's
a
whole
day,
or
maybe
it's
a
half
day.
That's
like
a
software
workshop,
and
the
way
it's
been
explained
to
me
is
that
they're,
essentially
asking
for
pitches
from
different
groups
or
who
would
run
you
would
get
the
money
to
like
run
this
software
workshop,
and
I
think
you
know
in
in
the
spirit
of
planning
ahead
for
for
the
next
year.
E
I
think
we
need
a
coherent
pitch
for
what
we
would
do
at
that
workshop.
It's
not
just
plot
a
map.
Save
the
map.
Do
a
sub
map
yeah
yeah
yeah,
like
it
needs
to
be
more
about,
like
a,
I
don't
know,
like
sort
of
a
mini
like
how
to
be
a
contributor
to
senpai
workshop,
but
I
think
that
would
be
a
great
thing
to
come
out
of.
E
This
would
be
a
skeleton
for
a
pitch
for
that
next
year,
because
I
imagine
that
you
know
spd
won't,
be
the
only
ones
to
kind
of
use
this
model
yeah
and
can
also
serve
as
like
a
starting
point
for,
like
part
of
a
summer
school
a
day
at
another
school
for
summer
school
or
maybe
a
sunday
summer.
School.
F
A
E
A
B
A
F
C
The
funding
model
for
running
a
summer
school
like
that
is
designed
that
you
get
in
n
lecturers.
Where
n
is
the
number
of
hours
of
lecture
you
want
and
each
lecturer
gets
a
small
amount
of
money
and
they
come
and
they
write
give
a
talk
when
drew
and
I
ran
it,
we
wrote
the
whole
thing
and
it
took
us
a
long
time,
a
lot
of
effort
and
we
got
paid
peanuts
for
it.
C
F
C
F
G
You
know
learn
how
to
do
your
analysis
in
python.
Sunfire
is
a
key
leg
in
that
stool,
but
you
know
if
we
want
people
to
go
away
and
say:
oh,
I
can
do
anything
I
want
now,
like
you
know
like
within
reason,
we're
going
to
have
to
talk
about
astrophy
we're
going
to
have
to
talk
about.
You
know
you
know,
maybe
other
visualizations
or
something
like
that's
that's
going
to
be
the
value
of
having
people
together
for
multiple
days
and.
C
G
Some
of
it
yeah
I
mean,
if
you're,
if
you're,
bringing
some
of
these
ideas
and
you're
not
going
to
immediately
limit
yourself
by
logistics
or
costs,
then
one
of
one
of
the
things
I
think
you'd
have
streams
right,
you'd
have
like
like
in
various
other.
Some
schools
have
been
like.
You
know,
beginner,
you
know
intermediate
advanced,
you
know
when
it
comes
to
science,
why
not
have
something
like
that
again
assuming
enough
people
enough
money,
but
just
brainstorming.
That
would
be
a
good
way
to
go
right.
Yeah.
It's.
C
A
A
F
Yeah
yeah
so
I'll
put
that
together
and
then
we'll
share
it
so
that
we
can
all
add
to
it
or
again.
If
you
see
a
conference
or
something
like,
we
should
talk
about
it
in
the
next
couple,
yeah
or
funding
opportunities
as
well.
Do
you
guys,
I
think,
that's
your
best
bet.
You
start
with
this
like
kind
of
tackling
it
onto
the
meetings,
then
you
build
up
a
template.
You
build
up
your
things,
you
want
to
say,
and
then
you
or
you
almost.
E
E
F
H
I
maybe
I
missed
this,
but
I
want
to
ask
if
anyone
has
experience,
doing
workshops
or
institutions
as
opposed
to
work
like
their
own
institution
or
instead
of
workshops,
because
I
I
never
got
around
to
doing
it,
but
I
was
going
to
do
a
senpai
workshop
at
mssl,
where
I
used
to
be.
I
think
that
would
have
been
really
useful
to
everyone
in
the
group,
not
all
of
whom
used
somehow
but
were
starting
to.
H
G
E
People
that
came
with
specific
questions-
yeah
I've-
I've
purchased
the
albert
a
couple
times,
but
I
think
maybe
I've
talked
to
you
guys
about
it
too,
but
I
would
like
to
start
at
governor
doing
like
some
high
office
hours,
it's
like
even
even.
C
A
H
E
G
G
A
Is
the
same
exactly
so
like
you
don't
have
someone
presenting
a
slightly
different
way
or
calling
something
or
explaining
something
slightly
differently?
If
you
have
attempted,
then
it
makes
things
more
common.
So
then,
when
those
users
talk
to
each
other
at
meetings,
they're,
like
oh
yeah,
I
know
I've
seen
that
thing
and
it
was
exactly
the
same
way.
It.
A
F
G
E
C
A
G
E
Just
perfect
yeah
I
mean
I
think,
when
we
put
this
on
the
schedule.
At
least
my
thinking
was,
I
don't
know
if
you
broke
the
stove
or
if
I
did,
but
that
I
have
just
and
laura
mentioned
this,
but
it
just
seems
like
we
have
an
increasingly
larger
number
of
ways
to
communicate
with
people
about
senpai,
whether
it's
the
mailing
list,
the
discourse
helio
knots,
l,
like
our
element,
channel
the
hc
element
channel
the
slack
paichy
slack.
E
Like
I
mean
I
think,
I'm
making
my
point
that,
like
our
channels
of
community
communication,
are
becoming
fractured
and
are
we
are
we
paying
attention
to
the
right
ones
and
giving
the
right
ones
the
the
amount
of
attention
amount
of
attention
that
they
need
to
relative
to
what
the
the
community
is
using
or
paying
attention
to?
I
think
that's
just
kind
of
the
question
I
wanted
to
pose.
B
F
E
C
E
E
I
don't
don't
know
that.
Well,
I
mean
yes,
I
have,
I
think,
made
that
point
before
I
don't
know
if
it's
him,
but
at
the
same
time,
if
that
is
a
platform
that
has
growing
momentum
in
the
community,
who
are
we
to
say,
you
know
screw
screw
your
platform
work?
This
is
where
we
communicate
like.
I
don't
think,
that's
the
right.
I
don't
think
that's
what
we
should
be
doing.
I.
H
Think
there
is
a
place
to
push
back
and
say
if
you
want
support
with
senpai.
This
is
where
we
give
it
and
we
give
it
here
because
it's
publicly
accessible
and
searchable,
and
so
we
don't
like-
and
there
are
good
reasons
for
that,
one
of
which
is
not
duplicating
our
efforts.
So
if
someone
has
the
same
issue
like
that,
will
come
up
on
a
google
search
and
I
think
we
should,
I
think
yeah.
I
appreciate
hearing.
What's
has
momentum
but
beyond
directing
people
from
healing
to
a
like
public
forum,
yeah.
C
E
Because
but
yeah,
I
think
the
last
thing
you
said
I
think
is
is
a
good
one
that
it's
we
are
still
like.
Those
communities
are
still
overlapping.
That
doesn't
mean
that
we
have
to
like
adopt
it
as
a
communication
channel.
That's
not
what
I'm
saying,
but
what
what
I
am
saying
is
that
I
don't
think
we
should
just
and
like
if
someone
asks
a
son
by
a
question,
we
don't
just
jump
in
there
and
be
like
oh,
if
you
want
help
go
to
like
this
good,
like
I'm.
B
A
H
C
A
A
E
A
E
E
I
mean,
I
think,
like
I
agree
with
all
of
your
points.
I
just
think
we
should
be
careful
about
the
way
that
we
phrase
that,
because
again,
I
think
there
is
like
this
idea
within
the
larger
solar
physics
community
that
they're
like
this.
This
group
has
like
anotherness
to
it
and
that
and
I
don't
and
I
don't
necessarily
think
that's
earned
sorry
you
mean
senpai,
has
anotherness
yeah
right
yeah
and
the
people
that
like
use,
senpai
and
develop
somebody
there's
just
this,
like
I
don't
know,
not
mystique,
but
you
know.
E
And
I
think
we
need
to
do
everything
we
can
to
to
not
give
people
that
impression.
But
so
I
don't
I
don't
disagree
with.
Like
anything,
that's
really
been
said,
I
think
it's
more
just
like
the
way
in
which
we
say
it
and
we
present
that
to
the
community.
F
A
F
C
C
A
E
There
is,
you
know
there
are
problems
with
it,
certainly,
but
I
think
it
has
become
like
a
really
informal
space
to
ask
questions
for
people
all
the
way
from
like
undergraduate
students
to
like
really
senior
people
and-
and
I
think
that
you
know,
and
that
kind
of
like
vulnerability,
that's
displayed
on
that
platform
is
extremely
valuable
and
it's
not
duplicated
as
any
other.
But
well.
I
think
I
think
that
does
happen
in
our
in
our
chat
room,
but
I'm
not
sure,
like
every
everyone
feels
they
have
access.
E
And
yeah,
I
like
I've,
answered
several
questions
on
there
and
I
think
stuart
has
and
others
I
don't
know
I
I'm
not
trying
to
especially
you
know.
I
realize
that
you
know
I'm
one
of
the
few
people.
That's
also
being
paid
to
do
this
and
I'm
at
you
know
a
very
large
u.s
centric
institution
and
the
one
that
created
this
possibility.
Yeah.
So
I
have,
you
know,
probably
a
really
implicit
bias
when
I'm
talking
about
this,
so
I
don't
want
to
make
it
seem
like
I'm
telling
people
to
pay
attention
to
this
platform.
E
E
H
I
wonder
if
we
need
to
like
actually
advertise
to
this
our
discourse
a
bit
more
and
like
definitely
do
something
yeah,
because
I
guess
I
get
the
impression
that
he,
you
know,
has
kind
of
appeared
to
fulfill
a
vacuum,
because
people
wanted
that
platform
and
somehow
didn't
have
a
similar
platform.
Until
we
were
on
discourse
yeah,
I
think
yeah,
I
think
yeah
doing
a
bit
more
advertising
of
like
these
are
the
the
some
part
you'll
get
better
helps.
Yes,.
F
A
A
C
A
C
A
G
Jump
on
your
side
of
the
argument
for
a
moment,
it's
like
it
seems
to
me
that
the
the
role
that
healing
notes
is
playing
especially
since
is
that
you
know
you're
a
researcher
you're
doing
research
you're,
not
like
a
developer
you're,
not
like
specifically
trying
to
know
about
sunflower
you're
trying
to
like
solve
a
problem,
and
so
what
you
used
to
do
is
you
turn
to
the
person
next
to
your
office
and
ask
like
how
to
do
something
right.
So
it
seems
like
healing,
is
creating
like
a
forum
for
solar
physics.
C
F
F
F
A
F
F
E
E
F
E
E
F
A
F
F
G
F
A
C
B
C
C
B
Oh
sorry,
right,
oh
yeah,
sorry
about
it!
Well,
maybe
is
it
time
to
run
with
dms.
C
B
C
C
C
B
F
C
A
A
A
C
B
A
C
This
is
I
had.
I
had
real
fun
making
this
graph.
I
don't
I
don't
like
matpot
love
anymore,
but
like
this
is
what
this
orange
here
is.
The
development
channel,
which
orange
this.
A
F
A
G
Apart
from
old
steward
and
albert,
I
talk
a
lot.
D
C
E
C
E
E
C
A
C
B
D
G
So
we
should
call
this
in
at
a
late
notice,
but
it's
something
I'm
interested
in
discussing
it's
something.
That's
come
up
over
the
years
with
various
degrees
of
frustration,
annoyance
and
inspiration.
G
Yeah,
so
what
I
would
like
to
get
out
of
this
is
to
show
you
an
example
of
a
sliceable
metadata
object
that
has
already
been
implemented.
It's
in
sunraster.
The
use
case
was
for
spice,
but
I
think
it's
a
lot
more.
G
I
think
it
could
easily
be
generalized
to
a
greater
extent
and
could
work
with
ndq
or
data
objects
generally,
so
I'd
like
to
sort
of
show
that
and
get
some
feedback
on
that
and
an
ambitious
goal
would
be
to
get
the
like
in
the
notes,
get
a
preliminary
kind
of
list
of
like
metadata
object
requirements
to
say,
like
you
know,
along
these
lines
and
also
to
hear
if
people
think
that
this
is
completely
unreasonable.
For
you
know
so
I'll
just
show
this
a
brief
notebook.
G
If
there
are
any
questions
or
clarifications,
please
jump
in.
I
think,
unusually,
though,
I
think
I'd
like
to
hold
discussion
until
the
end,
the
notebook
isn't
very
long,
and
hopefully
we
can
sort
of
get
a
full
picture
of
what
I've
got
to
show
here
and
then
we
can
like
tear
it
apart,
but
there
are
questions
verifications
that
help
understanding.
Please
do
completely
interrupt
any
time,
okay.
So
what
I've
done
here
is
I've
just
sort
of
generated
a
fix
here.
G
You
know
just
sort
of
locally
for
convenience
for
this
demonstration
called
spice
header.
So
this
is
just
a
fits.
A
fits
header
object.
So
this
is,
you
know
typical,
for
when
we
have
fixed
files
we
you
know,
I
think
most
of
us
are
aware
that
this
is
often
how
a
metadata
is
stored,
so
this
spice
meta
object
which,
as
I
said,
was
developed
for
for
the
spice
and
solar
orbital.
G
The
way
to
initialize
that
is
just
you
know,
give
the
object,
the
header
information
and
this
object-
I've
actually
subclassed
from
dictionaries.
So
if
you
want
to
then
see
the
metadata
tags
again,
you
can
just
use
the
dictionary
methods
right
so
keys.
You
can
see
all
the
different
names
and
then,
if
you
want
to
access
a
value,
then
you
know
you,
just
as
you
would
with
a
dictionary.
You
just
stick
in
the
balance
key.
However,
with
fits
headers,
we
also
often
have
comments
so
to
facilitate
that.
G
If
you
want
to
include
comments,
what
you
do
is
you
can
create
a
dictionary
of
comments
and
if
you
look
at
that,
the
keys
are
the
same
as
the
metadata,
but
instead
of
it
being
the
value.
It's
it's
the
comment,
and
then
you
can
include
that
when
you
instantiate
the
metadata
object
by
just
saying
comments
is
equal
to
that
comments
dictionary.
So
that
means
that
you
have
a
dot
comments
attribute
and
if
you
put
in
the
right
key,
then
you
get
the
comment
for
that
piece
of
metadata.
G
So
now
I
just
want
to
discuss
like
key
access
versus
attribute
access,
so
what
I
envisioned
to
what
I
think
would
be
good
is
if
we
could
define
a
hierarchy
of,
say,
abstract,
based
classes
or
meta
classes
anyway,
some
some
infrastructure
that
defines
a
standardized
set
of
names
for
common
pieces
of
metadata
and
then
what
we
can
do
is
we
can
combine
these
in
different
ways
based
on
our
needs.
G
So,
for
example,
we
might
have
like
a
remote
sensing
or
metadata,
that
is
with
remote
sensing
instruments,
and
so
then
they
all
have
a
standardized
name
and
maybe
then,
like
a
spectrograph,
you
know.
Maybe
there
are
things
that
surrounds
that,
so
we
can
mix
these
these
sets
of
standardized
names
and
then
we
access
them
not
just
as
a
key
in
a
dictionary
but
as
an
attribute
on
that
dictionary.
G
So
just
if
we
look
at
that
same
example,
we
look
at
observatory
if
we
access
that
as
a
key
like
a
dictionary,
we
get
solar
order,
but
if
I
do
metadot
observatory
in
this
case,
I
get
the
same
thing
now.
G
The
in
this
case,
there's
no
real
difference,
but
what
I
think
this
does
two
things.
Firstly,
as
I
mentioned,
the
attributes
allow
us
to
define
like
a
status
and
reliable
api
for
like
us
to
access
metadata,
no
matter
like
how
any
instrument
specifically
defines
that
underneath
and
also
so,
if
you're,
you
know
at
two
image
objects
from
two
different
instruments.
This
allows
you
to
like
very
much
like
map
already.
Does
you
know
you
sort
of
access,
the
observatory
or
the
dates
or
the
telescope?
G
G
G
This
does,
since
this
framework
doesn't
require
to
be
fits,
but
in
this
particular
example,
I
have
data
from
projects
in
this
framework,
I'm
not
nothing
dependent
on
being
it's
just.
C
G
Right,
so
actually
that
brings
up
the
exact
next
point
I
want
to
make
so
as
well
as
defining
this
standardized
sort
of
api
specific
metadata
map.
Does
the
attribute
model
allows
you
to
build
those
more
comp
of
those
more
useful,
complex
objects
right.
So
if
we
want
like
solar
b,
if
we
access
that
as
a
key,
it's
just
giving
us
the
value
from
the
the
header
object,
which
is
just
a
number
that
doesn't
really
actually
tell
us
very
much.
But
when
you
implement
the
attribute
or
the
property
you
can
pull.
G
So
if
we
look
up
here
solo
b
in
the
comment
it
gives
you
the
the
unit
and
if
we
know
how
this
instrument
is
defining
that
information,
we
know
how
to
get
it,
and
so
we
can
pull
those
pieces
of
information
together
and
instead
of
now
having
just
a
number
with
you
know
very
little
with
no
context.
We
can
combine
that
with
other
metadata
information
to
give
a
very
like
as
a
sort
of
a
self-contained
or
self-explanatory
and
far
more
useful
object.
G
So
this
is
an
example
where
we've
just
taken
a
number
unit
and
created
a
quantity,
but
you
know
we
could
easily
do
the
same
thing
with
the
time
object.
This
is.
C
G
G
Absolutely
right
so
so
this
brings
me
on
to
this
extra
part
of
the
notebook
which
I'm
suggesting
that
we
create
this
hierarchy
of
classes
that
define
the
standardized
names
and
the
standardizing
should
be
more
intuitive
than
like.
G
You
know,
like
the
keys,
because
the
keys
and
say
in
fits
like
have
certain
restrictions
on
them
that
you
know,
property
name
doesn't
like
to
like
observe
a
tree,
or
you
know,
like
various
other
things
like
that,
there
are
going
to
be
restrictions
or
conventions
that
might
make
sense
for
this,
but
once
we
have
a
data
object,
we
want
to
liberate
ourselves
from
those
things
because
they're
not
necessary
to
us
anymore,
right,
shackles
of
bits,
yeah.
Well,
I
mean.
G
So
in
this
particular
case
I
just
copied,
like
you
know,
solar
underscore
v0s,
but
we
don't
have
to
do
that,
so
I
don't
know
actually
if
this
will
share,
but
okay,
so
in
this
example,
we've
got
this
meta.
This
is
in
sunraster.
G
We
have
this
metadot
pi
file
and
you
can
kind
of
see
how
I'm
proposing
this
works
in
a
very
sort
of
low
level.
So
this
sort
of
general
meta
abc
and
it
defines
the
names
like
standardized
names
for
various
types
of
metadata.
Now
the
meta
abc
is
like
the
most
general
one,
the
solar
physics.
So
you
know,
detector
is
always
going
to
be
relevant,
instrumental
observatories,
processing,
level,
server,
location
date,
reference
time.
G
So
so
here
we
can
define
what
we
think
should
always
be
the
names
and
then,
if
we
want
some
slightly
specific
set
of
metadata.
For
example,
if
we
have
a
remote
sensing
instrument
rather
than
an
in-situ
one,
we
might
have
other
things
like
you
know
the
distance,
the
sun,
that's
probably
more
important
for
remote
sensing
instruments,
maybe
also
titu,
but
you
know
things
like
the
radius
of
the
sun
in
meters
or
or
or
arc
seconds
or
so
on.
G
So
we
can
like
define
things
that
are
relevant
to
each
sort
of
like
increasingly
specific
layer
of
data
type
or
inside
and
then
down
the
bottom.
Here.
I've
sort
of
taken
the
general
meta
and
remote
sensing
meta,
and
I've
inherited
that
and
now
I've
got
like
various
tags
that
are
specific
to
slit
spectrographs.
What
I'm
framing
in
this
example
are
specific
to
slit
spectrographs.
So
in
this
way,.
G
G
G
All
right,
so,
ultimately,
this
particular
implementation
metadata
object,
inherits
from
machine
like
I
support,
so
you
can
always
access.
You
know
you
get
the
metadata
as
keys
and
you
can
override
them
as
you
would
with
a
non-dictionary,
and
you
could
just
notification
but
for
reasons
that
hopefully
will
become
clear,
there's
a
certain
amount
of
housekeeping
that
should
be
done,
and
so
in
order
to
make
sure
that's
done
properly.
I've
included
a
specific
or
special
api
which
is
which
is
recommended
to
use
instead.
G
So
if
you
want
to
add
a
piece
of
metadata,
you
can
use
the
dot
ads
method,
and
here
I'm
just
giving
it
the
name.
My
key,
the
value
which
I
said
is
hello,
a
comment.
This
is
new
metadata
and
for
the
moment,
I'm
going
to
ignore
that
last
input.
I'll
come
back
to
that.
So
just
to
show
how
that
works,
you
know
now
I
can
access
it
as
just
as
a
as
a
key,
and
the
comment
that
I've
been
putting
in
here
for
me
is
also
available.
G
If
I
want
to
amend
it,
I
can
use
the
same
api,
but
I
have
to
set
this
override
equals
true.
I
mean
that
can
be
changed
as
a
default
or
whatever,
but
you
know
it
seemed
like
a
decent
protection.
So
if
that
means
now,
I
can
change
it
to
hello,
world
and
whatnot
so
and
again,
under
the
reason
for
this
is
underneath
under
the
hood
there's
a
certain
amount
of
housekeeping
going
on,
which
is
why
it's
preferred
and
then
removing
metadata.
H
G
G
At
the
moment
no,
but
I
would
say
that's
that's
I
haven't
that's
a
topic
we
can
discuss
and
how
yeah
that's
not.
No,
it's
a.
G
Is
at
least
by
default?
No,
because
my
thinking
was
that
the
the
attribute
should
be
standardized
degrees
names
and
you
just
start
like
adding
whatever
the
key
is,
if
there's
no
control
over
the
key
being
standardized,
so
my
initial
feeling
would
be
no,
but
I
think
that's
something
that
we
can
do
with
this
one.
Okay,
so
then,
coming
on
to
part
of
laura's
question
you,
you
can
restore
the
original
metadata
because
there
is
a
copy
of
the
original
header
available
at
dot
original
header,
and
this
has
never
changed.
G
G
A
G
Quietly
possible,
another
time,
if
I
type
that
very
critical
of
me
apologize
sorry
so
yeah
was
was
there
another
question.
G
All
right,
good
yeah,
so
going
back
to
this,
adding
again
addressing
laura's
question
in
this
case,
like
at
the
the
main
level
by
adding
and
amending
and
removing.
We
are
actually
changing
like
the
values
of
the
metadata,
but
there
is
a
copy
of
the
original
header
at
original.header
and
that
never
changes.
So
you
can
always
make
restore
or
reference
that
so
just
to
show
that
in
in
action
we
do
metadot,
remove
we
remove
date
end
and
now
there
should
be
a
key
here.
That's
gone,
but
it's
still
available
in
here.
G
I'm
showing
the
comments
here
but
like
nonetheless,
it
shows
that
the
key
and
the
information
exists
in
the
original
header,
and
so
I
can
just
add
that
back
in
you
know,
by
giving
that
the
key
and
the
value
and
the
comment,
and
now
it's
back
so
the
original
headed
metadata
is
always
available
in
its
original
form,
which
is
like,
in
this
case
a
fits
file
and
that
doesn't
change,
but
the
top
level
metadata.
G
You
can
change
so
now
to
come
on
to
the
sliceable
part,
and
this
was
one
of
the
main
motivators
for
what
I
was
trying
to
do,
and
it
comes
on
david's
question
about
an
example
of
the
hockey
thing.
G
I
think
the
best
example
is
exposure
time.
So,
if
you
have
a
time
axis,
every
image
along
the
time
axis
doesn't
necessarily
have
the
same
exposure
time.
It
doesn't
vary
like
like
a
coordinate.
Does
so
it's
not.
You
can't
describe
it
with
a
wcs.
It
is
metadata,
it's
arbitrary,
but
it's
not.
You,
don't
have
one
value,
one
exposure
value,
but
every
single
image,
so
you
want
to
record
what's
the
exposure
value
for
each
point
along
its
given
axis
with
a
simple
dictionary,
you
can
do
that
because
you
can
sort
of
put
in
an
8.
G
But
then,
if
you
slice
the
objects
now
you've
got
to
manually
like
fix
that-
and
you
know
instantly
these
two
things
come
out
of
step
and
stop
making
sense.
So
the
ultimate
aim
of
this
was
to
try
and
create
an
object
that
could
be
sliced
in
the
same
way
that
its
corresponding
data
objects
would
be
sliced,
and
this
metadata
this
active,
dependent
metadata
and
the
data
itself
could
be
kept
instead.
G
So
if
we
want
to
add
access
dependent
metadata,
the
first
thing
we
need
to
do
is
when
we
instantiate
a
meta
object.
We
have
to
define
the
shape
of
the
data
with
which
it's
associated,
so
here
at
the
end,
you
can
see
I'm
giving
it
a
tuple.
You
know
three
comma
four,
so
we
have
a
2d
data
array
with
you
know
y
equals
the
y
length
three
and
x
of
length
four.
G
So
I'll
just
do
that
and
you
see
that
if
we
can
see
what
the
shape
is
meta
dot
shape
so
now
carrying
on
an
example
of
exposure
time,
I'm
just
going
to
create
a
length
three
quantity
with
unit
of
seconds,
and
this
gives
me
exposure
time
for
for
each
image.
G
This
last
argument
is
the
axis
or
axis
to
which
the
the
metadata
that
you're
adding
corresponds
to,
if
you
add
it
as
none,
then
it's
not
access
dependent,
it's
like
any
kind
of
normal
string
or
something
that's
just
general
to
the
whole
cube.
But
if
you
want
to
add
something
like
exposure
time
in
this
case,
you
could
say
this
is
related
to
the
zero
data
axis,
and
in
this
case
then,
the
value
that
you
give
in
has
to
have
the
same
four
shape
and
the
data
as
the
axes
that
you
were
assigned
to.
G
F
G
G
Okay,
so
coming
to
the
very
according
to
the
end
of
this
demonstration.
So
the
reason
for
this
access
dependent
stuff
is,
as
I've
mentioned
before,
you
could
leave
to
slice
the
metadata
object.
So
say
we
want
to
a
slice
version
of
the
of
the
metadata
object.
What
I
do
is
just
apply
the
the
null
python
selection
api.
G
G
So
in
this
case
the
data
array
is
is,
has
a
there's
two
dimensions
and
you
know
with
length
three
and
length
four
and
so
that,
while
you
see
you
know
this
license
mentions-
and
you
know
the
indices
are
within
that
range.
So,
let's
slice
it
using
that
object.
And
now
we
look
at
exposure
time
because
the
zero
axis
we've
taken
the
first
two
elements
out
of
the
the
zeroth
dimension.
We
now
see
that
the
exposure
time
quantity
has
been
spliced.
G
So
just
those
first
two
elements
exist
and
if
we
also
look
at
shape
the
shape,
as
the
shape
of
the
data
has
now
been
updated
to
mimic
or
to
match
this
slice
item.
So
this
basically
means
it
has
an
array
and
we
apply
slice
item
to
it
and
with
the
exact
same
slice,
I
apply
it
to
the
metadata
and
it
just
sort
of
figures
out.
You
know
what
the
actual
permitted
data,
which
actually
do
they
correspond
to
slice
bits
like
slice
items
they
they
correspond
to
and
it
slices
it
as
you
would
expect.
G
G
So
you
could
do
that
by
attribute
or
again
by
a
dictionary
key
like
that.
So
that
is
that's
the
conception.
That's
the
idea,
I
think,
opening
the
discussion
at
this
point
or
any
other
comments.
I'm
sure
people
have
some
would
be
good.
H
G
G
Because
there's
got
to
be
some
sort
of
like
you,
can't
just
sort
of
go
back
and
forth
right.
There's
got
to
be
some
unique
relationship
between
your
pixel
and
the
world's
dimensions.
This
type
of
metadata
doesn't
have
that
yeah
right.
So
so,
in
this
case,
it's
not
according
you
can't
use,
coordinate
like
infrastructure.
You
need
access
to
where
metadata.
G
G
If
you
then
want
to
like
sub
map
or
extract
the
region
of
interest,
you
don't
want
to
have
to
then
manually
go
and,
like
you
know,
alter
that
metadata
to
make
sure
everything's,
consistent
and
you'd,
rather
that
all
that
okay
now
stale,
like
it,
doesn't
apply
to
anything
anymore
right
and
all
the
information
you
need
to
keep
that
consistent
is
actually
available,
and
so
it's
just
that,
like
you,
don't
need
to
do
it
like
human
manually.
All
the
information
available.
G
You
need
to
write
a
descriptor,
so
I
think
I
think
that's
that's
the
first
problem
I
was
trying
to
solve
for
this
access
dependence
cycle
metadata
and
the
other
one
was
again
and
trying
to
define
us
the
standard
api
metadata
which
is
already
kind
of
what
that
does
so
just
generalizing
that
away
from
maps
so
that
we
could
extend
this
to
to
other
data
objects
that
will
be
coming
online
with
such
as
ndq.
But
you
know,
spectrum,
radio,
spectra
or
whatever
you
know.
I.
A
H
G
See
the
time
series
metadata
just
keeps
those
two
copies
and
says:
oh,
these
two
yeah,
they
kind
of
apply
to
this
time
series
yeah,
but
you
can't
then
go
like
you
know,
exposure
and
give
me
an
array
right,
they're
separate.
So
what
this
has
allowed
you
to
do
is
when
you've
been
captioning,
you
don't
just
say
we
have
two
dictionaries
that
in
general
apply
to
this
period
of
time.
Yeah,
you
can
sort
of
say
right
for
each
time
step.
G
C
G
C
C
G
C
C
The
best
metadata
is
python
objects,
yeah
right
once
you've
when,
when
you're
once
you've
loaded
your
metadata,
don't
really
want
to
know
or
have
to
ever
think
about
fit
keys
ever
again
for
things
once
you've
made
your
standardized
tribute
names
right,
which
gives
you
back
things
like
observer
equipment,
you
don't
ever
want
to
have
to
think
about
which
basically,
they
came
from
and
always
make
sure
that
the
you
only
have
you
only
also.
You
only
want
to
keep
metadata
in
one
place
at
once.
Right
at
the
moment,
what
matters
is
provides
a
dynamic
view.
C
G
C
I
would
argue
that
we
want
to
move
away
from
that
pattern
and
we
want
to
do
things
on
the
fly
yeah.
We
want
to
take
a
header
pass,
all
the
information
understand
out
ahead
and
delete
them
and
then
also
implement
the
reverse
right
so
in
because
that
gives
you
a
workflow
where
you
load,
a
header,
have
now
have
a
bunch
of
hyphen
objects
that
contain
a
bunch
of
metadata
and
you
can
work
with
them.
It
allows
you
to
set
them.
C
So
if
you
wanted
to
change
the
observer,
coordinate,
you
do
observe
a
coordinate
equals
and
give
it
a
skype
board,
and
it's
like
cool
skype,
had
a
skype
call
before
forgot.
Now
it's
good
right
and
if
then,
if
you
want
to
save
that
to
an
file,
for
instance,
you
would
just
serialize
the
python
object.
So
the
answer
part
you
have
to
go
back
to
it.
You
wouldn't
have
to
serialize
the
fit
header
to
an
active
file,
but
if
you
want
to
serialize
it
to
a
fit
header,
then
it
also
knows
how
to
synthesize
itself.
E
But
if
you
have
updates
say
say
you
need
to
update
the
pointing
in
your
image
in
your
in
your
pitcher,
but
based
on
that
those
20
updates
are
like
expressed
as
updates
to
those
keys
right.
So
then,
how
do
I
then
there's
if
you've
erased
this
record
of
like
the
original
bits
that
I
have
translated
this
higher
level
layer,
then
how
do
I
do
that?
Be.
F
C
C
C
So
you
loaded
your
map
right
and
then,
if
you
look
at
that,
the
user
right,
the
point
is
that
the
expectation
is
anything
that's
in
there,
somebody's
going
to
touch
because
it
doesn't
understand
it
right,
yeah
and
then,
if
you
divide
it
by
the
exposure
time,
one
could
argue.
The
correct
thing
to
do
would
be
dropping
around
the
floor
because.
C
What,
if
doing
the
set
of
operations
you
could
do
to
an
app
or
an
ndq
right?
Where
reproject
is
an
excellent
example
right
once
you
reboot,
potentially
where
we
could,
we
could
make
it
a
case
where
the
user
has
to
explicitly
copy
that
over
afterwards,
like
you
know,
you
have
to
take
a
whole
header
and
bring
it
along
to
the
new
object
or
just
document
the
hell
out
of
it
right.
Anything,
that's
in
here.
G
So
I
would,
I
would
argue
that
this
is
a
very
specific
implementation
discussion
and
I
think
what
is
probably
more
useful
in
the
time
we
have
is
to
talk
more
at
like
slightly
higher
level
about
concept
and
api
okay.
So
I
think
I
think,
there's
a
few
requirements
on
that
right
is.
I.
C
Congratulations,
you
gave
me
a
dictionary
right
and
the
other
one
which
I
don't
know
maybe
is
more
controversial,
but
is
that
we
should
be
using
rich
python
objects
as
our
in-memory
representation
of
the
master.
C
G
G
And
any
other
scenario,
so
the
first
thing
I
was
trying
to
solve
to
your
question
was
to
get
to
which
was
cleaning
up
the
map
api.
So
you
can
take
a
lot
of
that
stuff
off
of
that
dot
and
then
you
have
map.metadot,
and
that
means
that
all
this
is
then
transferable
like
to
other
data
rockets
much
more
easily.
If
you
got
rid
of.
C
If
you,
if
you
had
meta
that,
did
that
right
what
you've
implemented
there
largely
is
map
in
terms
of
lines
of
code
in
your
generic
map
right
yeah,
it's
map,
we
could
basically
get
rid
of
all
the
map
sources
and
then
everything
would
be
generic
map.
G
C
G
G
C
H
G
I
think
that's
I
I
think
that
should
be
discussing
this.
I
don't
mean
maybe,
like
I
don't
know,
I
don't
think
absolutely.
C
C
One
does
not
simply
pass
the
header.
Wcs
is
when
it's
made
by
solar
physicists,
but
my
point.
G
C
Well,
that's
going
back
to
my
point
about
like
in
memory.
Representation
right
is
a
large
form,
and
that
is
wcs.
But
how
would
you
say
you
had
a
fit
header
and
a
right
right
and
you
wanted
to
construct
an
ndq?
Yes,
what?
How
do
you
split
the
fitch
header
into
wcs
and
metadata
for
like
which
is
a
bad
phrasing,
but
you
know
what
I
mean.
F
C
Right
ndq
doesn't
do
that
and
probably
doesn't
want
to.
So
that's
I
don't
know
how
you
you've
got
these
two
things
right
where
you
I
I
can't
figure
out
in
my
head
in
the
last
30
seconds,
like
what
that
api
looks
like
right.
How
do
you
take
a
fit
header
and
construct
an
ndq
output
which
has
wcs
meta
and
a
data
array
like
because
I've
allowed
that
to
customize
the
wcs
passing,
which
is
like
a
really
important
part
of
that.
C
C
G
C
G
A
G
C
It's
not
that's,
not
the
api
that
you,
the
astrophy
wcs
object
knows
where
the
observer
coordinate
is
in
like
sony,
vs,
coordinates
or
whatever
right,
there's
a
limited
selection
of
coolness
that
wcs
actually
understands.
So
the
core
information
is
there,
but
there
is
a
lot
of
so
you
have
to
get
the
keys
directly
out
the
all
structure.
G
C
C
E
H
G
Open
to
the
idea
that
I'm
missing
a
lot
of
this,
but
to
me
it
sounds
like
we've
done
it
this
way,
you
know
with
map
for
various
reasons
and
we're
not
letting
go
of
that
way
of
doing
it.
Like
I
mean
I
don't
see
how
you've
got
a
file,
you
know
you,
you
header,
you
extract
that
information.
You
generate
the
wcs,
you
know
and
alter
the
information
as
required.
Then
you
have
a
wcs
object,
which
is
now
at
wcs.
A
G
C
C
Because
then
you're
standardizing
a
metadata
api
for
the
entirety
of
our
statistics,
like,
I
think
so,
I
think
just
to
quickly
point
out
the
thing
right:
the
wcs
as
a
whole.
You
might
not
consider
metadata,
but
it
is
built
up
of
things.
I
would
consider
metadata
such
as.
G
H
Think
there's
a
thing
or
two
separate
things
that
we're
potentially
creating,
which
is:
what
is
the
storage,
how
what
storage,
architectural
and
inside
the
meta
object
right?
If
you
just
remember
the
map,
you
would
have
meta.wcs
and
that
would
be
the
python
object.
But
then
how
do
you
access
that?
What's
the
api
that
users
access
that
through
the
map
and
through
map,
you
can
have
an
api
guide
accesses
observable
by
doing
something
to
the
wcs?
H
A
G
Also,
it's
like
one.
It's
it's
one
like
piece
of
mediterranean
that
we're
talking
about,
but
like
there's
so
much
most
of
the
meditative
or
something.
C
H
H
C
G
G
A
B
C
G
B
C
A
A
E
B
H
A
E
Like
you,
don't
have
to
worry
about,
like
what
keys
I
mean
we
have
this
problem
right
in
map
now,
where
it's
like,
because
I
mean
it's
benefit
that
we
can
use
these
different
keys
to
construct
server
coordinates,
but
I
mean
you
really
have
like
what
keys
are
used
to
obstruct
that
construct.
That
observer
coordinates
are
different
depending
on
what
source
you're
using
right.
H
H
C
E
H
E
C
D
D
D
Fix
it
because
it
wasn't
doing
it
this
morning,
but
but
I
would
say
I
would
definitely.
I
A
F
B
C
H
E
Yeah,
well,
I'm
just
gonna
say
I
think
I
had
asked
jack
to
like
kind
of
give
an
overview
of
like
misconception
of
like
sort
of
the
the
you
and
you.
Maybe
you
can.
You
can
disagree
with
this
jacket
if
this
is
so
much
you're
going
to
present
but
kind
of
like
that,
the
data
land,
like
data
provider,
landscape
for
solar
physics,
so
like
what
are
the
the
vso,
the
stack,
the
json
etc,
and
how
do
they
sort
of
interact
with
each
other?
But
I
think
yeah,
I'm
gonna
say
more
broadly.
E
A
J
J
And
then
more
specifically,
so,
as
some
of
you
may
know,
the
vso
had
its
first
in-person
tech
meeting
in
three
years
after
the
test
meeting.
So
what's
that
10
days
ago
now
something
like
that,
and
so
we
discussed.
The
very
first
thing
we
discussed
was
interaction
with
the
dso
so
and
thanks
to
stuart
for
staying
up
late
and
honestly.
B
J
To
to
talk
with
us,
so
so
that
was
nice.
J
Never
mind
it's
asking
you
to
quit
and
whatnot.
I
won't
waste
your
time
with
that,
basically
so
like
in
a
broader
view
like
nasa,
is
making
a
big
push
to
sort
of
make
sure
that
data
has
better
metadata
and.
J
It's
easy
to
access,
so
the
open,
open
science
is
a
big
theme
going
forward
for
for
nasa,
so
that
includes
data
data
providers
and
open
source
code
like
some
pi.
So
nasa
is
like
really
kind
of
pushing
hard
in
these
areas
so
that
what
that
means
for
the
sdac
in
particular
is
that
there's
basically
well
we've
asked
for
and
in
the
past
couple
of
years
we've
got
more
funding
to
support
these
types
of
issues.
J
So
that's
good
in
my
in
the
s
tank
budgets.
We
have
support
for
specifically
for
senpai
developers,
so
we'll
be
looking
for
that
coming
forward.
J
Also,
on
that
same
note,
given
the
expanded
budget,
the
sdo
sorry,
the
vso
has
also
expanded.
We,
we
hired
somebody
new
last
year,
so
we
have
more
support
for
the
bso
now
continue
and
then,
as
needs
change,
we'll
we'll
work
on
getting
more
support
as
as
needed.
J
So
that's
kind
of
broader
aspect
like
I
think
you
know
it's
good
for
the
senpai
community.
To
be
aware
of
that,
you
know
this
like
there's
a
bigger
push
towards
like
anything.
We
can
do
to
promote
open
science,
open
source
code,
better
metadata
standards.
That's
that's
where
that's,
where
we're
going.
J
Yeah,
that's
a
good
question,
so
the
answer
to
that
is
qualified.
Yes,
so
the
if
you
we
have
some
funding
to
support
that.
So,
for
example,
sdac
has
been
funding.
J
Lasko
c1
data
lascal
has
three
coronagraphs.
The
first
one
was
apparently
always
fiddly
to
use
and
broke
after
the
soho
vacation
in
1998.
J
So
so
there's
an
effort
on
going
there
to
sort
of
bring
to
make
that
data
publicly
available
and
bring
the
standards
up
to
up
to
modern
requirements.
J
There
is
again
this
is
just
kind
of
nasa
related
but
like.
If
there
is
a
solar
physics
mission
out
there
that
could
do
with
improvements
to
the
metadata,
then
we
can
support
that
up
to
up
to
some
level
so
skylab.
J
Actually
you
know
what
skylab
I've
been
looking
at
skylab
data,
so
yeah
it's
going
back
a
bit,
but
we
had
a
student.
Actually,
he
was
looking
at
some
skylab.
J
J
I
We've
also
changed
the
methodology
for
some
of
maybe
new
data
providers,
but
with
older
data
in
that,
in
certain
cases,
what
we're
trying
to
do
is
actually
grab
the
files
so
that
we
can
parse
all
the
metadata.
That's
in
the
headers
at
least
so
we
can
at
least
provide.
We
can
look
at
that
metadata.
We
can
make
suggestions
to
people
sometimes
to
improve
it.
We've
do
it.
I
We've
done
that
with
naoj
and
h,
alpha
data
from
there
and
they
are,
they
went
back
and
reprocessed
a
lot
of
their
data
so
that
you
know
we
can
scrape
that
and
incorporate
it.
So
we
can
provide
better
meta
information
within
vso
about
older
data
providers
in
general.
Well,
all
data
providers,
but
you
know
some
of
the
older
data
sets
in
general.
E
J
Well,
I
can
tell
you
how
I
I
plan
on
doing
this,
so
you
should
know
that
as
dollars
is
supported
specifically
by
the
sdk
and
what
they
do
is
they
send
unsolicited
proposal
to
well
essentially
to
me,
and
then
it
gets
reviewed
by
members
of
the
community
to
not
me
and
a
decision
is
made
if
it's
worth
supporting
or
not
so
the
that's
the
way
that
at
least
some
of
this
funding
could
be.
It
could
be
worked
that
the
somebody
sends
it'd
be
easier
to
do
this
us-based.
J
Unfortunately,
that
someday
sends
sends
me
an
unsolicited
proposal.
It
gets
reviewed
and
then
it
gets
funded
or
not
so.
J
J
To
do,
and
then
you're
just
meant
to
act
as
like
funding,
given
that
the
community
has
decided
that
this
is
a
good
thing
so
like,
for
example,
if
somebody
wanted
to
say
like
something
controversial
like
I
want
to
remove
fido
from
senpai
and
make
its
own
package,
then
that
would
have
to.
J
I
would
get
that
proposal.
I
would
send
it
out
to
some
people
for
review
and
then
that
would
be
a
decision
would
be
made
like
for
something
like
that
and
clearly
the
you
know,
we'd
want
to
involve
folks
on
on
the
senpai
board
and
we'd
develop,
and
things
like
that.
So
yeah,
okay,.
J
Anyway,
so
that's
the
kind
of
broader
view,
the
I
don't
know
how.
How
would
you
like
to
cut
you
leaving
okay,
so
we
we
took
a
bunch
of
notes.
J
J
J
Like
a
specific
data
provider,
like
I
mean
like,
would
you
be
so
I
mean
an
example
was
saw
the
the
senpai
sword.
H
So
I
think,
speaking
from
the
seven
paisle,
there
are
two
conflicting
sides.
One
is
that
we
want
people
to
have
access
to
the
data
as
soon
as
it's
available
somewhere.
Even
if
that's
not
the
vso,
but
then
the
other
side
is
in
some
high.
We
don't
want
to
be
maintaining
a
client
for
every
mission,
which
is
why
the
vso
is
really
helpful.
I
That
is
where
we
occasionally
do
run
into
issues,
because,
generally,
what
we
do
is
involve
ourselves
in
a
conversation
with
the
people
who
are
providing
the
data.
One
of
the
examples-
and
we
talked
about
this
with
stuart,
for
example-
is
the
suvi
data.
We
were
specifically
asked
when
we
serve
suvi
data
not
to
serve
level
two,
but
as
stuart
points
out
the
level
two
is
up
there
and
publicly
available,
and
I
believe
you
can
get
it
through
some
pi,
but
we
won't.
We
won't
serve
the
level
two
until
the
the
you
know.
I
The
noaa
folks
say
it's
okay
for
us
to
serve
the
level
two
data.
H
C
C
H
H
J
H
J
Yeah,
well,
we
can
tell
you
the
the
success
rate
today
is
about
83
or
something
like
that
out
of
what
we,
what
we
say
we
will
provide
so
yeah,
that's
exactly
good!
So
well,
you
know,
there's
like
there's
the
data
we
say
will
serve
and
the
data
you
can
actually
get
to
today.
So
that
is
we're
like
plus
or
80
most
of
the
time.
But
that's
a
like
a
practical
mountain
yeah
yeah.
I
think
it's
a
and
some
maybe.
C
J
J
J
J
From
like
a
vso
request
to
back
down
to
a
senpai
user,
what's
your
feeling
about
that?
The
idea
that
we
were
thinking
was
to
have
something
a
little
less
opaque.
Perhaps
in
terms
of
the
error
response.
Would
that
be
something
that
some
pie
would
be
we'd
find
useful
or
willing
to
handle
like.
G
J
F
But
like
it's
important
because,
like
you
know,
sometimes
you
do
a
query
with
fido
and
you
get
back
a
random
error
and
then
someone
says
sunpower's
broken,
but
it's
actually
a
vso
client
issue
the
server's
down
for
an
hour,
and
it's
very
unclear
to
I
think,
even
myself
or
users
who
aren't
even
aware,
where
they're
getting
their
data
from
what
the
error
is,
and
they
just
think
it's
impossible.
So
there
was
clear,
more
error
message
of
like
oh,
the
vso
is
down
try
again
in
an
hour
or
raise
this
issue
with
vso.
F
I
think
they're
really
useful.
Or
do
you
remember
that,
like
plot
that
andy
made
jeff
you're
just
talking
about
like
linking
to
that,
even
just
to
show
that
the
data
is
probably
not
available
or
something
I.
A
Mean
the
one
thing
that
I
I
think
would
be
useful:
I
don't
know
how
difficult
it
would
be
from
the
vso
side.
It
is
to
have
a
page
with
a
dashboard
with
a
green
symbol.
If
you,
you
guys,
think
everything
is
working,
because
it's
so
often
someone
has
a
problem
with
a
message
in
the
chat
and
it's
like
oh
yeah,
this
server
is
being
down
or
we're
fixing
an
issue.
A
But
if
you
had
a
page
that
had
a
little
logo
saying
we're
aware
of
an
issue,
then
we
can
just
direct
people
there
and
it's
very
clear
that
you
know
what's
going
on.
I
think
that
would
be
super
useful
because
then
at
least
people
are
aware.
Okay,
there's
a
temporary
issue
being
worked
on.
It's
not
yeah.
I
don't
know.
J
So
as
long
as
the
vso
health
status
page
is
up,
then
that
would
be
the
kind
of
you
could
report
like
whatever
gobbledygook
error
comes
back
and
then
like.
H
E
E
G
C
C
I
So
in
our
discussions
with
stuart,
we
thought
of
a
couple
of
things.
One
obviously
is
this
better
passing
back
of.
You
know
less
opaque
error
messages,
which
is
obviously
one,
and
we
want
that
for
work.
You
know
for
to
actually
let
people
know
that
hey
it's,
not
the
vso
fault,
it's
the
fact
that
hao
is
down
again.
You
know
we're
doing
the
best
we
can
as
well.
So
we
don't
want
to
get
blamed
as
well
as
some
pie
not
wanting
to
get
blamed
because
they
pass
the
book
yeah
exactly.
C
I
So
what
we
talked
about
with
stuart,
as
we've
always
talked
about
creating
you
know
like
a
rest
api
for
the
the
registry
and
also
actually
adding
to
our
registry,
so
that
not
only
does
you
know,
is
it
easily
accessible
to
you
know
all
the
machine,
all
the
data
sources
we
have,
but
also
what
each
data
source
supports
in
terms
of
search
capabilities,
for
example,
but
also
we
want
to
incorporate
things
like
the
health
report
we've
been
generating
into
something
like
that
that
maybe
gets
updated
once
an
hour
with
you
know
when
we
know
that
there
is
an
issue
with
a
particular
data
provider.
I
You
will
also
know
that
I'll
be
able
to
find
out
that
we
know
there's
an
issue
with
that
particular
data
provider
and
so
we're
looking
at
providing
that
kind
of
service
as
well.
That
will
you
know
I
just
just
give
a
an
easy
machine,
readable
version
of
things
like
the
health
report,
so
that
we
know
what
the
status
of
each
data
provider
we
don't
imagine
doing
it
on
a
you
know,
on-demand
basis,
but
certainly
say
once
an
hour
or
something
like
that.
J
C
J
C
E
Providers,
I
think
an
example.
That's
happened.
Is
that,
because
is
it
says,
xrt
xrt
data
is
still
currently
unavailable
from
the
cfa
correct.
So
I
think
there
have
been
several
people
who
have
searched
for
xrt
data
using
senpai
and
they
think
well,
I
think,
usually,
we've
heard
sunpi
doesn't
provide
xrt
data
which
debate
the
merits
of
that
sentence.
But
that's
that's
the
con,
that's
the
feeling
and
then
and
then
the
second
level
is
okay.
I
I
I
I
will
stand
up
and
apologize.
I
have
a
trip
to
cfa
next
month
coming
up
where
I'm
actually
going
to
fix
that
at
last
that'd
be
great
real
time
update,
yeah.
G
F
F
J
H
So
no
but
we've
added
the
cedar
web
client,
which
was
kind
of
supposed
to
access.
The
spdf,
which
is
kind
of
was
specifically
despite
access
to
lots
of
in-situ
data.
J
J
E
We
have
kind
of
feedback
about
solar
institutes
or
over
data.
E
J
H
J
We
just
go
and
ask:
can
we
have
your
data
please
and
if
it's
open
data
then
like
there's,
usually
no
problem
aside
from,
like
you
know,
don't
hit
our
servers
too
hard.
That
kind
of
thing,
so
I
guess
we
you
know
with
the
via
the
vso
like
we.
J
You
know
we
ask
ground
based
observatories
like
naoj,
don't
have
any
there's
no
exchange
of
funds
between
nasa
and
net
and
naoj,
but
like
for
the
the
mission
of
like
providing
solar
physics
data,
then
that's
that's
what
we
do
and
you
know
it's
kind
of
a
favor.
Really
it's
like
this
is
another
way
of
getting
your
data
out
to
scientists.
C
I
mean
steve
kind
of
brought
this
up
the
other
day
about,
like
overlap
with
other
ways
of
getting
in
situ
data
in
the
general
period,
high
pi
hc
community-
and
I
think
I
just
don't
know
enough
about
it
to
know
where
that
is.
But
it's
certainly
something
we
should
consider
not
re-implementing
too
many
wills.
F
I
was
gonna
say,
like
maybe
we're
missing
a
voice
from
the
institute
right,
because
I've
had
heard
people
say
like
why
doesn't
senpai
do
in
situ
stuff?
So
there's
no
institute
people
there
to
complain.
There's
no
institute
things
there
like
if
you,
if
maybe
if
it
was
available,
more
people
would
ask
for
it
and
you
use
it,
and
maybe
you
know
the
fact
that
there
is
some
availability
and
that
they
might
grow.
Or
I
mean.
J
So
useful
senpai
to
the
downloading
situ
data
might
be
partly
a
branding
issue
of
like
in
senpai.
So
it's
got
something
to
do
with
the
sun,
so
it
must
be
remote
sensing
and
then
the
people
who
you
know
the
space
physics
community
on
on
this
side
of
the
pond
have
their
own
ways
of
accessing
the
data
they
want
to
access
and
that
is
like,
historically
not
in
senpai.
So
the
fact
that
it's
that
you
can
do
that
in
senpai
is.
J
F
J
Yeah,
so
I
mean
soho
had
this,
but
soho's
so
old,
that,
like
the
data
analysis,
environment
is
like
not
really
super
relevant
for,
for
what
we're
doing
right
now,
I
think,
like
in
solar.
Orbital
is
like
the
game
has
moved
on,
so
that
if
people
are
inside
of
a
python
environment,
then
using
some
pie
tools
would
be
like
much
more
comfortable.
G
J
So
it
did
like
so
at
the
test
meeting
this.
This
was
brought
up,
that
why
isn't
fido
just
by
itself
and
why
don't
we
have
like
one
downloader
to
rule
them
all
and.
E
Yeah,
there's
specifically,
it
was
well
at
least
I
I
I
I
guess
I
won't
speak
for
the
becoming
the
jack
just
me,
but
I
had
talked
with
the
heliocloud
people
saying
that
they
would
be
very
interested
in
a
like
data
search
functionality
that
is
outside
of
sometimes
because
it
plays
like
breaking.
If
there's
a
strong.
G
E
I
mean
I
think
it
goes
back
to
well
during
this
discussion
that
you're
having
before
about
like.
Why
should
if
people
want
to
download
sort
of
more
space
physical,
what
we
would
call
space
physics
data-
you
know
like
magnetospheric
data
stuff,
that's
kind
of
sort
of
outside
the
sort
of
research
scope
of
what
senpai
has
traditionally
been,
and
they
want
to
use
fido
to
do
that.
Why
should
they
have
to
download
some
guide
to
do
that?
G
For
it
to
take
the
other
side,
if
we
break
it
out,
are
we
just
creating
a
python
version
of
vso,
yet
just
another
standalone
floating
hard
to
come
across
unless
you
already
know
tool.
E
E
But
but
but
practically
speaking,
I
I
definitely
like
I
mean
it's
this
whole
problem
of
like
a
proliferation
of
or
packages
that
makes
it
harder
for
us
to
maintain
lots
of
packages
and
also
it's
like
you
know,
another
thing
that
we
have
to
install
and
maintain
separately
from
the
core
code
base
like
there
are
reasons
practical
reasons
not
to
do
that,
but
there's
nothing
about
like
like
the
really
powerful
functionality
of
fido.
That's,
that
is
solar
system
right.
Okay,
the
clients
are,
but
certainly,
if
I
know
themselves.
E
E
J
J
But
that's
habit,
you
know
so
the
drive
for
this
the
drive
for
like
is,
should
we
have
one
downloader
to
use
them
all
is
more
of
the
like
nasa
organizational
like
it's
all
one,
big
heliophysics,
and
so
why
don't?
We
have
one
big
heliophysics?
Why
don't
we
have
one
package
to
download
all
these
data
instead
of
this
like
bits
and
pieces
here
and
there.
J
J
J
It's
not
it's
not
so,
like
I
mean
the
scope
question
is,
is
you
know
definitely
one
to
keep
an
eye
on,
but
like
it's?
Not
it's
not
a
hard
one
yet
and
nobody's
asking
like
I
resist
the
urge
to
sort
of
have
these
unification
efforts
because
they
tend
to
a
lot
of
effort
for
not
much
gain.
I
mean
you
can
already
like.
J
We
have
great
data
access
as
it
is
like
writing
a
another
client
to
unify
in
situ
and
remote
sensing
data
is
not
or
reorganizing
the
way
that
things
have
been
done
right
now,
just
because
it
looks
better,
it's
not
something
that
I'm
keen
on
doing.
D
I
just
wanted
to
say
something
before
I
go,
that
someone
brought
out
that
if
you
spun
out
fido
that
it
would
just
become
like
a
vso
python
interface,
but
the
data
sources
available
on
sunpack
is
on
fire.
Sorry
go
beyond
just
vso,
so
spinning
out,
fido
is
fine,
but
it
wouldn't
just
become
a
vso
client.
D
E
D
G
C
C
Just
to
say
it,
my
opinion
on
moving
fido
out
is
like
two
problem.
Nobody,
nobody
has.
There
would
be
no
point
in
doing
it
unless
there
was
users
right
and
I
think
by
that
it
means
like
if
happy
go
ahead
and
write
their
fido
plugin.
Where
is
that
going
to
live
like
do
they
want
to
put
that
in
their
own
happy
package
in
senpai
or
in
a
third
place
where
fido
live?
If
we
spawn
out
fido,
is
it
just
fido
and
no
clients
that
live
in
they
live
in
the
fido
package?
E
C
There
is
like
a
chunk
of
the
phytoactive
system
is
like
the
core
actors.
Non-Client
specific
attitudes
in
fido
are
are
a
yeah,
but
they
are
a
reflection
on
the
data
models
of
the
clients
that
we
have
implemented.
They're,
not
they're
there.
There
is
a
data
model
in
those
it's
heavily
biased
by
the
vso,
but
there
is
a
data
model
in
those
core
assets
we
provide.
They
are
probably
not
applicable
to
to
happy
as
much
as
they
are
to
the
sort
of
data.
C
C
C
Would
it
still
be
governed
by
us
and
would
we
have
to
spin
up
a
new
community
around
it,
whereas
the
other
but
like
if,
if
multiple
large
heliophysics
people
rocked
up
and
were
depending
on
it,
they're
going
to
want
to
get
involved
with
the
decisions
made
around
it
like?
What?
What
is
the
community
implications
of
moving
out
to
somewhere
else
and
is?
Is
the
fact
that
it's
not
moved
out
somewhere
else,
actually
hindering
them
coming
up
and
doing
it
from
that
community
aspect?
Do
the
heliophysicists
feel
like
they
don't
really
have
like
that?
G
Summarize
there
are
several
arguments
for
but
no
one's
making
those
arguments
so
until
somebody
does
the.
C
Only
the
only
coherent
argument
I've
heard
is.
Why
would
I
need
to
import
somebody,
and
that
is
not
a
valid
argument
to
me.
There
are
lots
of
valid
arguments,
but
I
don't
think
anyone's
made
them
to
us
yet,
and
I
don't
know
whether
that's
just
because
they
don't
feel
like
they
can,
or
they
have
just
immediately
dismissed
it.
I
don't
know
just
like
there
was
so
so
much
going
on.
E
D
C
I
think
the
response
to
people
who
ask
this
question
is:
if
you
would
like
to
see
that
happen,
you
have
to
come
and
talk
to
us
about
it
like
open
an
issue
on
the
repo
like.
I
want
to
use
fido
for
this
thing,
but
I
feel
like
it
being
in
some
places
in
some
way.
Stopping
me
let's,
let's
talk
about
it
right
like
we're,
not
going
to
just
do
that,
because
you
might
want
to
use
it
if
you
want
to
use
it
and
that's
a
problem
for
you.
E
J
Well,
so
I
think,
on
the
via
side,
we
got
a
couple
of
things
of
this
that
the
the
health
report
seems
to
be
a
good
thing
for
some
pi
to
direct
their
users
to
we
can
write
better
error
messages
that
can
percolate
back
to
to
some
pi
users.
That's
a
good
thing.
B
B
G
But
right,
but
then
also
made
it
publicly
available.
So
in
the
case
of
we
consider
to
be
so
as
like
some
sort
of
official
tag,
and
they
don't
consider
that
I
mean.
A
A
A
Have
no
idea
the
number
of
times
I
got
to
do
something
and
something
somewhere
is
down
not
necessarily
the
vso's
fault,
not
necessarily
their
data
flyer's
fault,
but
something
somewhere
is
broken.
However,
I
have
a
client
that
I
wrote
that
goes
in
yesterday,
every
time
no
bother
because
it
sits
on
a
web
server
and
it's
just
there
happens
all
the
time
with
a
near
real-time
data.
Obviously
we
want
solarmonitor.org,
that's
what
we
hit
all
the
time
and
it
goes
down
so
often
in
other
ways
and
there's
many
other
kinds
of
the
same.
A
That's
great,
and
I
understand
why,
from
a
maintenance
perspective
and
a
co-perspective
and
lots
of
reasons,
but
you're
really
hampering
people
getting
the
data
when
they
wanted
them.
I
I
think
that
that
is
a
big
problem,
because
sometimes
people
will
try
it
and
if
it
doesn't
work
first
time,
they're
never
coming
back.
B
C
So
like
going
back
to
like
duplicating
stuff
with
the
vso
suvi
is
a
good
example
of
this
for
the
the
stuff
that
they
do
index.
If
we
wrote
an
ara
and
our
a
I
a
nrt
client,
it
would
be
a
similar
deal
right.
We're
just
deliberately
duplicating
results
with
the
vsos
like
how
does
that
affect
the
ui
of
fido
right
like
how
does
that
that
adds
significant
complexity
to
how
people
interact
with
fio
so
like
what
and
what's
the
trade-off
with
that,
and
also
not
a
lot
of
the
data
we
do.
Duplicate
jsoc
in
particular.
C
Right
has
a
completely
different
data
model
and
search
cell,
so
you
basically
never
hit
the
two
simultaneously
with
one
one:
query:
no
walls
right
and
whereas
with
suvi
and
other
things
you
like
scraper
clients,
you
can
very
easily
construct
a
thing
where
you're
duplicating
all
the
results
and
that's
a
lot
of
extra
code
for
us
and
a
lot
of
complexity
for
the
users,
when
the
reason
that
it
is
on
might
not
be
on
vso
might
just
be
that
your
web
server's
down
right
like
if
your
web
server
goes
down.
A
C
I
mean
to
some
extent
you're
also
trading
off
reliability
with
simplicity,
right,
like
in
some
ways
directly
querying
the
jsoc
is
more
reliable
than
going
through
bso,
especially
for
like
very
large
queries,
but
not
not
putting
that
in.
As
the
leading
example
like
that's
that's
it
I
mean
the
fact
our
jsoc
client
exists
and
we
can
do.
That
is
very
useful
and
it's
great
that
we
have
the
functionality
to
do
that.
C
A
I
mean
radio
spectra
is
another
example
from
one
where
the
data
is
just
all
over
the
shop
and
like
you,
have
no
choice
but
to
write
a
whole
bunch
of
clients
to
get
it,
and
one
of
them
is
the
store
so,
like
I
don't
think,
bso
is
going
to
ever
index
the
radio
data
from
solar
orbiter.
It
doesn't
go
into
the
institute
data
either.
I
don't
think
so
like
there's
an
example
of
some
data
that
we
can't
get
unless
we
write
a
client
for
store.
I
That
might
be
a
case
of
something
I
mean
we
don't
have.
We
know
that
we're
like
we're
very
light
on
radio
data
within
vso,
and
that
is
something
that
we
are
going
to
look
at,
but
also
you
know
this
may
be
a
case
where,
if
it
falls
between
the
cracks,
we
could
cover
it.
I
It's
just
that
if
we
know
it's
designed
to
go
into
spdf
or
something
like
that,
we
tend
not
to
look
at
that
kind
of
data
so,
but
that
you
know
if,
if
somebody
was
to
request
access
to,
you
know
ask
vso
to
provide
access
to
the
data,
then
that's
that's
one
of
the
ways.
We
know
that
people
actually
want
access
to
the
data,
and
so
we
can
look
at
that.
J
So
I
have
a
question:
what
is
the
degree
of
overlap
between
some
pi
provided
clients
and
vso?
How
many
instances
are
we
talking
about
here?
Where's
the
list.
C
What
we've
got
you
might
have
to
update
us
on
what's
currently
available
in
the
vso
as
I
hit
this
list,
but
we
have
a
client
that
accesses
the
level
zero
cs
eve
data.
I
don't
think
that's
in
the
vso
thanks
pierce
yeah
there's.
We
have
a
fermi,
gbm
client.
A
C
J
C
I
I
Yes,
lyra
data
will
be
on
real
shortly.
That
was
part
of
our
interface
with
esa.
A
I
I
think
technically,
we
are
meant
to
actually
serve
the
xrs
data,
but
we
don't
at
the
moment.
I
On
that,
I
believe
we
have
a
ticket
for
it.
H
H
And
I
think
this
was
really.
I
think
this
was
a
really
big
thing
with
somebody
right,
so
I
thought
it
should
have
gone
somewhere
else.
I
think
correct
me
if
I'm
wrong,
but
still
you
shouldn't
have
because
it
was
mentioned
going
to
be
in
this
world,
but
I
thought
it
was
really
important
that
we
had
support
first
person.
A
A
C
C
Worry
yeah,
I
I'm
not
sure
I
actually
want
to
do
that.
I
was
just
bringing
it
up
as
a
like
technical
option.
I
so
just
saw
is
an
interesting
question.
My
issue
with
saw
early
on,
was
partly
the
so
we're
getting
implemented
and
also
we
didn't
have
to
talk
to
them,
but
we
have
now
rectified
that
problem.
We
have
contact.
We
saw
their
up
first
help
supporting
us
writing
here.
C
I
think
the
stock
line
has
implemented
could
probably
do
with
a
bit
more
push,
but
I
haven't
really
looked
at
it
too
much
recently,
so
maybe
it's
better
than
last
time.
I
tried,
and
I
know
that
heaven
said
there's
a
few
things
that
were
changing
on
the
rpas.
I
don't
know
whether
they're
being
reflected
anyway
ramble
rattle
ramble.
We
can
move
into
core
eventually,
like
I
think.
E
I
I
I
A
A
B
E
C
C
C
Yeah,
I
think
I
think,
from
my
perspective,
we've
saw
now
that
we
have
that
initial
relationship
with
the
team
before
it
goes
into
court.
I
would
like
the
sorting
to
basically
sign
off
on
it
like
yeah.
I
want
them
to
agree.
Yes,
we're
going
to
help
you
and
yes,
what
you've
implemented
here
is
how
we
expect
it
to
work
like
I
think,
maybe
the
best
way
to
get
that
feedback
is
crack
the
pr
and
get
them
to
look
at
it.
Maybe
it
isn't.
E
C
Yeah,
I
I
think
this.
This
is
actually
the
first
time
we've
done
this
right.
This
is
the
first
time
where
we've
set
up
a
relationship
with
the
data
provider
before
we
put
the
code
in
core
right.
So
I
think
this
is.
This
is
a
good
model.
This
is
good
right,
but
we
actually
need
to
get
their
input
on
on
it
before
we
merge
it
yeah.
I
think.
A
J
B
J
J
G
J
C
That,
but
at
the
tech
meeting
that,
like
idea,
came
up
with
like
python
vso
data
providers
based
on
binary
clients
like
yeah,
is
that
I
don't.
I
don't
know
what
like
the
situation
is,
but
I'd
be
happy
to
like
work
on
that
in
the
theory
where
I
have.
The
time
like
you
know,
like
it'd,
certainly
be
interesting.
If
a
senpai
their
fido
client,
could
you
know,
make
it
easier
for
the
vso
to
implement
a
thing
eventually.
G
C
G
Not
so
much
more
like
a
technological
base,
I
meant
more
of
like
a
human
level.
Yeah
like
there's
a
channel
of
communication
between
songfire
and
vso,
saying
we're
getting
like
requests
for
this
data
users
are
trying
to
find
this
data.
We
can
write
our
own
client
for
it.
You
know.
Maybe
we
can
like
raise
it
as
a
vso
ticket
that
you
know
we
have
a
client,
but
you
perhaps
you
should
consider
serving
that
and
then
what
you
do
we'll
get
rid
of
our
client.
C
I
think
one
thing
to
consider
in
that
workflow
that
I
can
definitely
see
working
is
if
the
va,
if,
if
we
just
write
a
scraper
client
like
suvi
and
then
the
vso
start
chatting
to
the
team
and
they're
actually
like.
Oh,
we
can't
serve
that
data.
We
end
up
in
a
split
brain
situation
where
we've
written
we've
done.
One
thing
and
vso
has
done
another
and
now
we're
like.
Oh
well,
now
it's
in
both
places,
but
we
also
can't
just
delete
ours
because
does
some
stuff
that
the
vso
one
problem,
but
if
we
open.
G
The
line
of
communication
like
at
the
very
start
when
we're
considering
writing
a
client.
If
we
then
sort
of
link
the
vso
into
that
discussion,
saying
you
know,
we've
had
requests
for
this.
This
is
what
we
plan
on
doing
then
the
vso
can
potentially
say
wait
a
minute
like
we've
spoken
to
the
team
and
they
said,
for
whatever
reason,
we're
not
providing
this
for
at
their
request
and
then
we're
more
informed.
G
F
B
F
I
I
think
we'd
be
more
than
happy
to
provide
some
kind
of
you
know,
institutional
contact,
you
know
and
and
to
work
with
you
guys
on
these
I
mean
we,
we
don't,
we
don't
want
to
be
the
people.
I
think
who
say
you
can't
do
something,
but
we'd
certainly
like
to
be.
You
know
know
when
you
guys
want
to
do
something,
because
you
know
we
can.
We
can
perhaps
help
facilitate
that
on
a
more
you
know,
on
a
more
institutional,
unsupported
level.
C
Yeah,
so
in
that
spirit,
going
through
those
clients
that
we
already
had
what
I
didn't
get
from
going
through,
that
list
was
something
else
and
jackie.
G
Said
no,
we
don't
serve
that
data.
The
next
question
is
like.
Is
that
just
because
you
don't
because
you
haven't
had
requests
or
would
you
do
you
have
a
plan,
or
would
you
be
willing
to
consider
taking
on
that
data
or
should
we
you
know,
should
the
model
then
be,
but
that
specific
client
we're
going
to
keep
it
for
like
indefinitely
like?
Would
you
consider
serving
a
gang
data,
for
example,
or
the
synoptic.
I
That's
something
we
should
actually
be
doing
so
you
know
that's
something
that
we
should
go
and
look
at
and
make
sure
that
we
have,
and
then
we
can
tell
you
yeah,
we
have
it
and
make
sure
you
can
get
it
reliably.
A
D
B
J
On
the
other
hand,
it
is
data
that
you
know
people
have
asked
senpai
to
make
available,
and
given
that
you
know
this
is
solar
physics,
relevant
data,
then
you
know.
A
I
mean
do
we
have
any
concerns
about
like
pulling
all
the
data
from
the
states
like
fought
over
here
in
terms
of
their
bandwidth,
and
things
like
that
like
is
it?
Is
there
any
any
distortion
to
have
like
mirrors
of
the
vso
in
europe
and
stuff,
just
to
reduce
data
transfer
and
stuff
like
that,
or
is
that
a
concern
that
we.
F
C
E
Yes,
it
is
stock
has
a
mirror
which
is
back
up
and
running
now.
Their
power
outage.
A
I
C
C
C
C
I
Well,
I
mean
so
ed
and
I
normally
sit
and
obviously
jack
sit
on
the
senpai
chat.
Looking
for
any
vs,
you
know
things
come
up
which
should
be
so
related.
You
know
anything
we
know
which
is
data
provider
related.
We
try
to
pass
on,
but
stuart
obviously
has
been
our
main
conduit.
He
sits
on
the
the
vso
channel
as
well
and
is
more
than
happy
to
lob
hand,
grenades
at
us
so.
A
G
C
H
B
J
Any
of
the
chats
so
like
for
the
pso
developers,
sorry
senpai
developers
chat.
C
I
think,
like
the
chat's
good
like
is
it
I'm
really
really
happy
that
we
have
good
chat
system
with
vsr?
It's
really
helpful
when
we're
asking
good
questions
on
top
of
it.
I.