These are chat archives for ipython/ipython

27th
Feb 2015
Jonathan Frederic
@jdfreder
Feb 27 2015 01:55
@minrk took forever, but here's the results of my pass on examples/Notebook - ipython/ipython#7886
Min RK
@minrk
Feb 27 2015 03:39
@jdfreder great, I'm looking at the IPython Kernel ones
michigraber
@michigraber
Feb 27 2015 14:14
hey guys! new here .. have a question regarding installation of 3.0.0rc1: i work for an astronomy collaboration with own package management system (eups). we would like to have said ipython version distributed through our system, independent of pip. however, trying to install the package through $ python setup.py install results in an error, apparently due to missing submodule managed packages .. is there an easy way to resolve this issue, without pip? is this going to be an ongoing issue also for the final release of the 3.0.0 version? thx!!
Jessica B. Hamrick
@jhamrick
Feb 27 2015 16:42
If it’s missing submodules, then you probably need to do git submodule init && git submodule update first (you can also clone the repository initially with --recursive)
Min RK
@minrk
Feb 27 2015 17:07
@michigraber did you grab a tarball from GitHub instead of using git clone or PyPI? If so, git archive omits submodules, so it won't work. When we make sdists of real releases, the submodules are included, so this won't be a problem once we make a real release.
Min RK
@minrk
Feb 27 2015 17:15
We've asked GitHub for the ability to disable tarball downloads based on git-archive, since they are of no use to anyone, and highly prominant links on the repo.
michigraber
@michigraber
Feb 27 2015 19:03
@minrk @jhamrick thanks! yes, it is complaining about missing submodules and yes, I did grab the tarball from the release section on github. if i understand you correctly the way to go would be to do a git clone from https://github.com/ipython/ipython.git, clean it from .git files and tar it for our package management system, right? would i still have to get the submodules separately by git submodule init && git submodule update?
Matthias Bussonnier
@Carreau
Feb 27 2015 19:09
@michigraber depends on how you clone.
if you clone --recursive, you should not need to
you can manually git submodule init/update
or run python setup.py submodule.
you can also get the tarball from PyPI
Which should be ok, only github one is wrong.
And which package management system are you using ?
michigraber
@michigraber
Feb 27 2015 19:12
@Carreau thanks! the package management system is a 'custom' one for our collaboration called eups
Jason Grout
@jasongrout
Feb 27 2015 19:29
Jonathan Frederic
@jdfreder
Feb 27 2015 19:33
Thanks!
I was hoping you or someone would read it :)
@jasongrout I can't seem to get the diff to render.
I wish I could see it , because the same doc is in IPython now: ipython/ipython#7870
Cedric GESTES
@cgestes
Feb 27 2015 20:54
how are we supposed to load the css associated to a widget ?
Jason Grout
@jasongrout
Feb 27 2015 21:01
@jdfreder, I think we ought to make require-less a default component to answer @ctaf42's question
Cedric GESTES
@cgestes
Feb 27 2015 21:02
and then _view_css attribute on widget ? :D
Jason Grout
@jasongrout
Feb 27 2015 21:02
no, I think the widget's javascript should require whatever less/css files it needs
so the css/less is injected in the page when the js is loaded by require
Min RK
@minrk
Feb 27 2015 21:05
@jasongrout that's probably a good idea, and probably will be more apparent once the widgets aren't part of style.min.css, since they won't be part of the notebook repo.
I think we'll find lots of things we need to improve (css, js loading, etc.) when the widgets are a separate package.
Jonathan Frederic
@jdfreder
Feb 27 2015 21:15
  • template loading too
Jason Grout
@jasongrout
Feb 27 2015 21:21
indeed, requre-less I think also includes/needs require-text
jdfreder @jdfreder wonders if he should rename all of his ipython-* repositories to jupyter-*.
Min RK
@minrk
Feb 27 2015 21:35
@jdfreder yes, of course!
(if they are Jupyter-related)
IPython is still a thing, so if it's Python-specific rather than frontend, IPython is appropriate.
Thomas Kluyver
@takluyver
Feb 27 2015 23:04
I am initiating the release process. If there are any last minute problems, please shout now.
Min RK
@minrk
Feb 27 2015 23:05
@takluyver do you want to merge the two outstanding PRs on the examples?
That's all that's marked for 3.0
Thomas Kluyver
@takluyver
Feb 27 2015 23:05
yep, let's get those in
Min RK
@minrk
Feb 27 2015 23:06
ok
you want to run the release steps?
Thomas Kluyver
@takluyver
Feb 27 2015 23:07
yep :)
Min RK
@minrk
Feb 27 2015 23:07
Make sure your setuptools, pip, and wheel packages are latest
Scott Sanderson
@ssanderson
Feb 27 2015 23:07
welp, gitter renders that much larger than expected...
Min RK
@minrk
Feb 27 2015 23:08
I don't think our release scripts mention that. When I did the beta, my py2.7 had a slightly old setuptools, which did different normalization for the version string.
Thomas Kluyver
@takluyver
Feb 27 2015 23:08
in the instructions, should PREV_RELEASE be 2.0.0 or 2.4.1? It's not clear what uses that
Min RK
@minrk
Feb 27 2015 23:08
I don't think PREV_RELEASE matters, but it would be 2.0.0
Jonathan Frederic
@jdfreder
Feb 27 2015 23:09
@ssanderson trust me, the largeness is appropriate in this case!
lol
Min RK
@minrk
Feb 27 2015 23:16
does the release process say to rebuild the git stats? They haven't changed much since the rc.
Thomas Kluyver
@takluyver
Feb 27 2015 23:16
yeah, it's just fetching them
is there a way to get only the changes since the RC?
Is it short enough that we should do it manually?
Min RK
@minrk
Feb 27 2015 23:17
there's probably a way, but it might take longer than running the script to figure it out
Thomas Kluyver
@takluyver
Feb 27 2015 23:18
actually, it does say "fetching GitHub stats since 2015-02-13 21:00:00 (tag: 3.0.0rc1, milestone: 3.0)"
not sure where it got the 3.0.0rc1 from..
Min RK
@minrk
Feb 27 2015 23:18
git describe
Thomas Kluyver
@takluyver
Feb 27 2015 23:18
well, that's clever
Min RK
@minrk
Feb 27 2015 23:18
but the slow part is getting PR info from GitHub
that's not affected by the since-tag
only the milestone
Thomas Kluyver
@takluyver
Feb 27 2015 23:19
Ah, so it has to get all the PRs
so I'll need to add to the issues section, but replace the PRs section from what's already checked in?
Min RK
@minrk
Feb 27 2015 23:21
It does all this to get the right numbers for the backport releases
Thomas Kluyver
@takluyver
Feb 27 2015 23:22
that doesn't take five minutes, though...I'm starting to wonder if it's stuck
oh, I see what you mean
Min RK
@minrk
Feb 27 2015 23:22
I mean the thing that's really expensive is only because we can't just look at the git log for author info on the backport releases
Thomas Kluyver
@takluyver
Feb 27 2015 23:23
mhm
Min RK
@minrk
Feb 27 2015 23:23
but that's not true for a mainline release
so we could save a bunch of time by skipping the author/git check when it's a major release
Thomas Kluyver
@takluyver
Feb 27 2015 23:24
it's probably not worth spending too much time on release tools just now, though, given that it's all going to change for the next major release.
Min RK
@minrk
Feb 27 2015 23:24
sure
oh: issues, pr counts are backwards
Thomas Kluyver
@takluyver
Feb 27 2015 23:27
alternatively... it was waiting for me to enter a password
hmm, I seem to be having some trouble with this script
perhaps it's not Python 3 ready?
Min RK
@minrk
Feb 27 2015 23:30
I just finished running it. I can push a commit of the results to master, if you just want to pull
Thomas Kluyver
@takluyver
Feb 27 2015 23:30
ok, go for it
Min RK
@minrk
Feb 27 2015 23:30
maybe it hit API limit?
Thomas Kluyver
@takluyver
Feb 27 2015 23:31
422 Client Error: Unprocessable Entity
Min RK
@minrk
Feb 27 2015 23:34
yeah, that's failing to create the auth token
so you probably hit the unauthenticated API rate limit
numbers pushed
Thomas Kluyver
@takluyver
Feb 27 2015 23:35
great, pulled
Min RK
@minrk
Feb 27 2015 23:35
Make sure to add today's date in whatsnew/version3.rst
Thomas Kluyver
@takluyver
Feb 27 2015 23:39
IPython.core.release has a comment saying "release.codename is deprecated in 2.0, will be removed in 3.0"
shall I remove it?
Min RK
@minrk
Feb 27 2015 23:43
sure
or save that for 3.1
in case there's some random bit of code that uses it elsewhere in IPython
though grep turns up nothing
Thomas Kluyver
@takluyver
Feb 27 2015 23:45
probably safest to leave it.
Min RK
@minrk
Feb 27 2015 23:46
sure
ha, people keep pushing that 1.0.0a1 tag back up
Min RK
@minrk
Feb 27 2015 23:53
I've removed it a whole bunch of times
Thomas Kluyver
@takluyver
Feb 27 2015 23:55
that's bizarre, I cloned this repo fresh on this laptop just the other day
oh FFS, I don't have ssh access to archive.ipython.org
it's made it to PyPI
Min RK
@minrk
Feb 27 2015 23:56
I can add you. Do you know which of your 6 GitHub keys you are using?
(I'll assume oldest)
try: ssh ipython@archive.ipython.org
Thomas Kluyver
@takluyver
Feb 27 2015 23:57
oldest is probably wrong, hang on
ssh-rsa AAAAB3NzaC1yc2EAAAADAQABAAABAQDdsPGQM7WWrW1VVLs/4yRDyXrTdYo68W+ItWzkTuIcHWUd8LkkDnFVqrmRWugJoGnheCqXMLQFJeRqOspnB1qsaLnCN2k9SJVHcSTEzWYX2UcUv6+7Es9Vua7b7Vzx8s7+GgvoS8F4S8BKMTEARHA15qaAJ+/RfUzKYCeb/CWw5SrVWtxL78egFS6EYY8ZNQvclCpxRtCCVn2W9kkN2DNLrjs8o5KGDsmOe9zZOg+oFavh30z1kv3QjaTV+OPl2ES4g3LqkX3fQS7Ix4yBQvmihPIF4MS1O8nQgLYXwNDl67heBv9JRl7p1pl1VTE+IKUTEBv7DBMXubH50L8CC/Jv takowl@gmail.com
Min RK
@minrk
Feb 27 2015 23:57
ok, try it now
the archive bits aren't a big deal, I can upload the files, if you want to start building the docs
Thomas Kluyver
@takluyver
Feb 27 2015 23:59
ok, working. Now I just need to fish out the relevant bits of the release script