►
From YouTube: SunPy Developer Discussion
Description
No description was provided for this meeting.
If this is YOUR meeting, an easy way to fix this is to add a description to your video, wherever mtngs.io found it (probably YouTube).
C
C
B
A
B
A
A
I
have
got
the
cover
of
what
his
actual
name
is.
I
only
know
him
buys
IRC
tag.
All
the
way
to
describe
him.
I
have
had
to
someone
who's
hanging
around
in
the
IRC
channel.
Didn't
direct
he's
interested
in
G
sub
is
being
overhead
him
work
on.
You
me
down.
He
played
around
with
it
a
bit,
try
to
fix
up
and
tests
and
also
as
darkness,
some
and
done
the
starts
of
some
documentation,
switches,
PR,
16,
20
23,
which
frankly
I
might
just
merge.
A
A
A
D
C
A
C
D
C
A
C
A
That's
a
lot
of
that
that's
space
Ortiz,
that's
one
and
then
another
slip
and
extra
nine
characters
every
time.
My
for
a
second
also,
if
you
include
microseconds-
and
this
is
another
15
characters,
and
then
you
should
also
into
your
doing
it
full
iso
date.
Formatting
is
usually
include
to
zed
for
the
timezone.
So
that's
16
and.
A
A
D
A
C
A
E
D
D
F
B
A
C
C
C
A
Been
over
this,
it's
time
range
is
unnecessarily
I
can't
know
it
shouldn't
use
time
range.
We
have
a
timer
and
object
just
much.
Every
years
they
had
once
worked,
it
will
break
so
hard
if
we
use
it.
There
are
good
reasons
that
is
done.
My
best
I
can
make
time
range
be
supported,
as
in
Fido
attribute
would
cause
so
many
circular
import
problems
because
you'd
have
to
import
net
from
time.
It
would
be
horrible.
You
can't
do
that.
Let
right
so
time
is
a
subclass
of
simpler
tribute
right,
which
is
defined
inside
net.
A
C
A
A
C
E
A
A
C
A
A
C
B
You
do
you
do
get
a
look
at
em,
16
24,
which
is
a
slightly
different
issue
that
the
what
we
display
is
correct,
then,
even
though
what
is
displayed
as
a
result
is
incorrect.
The
files
that
you
expect,
at
least
from
the
light
of
client
and
seemed
to
be
fading,
sorry
gets
like
this
Phillips
I
missed
an.
E
E
B
So
here's
another
point.
So
if
we
have
met
night
right
as
the
as
the
end,
when
you're
specifying
a
the
FG
specify
a
date
only
and
then
you
have
midnight
get
added
on
automatically,
then
in
1624,
how
many
files
should
I
get
down
and
the
original
pizza
and
that
first
piece
of
good
beer
to
get
the
file
at
the
seventh
of
March,
because
that
doesn't
quit
due
to
the
file
name.
Doesn't?
Would
you
know
midnight
there?
A
A
A
B
So
lighter
files
go
from
like
a
single
day
files
write
a
correction,
and
so,
if
I
specify
like
just
using
the
date,
the
the
first
second
and
third
right
do
I
go
from
the
01
/
or
12
01
/
14
January.
First
two
january
thugs
yeah.
Then
it
was
lighter
I
mean
if
I
we
should
lighter
the
tongue
to
your
tea
I.
D
A
A
Mean
it's
not
something
I,
don't
think
it's
something
we
gave
too
much
thought
to
before,
but
actually
changing.
That
would
be
a
very
large
behavior
change.
Whatever
should
I
mean
yeah
do
speak
line
I
mean
that
the
thing
that
would
have
to
catch
that
is,
we
would
need
very
special
input
handling
on
the
timer
Tribune,
because
we
wouldn't
the
time
range
object
to
do
that.
It's.
D
A
E
A
D
A
A
D
C
D
A
B
A
D
C
The
solution
is
I,
think
the
following
one:
we
need
to
document
whether
the
the
the
time
range
is
open
or
closed
right
on
which
side
right.
You
need
to
do
that
here.
You
then
the
other
thing
is,
I
would
say
we
need
to
specify.
We
need
to
force
them
to
give
us
a
time.
You
know
up
to
the
minutes
and
then,
assuming
that
we
fill
out
zeros
with
the
rest
of
it,
then
you
remove
this
ambiguity
about
midnight.
I
think
the
day
issue.
Is
it
as
a
problem
right?
C
E
C
A
C
C
D
On
the
right,
then,
it's
open
on
the
right.
So,
whatever
time
you
specify
on
the
right
it's
up
to
and
including
that
I,
don't
think
you'd
mean
that's
right,
so
you
don't
have
to
get
hours
and
minutes
if
you
just
have
a
convention
that
it
builds
out
the
rest
with
zeros
and
it's
open
on
the
right.
That's
again,
that's
actually
how
how
it
behaves
most
of
the
time
anyway.
D
That
is
currently
what
it's
doing
so,
that's
intuitive
as
up
to
and
including
you
later
once
you
know
that
and
then
you,
if
you
you,
make
mistake
as
you're
getting
getting
used
to
it.
Then
the
patterns
that
you
get
in
the
results
that
you
return
tell
you
that
pretty
much
intuitively
it's
open
on
the
right
and
fill
in
the
zeros
with
any
kind
elements
that
you
don't
you
don't
include
it.
A
So
if
we
want
to
it's
true
at
the
moment,
we
have
an
open
interval
on
the
right,
because
if
you
specify
any
time
to
any
precision,
it
includes
that
time
right
if
we
wanted
to
close
the
interval
right
so
that
we
don't
include
the
second
time
the
way
we
would
have
to
do
it
from
having
a
quick
look
at
time.
Delta
thing
on
have
a
great
time
too.
Much
of
this
is
not
completed.
D
D
A
A
moment
the
moment
is
currently
closed.
It
includes
the
end
point
if
we
wanted
to
make
it
so
it
doesn't
include
the
end
point.
We
would
just
subtract
one
micro
second
from
the
time
specified
on
the
right
hand,
side,
and
then
it
would
always
be
one
microsecond
back,
which
means
it
wouldn't
actually
include.
The
time
passes
the
right
time.
B
C
A
D
I
think
in
this
issue
this
this
discussion,
whether
we
force
people
to
add
hours
and
minutes
if
Toews
you
run
from
whether
the
interval
doesn't
close
and
issues
they're
talking
about
here,
it's
actually
a
separate
thing
all
together
for
you,
it's
a
it's
a
AP
I.
You
know
issue
really
is
how
we
want
people
to
use
it,
not
how
actually
work
yeah.
E
C
B
A
A
A
A
E
A
A
A
C
A
A
Say
just
put
up
an
issue
and
don't
suggest
the
implementation
yeah
sure,
but
I
mean
if
someone
wants
to
come
up
with
a
better
implementation,
have
fun
I,
can't
think
of
one
yeah
I
mean
but
yeah
put
the
issue
up
we'll
deal
with
it
later
and
okay.
So
that's
it
by
inaudible.
Sorry,
man
issue
is
already
up.
You
should
already
others
it.
1624
is
the
issue
essentially,
okay,
oh
sure,
but
we
should
have
a
six.
We
should
have
a
specific
one.
Other
is
your
oppa
mother,
1630.
C
C
A
E
C
Care,
you
can
maybe
have
it,
you
can
maybe
have
it
have
a
property
that
tells
it
whether
it's
open
or
closed,
but
the
time
razor
object
itself
should
judge
should
not
change
the
time
that
you
give
it.
But
then
then
that's
closed
and
then
it's
closed
interval
know
you
could
say
you
could.
We
can
maybe
add
an
attribute
that
says
you
know
this.
The
left
imitable
is
open.
The
right
implement
all
is
closed.
Just
like
when
you
write
an
actual
interval
right,
you
would
use
the
square
brackets
or
the
boys
room
go
brackets.
C
B
C
B
C
Would
say
we
could
we
can
document
it
in
the
time
earning
object,
and
the
time
attribute
just
say
this
is
this
is
how
this
is
meant
to
be
treated,
and
it
should
always
be
consistent
with
that
right
and
then,
if
we
wanted
to
act
differently
than
we
can
add
an
attribute
to
those
objects
that
say,
I
want
this
to
be
treated
differently.
Right,
I
want
it.
A
A
Yes,
so
is
this
heads
they
simple
sup
from
some
fight
on
their
importance
right,
yeah,
a
has
these
things
in
it
so
instrument,
time
and
wavelengths
are
universal
over
all
clients,
and
then
you
have
client
specific
in
their
own
namespace
yeah.
So
that
is
just
a
importing
right.
Yeah.
So
are
you
asking
whether
we
should
provide
a
shortcut?
So
do
we
want
to
shortcut?
Fido
is
really
the
question
here.
C
A
C
A
A
A
B
C
B
A
And
just
yeah
I
just
think
for
the
documentation,
because
it's
something
that
people
are
likely
to
do
from
the
command
line.
I
know
I.
Do
it
from
the
command
line
a
lot.
In
fact,
one
of
my
primary
uses
of
some
pie
as
a
user.
There
aren't
many.
The
one
thing
that
I
do
occasionally
do
is
go
I,
wonder
what
that
flare.
Look
like
this
time
and
I
so
much
with
the
typing.
C
A
C
Units,
yes,
so
I
would
just
say
that
yeah
there
is
a
there
is
a
balance
that
needs
to
be
struck
between
the
two
I
could
go
either
way.
Metra
boots
is
not
very
helpful
anyway,
yeah
so,
and
then
I
do
like
this
idea
of
like
a
time.
So
if
you
want
to
use
a
and
the
docs,
that's
fine
with
me,
but
we
should
not
make
this
something
we
do
often,
because
really
it's
only
like
one
extra
key
stroke
right,
you
type
a
then
tab.
C
B
B
C
Where
are
you
deck
with
the
map
cube?
Oh
I
should
make
a
pee
on.
B
C
C
The
pixels,
the
the
stuff
and
the
pixels
kind
of
staying
the
same
right,
mm-hmm
and
and
so
as
we're,
adding
that
we
need
to
be
more
careful
that
what
users
are
doing,
and
it
would
be
easy
to
give
that
little
warning
on
right
during
the
standard
deviation
of
a
pixel,
is
only
worthwhile
if
the
stuff
inside
the
pixel
hasn't
moved
over
to
the
next
pixel
by
the
time
you're.
At
the
end
of
your
series
of
maps.
C
B
E
A
C
A
A
A
C
C
C
C
We
go,
I
was
trying
to
use
map
tube
and
I
was
very
frustrated
because
currently,
map
cube
is
really
provides
very
little
functionality.
It's
basically
like
slippery
I
to
list
basically
yeah
yeah,
it's
basically
a
list
and
it
provides
the
plotting
animation,
which
is
nice,
but
I
should
have
linked
to
the
actual
file.
No
see
that's
interesting.
C
Well,
we
with
Jack
and
I.
We
started
talking
and
we
both
got
started
griping
about
it,
and
you
know
that
you
know
we
have
this
kind
of
higher-level
goal
of
treating
everything
properly.
So
here's
the
actual
file
treating
it.
You
know
as
a
n
dimensional
data
cube
blah
blah
blah
all
that
stuff
and
I.
Think
that's
what
stopped
us
at
first
from
doing
anything
even
from
providing
a
nicer
implementation
of
map
cube
because
we
were
all
you
know:
ND
dedos.
A
A
C
C
I
cute
eyes
it
essentially,
and
then
it
also
is
being
sort
of
in
some
places
it's
kind
of
assumed
like
when
you
do
a
sub
map.
If
you
know
it
could
break
if
they
weren't
really
the
sucking
at
the
same
places
right,
maybe
the
sub
map
would
apply
to
you
know
a
fraction
of
the
maps,
but
not
all
of
them.
Okay,.
A
Right,
okay,
so
I
think
there
are
two
separate
things
here:
one
I
kind
of
like
and
the
other
on
I
kind
of,
don't
all
succeed
in
the
festival,
so,
first
off
the
concept
of
doing
things
like
sub
map,
super
pixel,
resample,
attribute
access,
etc,
etc.
On
a
map
cube
not
as
currently
standing
matthew,
something
that
I've
had
in
mind
for
a
while
would
have
never
had
sat
down
and
found
time
to
implement
was
I
think
it
would
be
really
cool
if,
with
the
current
map,
you'd
implementation
ignoring
spatially
aligned
at
the
same
size.
A
A
C
So
Jack
and
I
also
talked
about
that
and
you'll
notice.
We
have
implemented
some
of
that
and
basically,
what
this
is
doing.
It
just
gets
you
the
first
one,
because
on
import,
it's
got
it.
It
checks
whether
some
of
these
things
are
the
same,
and
we
then
that
kind
of
goes
into
this.
What
are
we
assuming
on
this
map,
cube.
A
But
okay,
so
this
is
my
thing:
I
haven't
gone
into
the
assumption
yet
I
I,
like
the
fact
that
Matthew
doesn't
make
any
assumptions
about
the
homogeneity
of
the
things
it
holds.
I
agree.
This
is
very
limiting
in
certain
situations,
where
very,
very
limiting
in
certain
situations.
Composite
lab
also
tell.
A
Is
the
temporarily
alive
right,
not
the
explicit
assumption.
The
implicit
assumption
is
that
they're,
temporary
temporarily
aligned
yeah
but
Matt
cube.
Maybe
maybe
this
shouldn't
be
called
knocking?
Maybe
the
fact
that
Matt
cute,
maybe
Matt
cube,
is
misleading
for
the
current
Matthew
I
was
trying
to
think
of
a
new
name
for
this
new
thing.
You
were
right
making,
but
actually
thinking
about
it.
Maybe
map
coo
is
the
wrong
name
for
what
is
currently
mapping
made
a
very
loose,
very
non-restrictive
collection
of
maps.
Very.
D
A
To
be
able
to
awkward
lists
of
maps
together,
yeah
yeah
angry,
so
perhaps
we
should
cool
make
map
you.
Something
else
provide
this
functionality
to
do
like
Matt
list
map
list
would
be
good
and
provide
provide
functionality
to
apply
all
the
map
API
to
Matt
list
and
get
it
ripples
back
yeah
and
then
try
and
actually
make
a
cube
object.
C
A
A
A
C
A
C
A
I
mean
you
need
something
which
is
you.
It
needs
to
be
obvious
that,
like
it,
needs
to
really
enforce
those
assumptions
and
make
sure
that
it
actually
makes
sense,
because
then
yeah
that
functionality
would
be
great.
If
you
can
make
those
assumptions
that
functionality,
if
you've
really
really
useful
and
you
wouldn't
have
to
go
very
far
down
the
rabbit
hole.
But
you
can't
make
the
current
implementation
of
map
cube.
Do
that
so.
B
A
A
A
B
C
A
A
A
C
C
A
You
are
talking
about
making
cute
something
very
similar
to,
albeit
substantially
less
complicated
and
the
WCS
q
prototype
that
my
tio
rope
yeah
and
it
needs
to
eat
I.
This
should
be
something
else
other
than
Matt
goob.
It
needs
to
have
a
separate
name
and
not
be
Matthew.
I,
like
this
approach.
I
like
this
kind
of
like
top-down
approach
to
it,
I
think
it
will
actually
make
something.
That's
usable,
but
it's.
C
B
C
C
A
Ok,
so
I
think
you
could
I
think
with
the
map.
You
think
I
think
if
you
on
the
instance,
if
you
create
that
object,
build
the
3d
array
and
then
replace
the
dot
data
or
underscore
data,
whichever
one
it
is
attribute
of
the
input
map
classes
with
just
slices
of
the
number
array,
I
think
it'll,
where
I
think
that's
literally
all
you
have
to
do
and
then
you're
only
storing
the
data
once
and
it
will
point
back
to
the
same
right.
Whoo.
C
B
C
C
Go
in
and
take
the
e
to
the
maps
and
just
replace
the
dot
data
with
this
slice
of
this
of
the
map
cube
data,
and
so
now
you
only
have
it
once,
and
so,
when
you
call
when
you
ask
for
data,
the
cube
it'll,
give
you
the
thing
and
then,
when
you're,
using
the
Maps
individually,
it's
taking
a
slice
of
the
original,
the
data
cube
and
so
you're.
Not
it
all
solved
this
issue
it.
C
A
A
A
A
It
was
then
you're
holding
the
memory
you're
holding
the
data
in
memory
twice.
You
only
want
the
arrays
wants,
because,
when
you're
dealing
with
cubes
of
image
data,
you
really
are
you
run
into
array
size,
memory
limitations.
You
really
only
want
that
data
in
ran
month,
I
mean
what
I've
just
did.
The
problem.
I've
just
described,
is
exceedingly
pretty
niche
right
who
actually
ever
uses
the
delete
statement
in
Python
I
mean
where,
let's
just
that
is
solved
by
documentation.