These are chat archives for astropy/astropy

2nd
May 2014
Stuart Mumford
May 02 2014 11:17
hello
Thomas Robitaille
@astrofrog
May 02 2014 11:17
hi there!
Stuart Mumford
May 02 2014 11:17
right so the organisation chat rooms are private
but the repo ones are public
Thomas Robitaille
@astrofrog
May 02 2014 11:17
right, maybe
strange
oh well, this well
*this works
Stuart Mumford
May 02 2014 11:18
This is pretty awesome
Thomas Robitaille
@astrofrog
May 02 2014 11:18
Yes, I'm going to advertise it on astropy-dev
I like this:
def poisson(lam):
"""
Poisson Probability Distribution Function.
"""
p = []
k = 0
while len(p) == 0 or p[-1] > 1.e-16:
p.append(lam ** k * exp(-lam) / factorial(k))
k = k + 1
return p
except that it's the opposite of IPython (shift-enter to continue on next line)
Stuart Mumford
May 02 2014 11:19
I am very attached to IRC tho
if you switch to compose mode
it goes multi line
them ctrl+enter to send
Thomas Robitaille
@astrofrog
May 02 2014 11:21
oh
how do you switch to compose mode
Stuart Mumford
May 02 2014 11:22
button on the rhs of the typey box
Thomas Robitaille
@astrofrog
May 02 2014 11:22
ah, I see, awesome
this would be a good chat place during hangouts to paste code
Stuart Mumford
May 02 2014 11:23
indeed
this is pretty close to being damn bitchin
Ghost
@ghost~53638049048862e761fa0380
May 02 2014 11:24
Good morning / afternoon.
Thomas Robitaille
@astrofrog
May 02 2014 11:24
Hi @taldcroft!
Stuart Mumford
May 02 2014 11:24
you get email notifications if you get mentioned in a message and you arn't on the page
hey @taldcroft
Thomas Robitaille
@astrofrog
May 02 2014 11:24
do you see the conversation above? or just from when you joined onwards?
Stuart Mumford
May 02 2014 11:24
it would be cool if you could post GH issues etc.
Ghost
@ghost~53638049048862e761fa0380
May 02 2014 11:24
I guess this is the way to suck 100% time into astropy?
Stuart Mumford
May 02 2014 11:25
they are working on a mobile app
so you would get mobile notifcations
Thomas Robitaille
@astrofrog
May 02 2014 11:25
you mean create new issues? or refer to existing ones? astropy/astropy#2415
Christoph Deil
@cdeil
May 02 2014 11:25
Ahh ... much slicker than the IRC webchat ... :clap:
Stuart Mumford
May 02 2014 11:25
create new issues
we can do that using the bot in the sunpy IRC channel
Ghost
@ghost~53638049048862e761fa0380
May 02 2014 11:25
This feels like it could be a home for me. I never got into IRC so much...
Stuart Mumford
May 02 2014 11:25
I love IRC
Thomas Robitaille
@astrofrog
May 02 2014 11:26
I really like the code syntax highlighting. Also you can use it kind of as a non-instant chat
Stuart Mumford
May 02 2014 11:26
but this has many more awesomes
Ghost
@ghost~53638049048862e761fa0380
May 02 2014 11:26
Yeah, this week I needed to chat with @eteq a few times.
Stuart Mumford
May 02 2014 11:26
and there is a WIP irc client for this
Thomas Robitaille
@astrofrog
May 02 2014 11:26

# awesomes

Stuart Mumford
May 02 2014 11:26
which makes it very awesomes
I am on IRC
Thomas Robitaille
@astrofrog
May 02 2014 11:27
anyway, I'm back to work but hanging out here on the side
this might be useful for the feature freeze rush
Stuart Mumford
May 02 2014 11:27
python hello

# MARKDOWN

oh cool
I type markdown in the IRC window and it still renders
Thomas Robitaille
@astrofrog
May 02 2014 11:28
:+1:
@keflavich
May 02 2014 11:44
whoa awesome
Should this be linked from READMEs in every repo, then?
ah, and where are they archived?
Thomas Robitaille
@astrofrog
May 02 2014 11:46
not sure about archiving - since it's an IRC channel we can always log it externally
@keflavich
May 02 2014 11:46
yeah, but it would be cool to have it version-controlled too =)
yes, certainly
does it notify you especially if I say your name, astrofrog?
Thomas Robitaille
@astrofrog
May 02 2014 11:48
my fan is on max speed though
somehow this is guzzling CPU
Stuart Mumford
May 02 2014 11:49
@keflavich, use @
@astrofrog, use the IRC client ;)
tho the code typesetting doesn't work
also @astrofrog what browser?
Thomas Robitaille
@astrofrog
May 02 2014 11:51
safari
@keflavich
May 02 2014 11:57
@Cadair thanks, same as github, makes sense
Andy Casey
@andycasey
May 02 2014 12:12
ah, much better
are the chat sessions logged somewhere? it's good to get some context (and chat history) when joining but surely there's a limit on the number of rows that appear upon joining
Thomas Robitaille
@astrofrog
May 02 2014 12:13
good question
May 02 2014 13:15
Ooo, shiny
Thomas Robitaille
@astrofrog
May 02 2014 13:39
It would be cool to be able to log this chat as markdown wiki pages, then we could provide an archive on the wiki
jehturner
@jehturner
May 02 2014 14:06
By golly, CPU 164%!
Thomas Robitaille
@astrofrog
May 02 2014 14:07
@jehturner - ah, so it's not just me!
seems to go down after a while
@keflavich
May 02 2014 14:07
I'm on Chrome, and not seeing anything abnormal
jehturner
@jehturner
May 02 2014 14:07
May 02 2014 14:21
I'm also on Chrome, only hitting 4% CPU
Brigitta Sipocz
@bsipocz
May 02 2014 14:26
So far it looks OK with Firefox, sometimes there's a spike but that may come from another tab
Thomas Robitaille
@astrofrog
May 02 2014 15:29
@taldcroft - is astropy/astropy#2180 good to merge?
Everyone: if you have ideas of things that should be added to the 0.4 what's new, please list them here: astropy/astropy#2416
Erik Tollerud
@eteq
May 02 2014 15:41
hello team
Thomas Robitaille
@astrofrog
May 02 2014 15:44
hola @eteq
Erik Tollerud
@eteq
May 02 2014 15:44
Oh, boy, new toys! :clap:
Thomas Robitaille
@astrofrog
May 02 2014 15:44

# yay!

:wink:
Erik Tollerud
@eteq
May 02 2014 15:46
I didn’t quite follow what @Cadair said above: this is public, or private?
Thomas Robitaille
@astrofrog
May 02 2014 15:48
well, it's public to anyone with a GH account
Erik Tollerud
@eteq
May 02 2014 15:49
ok, gotcha
Thomas Robitaille
@astrofrog
May 02 2014 16:05
Does anyone have any objections to merging astropy/astropy#2094 if Travis passes? (the implementation of the new configuration system) cc @eteq
Erik Tollerud
@eteq
May 02 2014 16:13
Looks good to me!
Also, the “read by” feature here is pretty useful
Stuart Mumford
May 02 2014 16:15
@astrofrog, merge all the things!
hey @eteq
Erik Tollerud
@eteq
May 02 2014 16:15
Stuart Mumford
May 02 2014 16:16
eteq lol
how is the APE5 stuff?
Thomas Robitaille
@astrofrog
May 02 2014 16:17
Oh yeah, 'read by' is cool
no excuse for not replying :-p
Stuart Mumford
May 02 2014 16:20
does it show read by for me?
I am on IRC
Thomas Robitaille
@astrofrog
May 02 2014 16:26
I'm signing off for a few hours, should be back online later!
:wave:
Erik Tollerud
@eteq
May 02 2014 17:24
Oops, missed your message, @Cadair - apparently it doesn’t send me a notification if you don’t actually “@“ me. APE5 stuff is on course - it could be PRed now, actually, but I want to try to get some final stuff cleaned up
Ghost
@ghost~53638049048862e761fa0380
May 02 2014 17:35
Looks like Travis is stuck, I have two PR's that have been sitting for an hour with no tests even started running.
Erik Tollerud
@eteq
May 02 2014 17:36
https://travis-ci.org/astropy/astropy/builds/24277936 seems to be running… I think it’s just overloaded?
Ghost
@ghost~53638049048862e761fa0380
May 02 2014 17:37
Erik Tollerud
@eteq
May 02 2014 17:38
yeah, that makes sense: it’s later in the queue than 5727, which is what’s running right now
Ghost
@ghost~53638049048862e761fa0380
May 02 2014 17:38
I guess it's the feature-freeze rush.
Erik Tollerud
@eteq
May 02 2014 17:38
yep
I wish we could just buy extra build time on travis when we’re up against the wall...
but the good news is that the feature freeze still allows PRs “ready for final review” to be merged later as long as there aren’t major revisions or something
Ghost
@ghost~53638049048862e761fa0380
May 02 2014 17:39
That would be good. Doesn't the Astropy project have some small amount of cash via numfocus?
Erik Tollerud
@eteq
May 02 2014 17:40
Yep, so that would be perfect for something like this
but I don’t think travis does that...
maybe I’ll put in a feature request
Ghost
@ghost~53638049048862e761fa0380
May 02 2014 17:41
Humm, OK, I figured that was a way that they monetize the project.
Erik Tollerud
@eteq
May 02 2014 17:46
It’s possible… But I think they charge for private repo travis'
Also, @taldcroft , eteq/astropy#22 looks good except for one comment I left
Ghost
@ghost~53638049048862e761fa0380
May 02 2014 18:45
@eteq - pushed up a commit on the to-string branch.
Erik Tollerud
@eteq
May 02 2014 18:45
got it - I like it, but it needs a rebase because of something I must have changed recently
Ghost
@ghost~53638049048862e761fa0380
May 02 2014 18:56
rebased.
rt_args = 'frame0,frame1,equinox0,equinox1,obstime0,obstime1'
@pytest.mark.parametrize(rt_args, rt_sets)
def test_round_tripping(frame0, frame1, equinox0, equinox1, obstime0, obstime1):
Do you know why doing this pytest.mark.parametrize made the tests run so slowly? It now takes like a minute to get through that one test_round_tripping. BTW, stepping away for a few now.
Erik Tollerud
@eteq
May 02 2014 19:01
For me it sped things up…
Although now it actually runs all of them even if they fail
igg too for ~1 hr, but will be havk
*back
May 02 2014 19:35
@eteq you there?
or @taldcroft
Thomas Robitaille
@astrofrog
May 02 2014 20:03
I'm back
btw - for info, this chat is public even when not logged inp
in
May 02 2014 20:11
cool
Ghost
@ghost~53638049048862e761fa0380
May 02 2014 20:22

@eteq - here is a tabular summary of transform success and timing: https://gist.githubusercontent.com/taldcroft/356d04523a53126aa494/raw/coord_transforms.dat

Transform times are in millisec, some are over 20 msec.
May 02 2014 20:26
interesting
Ghost
@ghost~53638049048862e761fa0380
May 02 2014 20:33
BTW most of that must be setting up the transforms, since if I make the coordinates be 1000 elements long then the times go up to 20 to 60 msec.
Erik Tollerud
@eteq
May 02 2014 20:41
Yeah, that’s similar to the old ones, I think
A lot of that may be the quantity initializer
Ghost
@ghost~53638049048862e761fa0380
May 02 2014 20:42
I'm looking at the fk4 <=> fk4noeterms fails now to see what's happening.
Erik Tollerud
@eteq
May 02 2014 20:42
but thanks for doing it! I’m still trying to work out the 6 failures - it has to do with how the FK4<-> FK4NoE transforms pass on their obstime… it has to be handled manually because the E-terms are a weird thing, and probably there’s just some bruied typo
ah, ok, awesome! let me know what you find, then - I’m working on getting the matching stuff in.
and @adrn just put up a PR for from_name, so after that we’re pretty much there
Ghost
@ghost~53638049048862e761fa0380
May 02 2014 20:43
I was just modifying the code to include exactly what attrs are being used in each transform to be able to reproduce outside the test script.
Erik Tollerud
@eteq
May 02 2014 20:45
@adrn - that’s what we assumed before, right? That simbad always gives out ICRS?
May 02 2014 20:47
that's right
and we specifically pull out the coordinate string that starts with J2000
oh, sorry, no it just starts with J
Erik Tollerud
@eteq
May 02 2014 20:48
ok, cool - just quick checking a few things and then I’ll merge.
Actually, @adrn, do you think you have some time now to try to get the match_* stuff to work on SkyCoord? If so, I might take a shot at trying to “dynamicize” the SkyCoord (replacing the constants at the top so that user coordinates get pciked up on the fly), but if not, I can work on the matching stuff
May 02 2014 20:57
I could have a go at it, but not for another hour or two
Erik Tollerud
@eteq
May 02 2014 20:58
ok, let’s re-evaluate then, then
May 02 2014 20:58
cool
Ghost
@ghost~53638049048862e761fa0380
May 02 2014 20:58
OK, now I understand what you meant by "it has to do with how the FK4<-> FK4NoE transforms pass on their obstime…", but I can't fix it.
I know what to do for the dynamicizing, so I can do that quick.
Erik Tollerud
@eteq
May 02 2014 20:59
ok, that sounds perfect
give me a sec to push up some small changes that might conflict, @taldcroft
and then I’ll continue on the match_* stuff
Ghost
@ghost~53638049048862e761fa0380
May 02 2014 20:59
ok.
Erik Tollerud
@eteq
May 02 2014 21:07
ok, done
Thomas Robitaille
@astrofrog
May 02 2014 21:07
APE5 is going to be an epic pull request!
Erik Tollerud
@eteq
May 02 2014 21:08
indeed!
Thomas Robitaille
@astrofrog
May 02 2014 21:08
Erik Tollerud
@eteq
May 02 2014 21:08
and you, @astrofrog, and @Cadair , and @mhvk :wink:
Thomas Robitaille
@astrofrog
May 02 2014 21:09
we did the easy stuff :)
Ghost
@ghost~53638049048862e761fa0380
May 02 2014 21:12
epic indeed.
@eteq - did you just merge the PR you were mentioning?
Erik Tollerud
@eteq
May 02 2014 21:13
yep, @taldcroft - and also @adrn’s from_name stuff
Ghost
@ghost~53638049048862e761fa0380
May 02 2014 21:13
I mean the "small changes". I'm ready to go with the dynamicizing.
copy.
May 02 2014 21:21
I wonder if L$\alpha$tex works
Erik Tollerud
@eteq
May 02 2014 21:22
hah… give ‘em time?
maybe they have an issue tracker...
May 02 2014 21:22
Erik Tollerud
@eteq
May 02 2014 21:23
@taldcroft : I’m noticing some unusual behavior:
>>> SkyCoord(1*u.deg, 2*u.deg,frame='icrs').data
<UnitSphericalRepresentation lon=1.0 deg, lat=2.0 deg>
>>> coo.SkyCoord(1*u.deg, 2*u.deg,frame='icrs').data.distance
AttributeError: 'UnitSphericalRepresentation' object has no attribute 'distance'
>>> coo.SkyCoord(1*u.deg, 2*u.deg,frame='icrs').distance
<Quantity 1.0>
Ghost
@ghost~53638049048862e761fa0380
May 02 2014 21:24
In both cases it's passing through to self._coord.<attr>
Are those not expected values?
At least it should be doing that.
Erik Tollerud
@eteq
May 02 2014 21:25
ahhhh, I see - I thought you were going straight to the representation
it’s fine then - I was just confused
Ghost
@ghost~53638049048862e761fa0380
May 02 2014 21:25
OK. (whew!)
Just pushed the dynamic PR.
Thomas Robitaille
@astrofrog
May 02 2014 21:27
@eteq - just to double check, we can merge astropy/astropy#2094 as soon as it's rebased and passed, right? It's a very delicate PR in that it conflicts with quite a few others...
Erik Tollerud
@eteq
May 02 2014 21:29
As far as I’m concerned, yes. Hope it doesn’t doo too much with APE5?
It shouldn’t because no config stuff is there, I think, @astrofrog
Thomas Robitaille
@astrofrog
May 02 2014 21:30
I'm hoping it won't conflict either, fingers crossed
Erik Tollerud
@eteq
May 02 2014 21:33
@taldcroft - btw, you might want to take a look at the SkyCoord docstring I added, and finish it up with any initialization cases I missed
Ghost
@ghost~53638049048862e761fa0380
May 02 2014 21:46
Will do, but that will probably have to wait till the weekend, I'm being compelled to attend beer event on the roof...
One thing, I think the end of SkyCoord.transform_to() should be return self.__class__(..) not return SkyCoord(..) for subclassing.
Thomas Robitaille
@astrofrog
May 02 2014 21:48
Ack: looks like Shining Panda finally pulled the plug :(
The MacOS X instance is still running, I just need to publish over FTP instead
Thomas Robitaille
@astrofrog
May 02 2014 22:14
Note to everyone we should merge astropy/astropy#2094 before merging any other PR, since it's not easy to rebase, and @mdboom's already had to do it several times (thanks!)
Erik Tollerud
@eteq
May 02 2014 22:15
@taldcroft - that makes sense re: self.__class__
I can just change that in coordinates-ape5