These are chat archives for ipython/ipython

23rd
Jun 2015
Kyle Kelley
@rgbkrk
Jun 23 2015 01:19 UTC
Weee, deploying the new versions of the docker images knocked out tmpnb for a moment there
Hopefully that will be back up soon
I did the wrong thing...
Douglas La Rocca
@douglas-larocca
Jun 23 2015 07:32 UTC
hello all, I wrote a quick %%futurize magic to do fast 2to3 conversions in the notebook: https://github.com/douglas-larocca/ipyfuturize
Spencer Lyon
@sglyon
Jun 23 2015 14:25 UTC
Quick question. Is there a way to have the --to markdown nbconverter use the language key in the metadata and wrap input cells in fenced code blocks with that language?
Spencer Lyon
@sglyon
Jun 23 2015 14:50 UTC
(if this is the wrong place to ask the question please let me know. I’d be happy to re-post elsewhere)
Kyle Kelley
@rgbkrk
Jun 23 2015 15:35 UTC
That would be great to see @spencerlyon2!
Sylvain Corlay
@SylvainCorlay
Jun 23 2015 15:35 UTC
@spencerlyon2 welcome to the gitter chat
Kyle Kelley
@rgbkrk
Jun 23 2015 15:35 UTC
I haven't used the --to markdown option, maybe @rossant has ideas
Spencer Lyon
@sglyon
Jun 23 2015 15:35 UTC
@SylvainCorlay thanks, first time on gitter
(in any repo)
Sylvain Corlay
@SylvainCorlay
Jun 23 2015 15:35 UTC
nice to see you here
Spencer Lyon
@sglyon
Jun 23 2015 15:36 UTC
Likewise. @rgbkrk I might have time to take a stab at this sometime this week, but I’d need to be pointed to where the source for the markdown converter lives.
Kyle Kelley
@rgbkrk
Jun 23 2015 15:36 UTC
and here's the markdown exporter
Most important is the markdown filter
Spencer Lyon
@sglyon
Jun 23 2015 15:38 UTC
Thanks
Cyrille Rossant
@rossant
Jun 23 2015 15:40 UTC
@spencerlyon2 I am not familiar with nbconvert -- I made an alternative converter, but it doesn't support what you want. There is a pending PR by @bollwyvl that introduces cell metadata. Once it is merged I imagine it would be possible to implement it. That's certainly a nice feature to have
Jonathan Frederic
@jdfreder
Jun 23 2015 15:43 UTC
@spencerlyon2 you'll want to take a look at the null.tpl, skeleton file. It will give you a good idea of how the document is structured. Without that, it can be hard to understand the markdown template.
Spencer Lyon
@sglyon
Jun 23 2015 15:43 UTC
@rossant that’s good to know. Thanks for the heads up.
Jonathan Frederic
@jdfreder
Jun 23 2015 15:44 UTC
Spencer Lyon
@sglyon
Jun 23 2015 15:44 UTC
@jdfreder Thanks, I was just looking for it.
Jonathan Frederic
@jdfreder
Jun 23 2015 15:44 UTC
Spencer Lyon
@sglyon
Jun 23 2015 15:45 UTC
After a very quick scan, it looks like this might be doable by changing this line
Jonathan Frederic
@jdfreder
Jun 23 2015 15:46 UTC
yup
I think (this is off the top of my head) the language is cell.cell_type.
Spencer Lyon
@sglyon
Jun 23 2015 15:46 UTC

Something like

{% block input %}
```{{ cell.language }}
{{ cell.source }}
```
{% endblock input %}

might work, assuming cell.language exists

Jonathan Frederic
@jdfreder
Jun 23 2015 15:47 UTC
mmm, no I think you need:
^ yeah that
cell.cell_type is not what you want, sorry
and I don't think cell.language exists...
Spencer Lyon
@sglyon
Jun 23 2015 15:49 UTC
no problem, that would have been too easy :smirk:
Jonathan Frederic
@jdfreder
Jun 23 2015 15:50 UTC
nb.metadata.kernelspec.language
:)
I think that's it
Spencer Lyon
@sglyon
Jun 23 2015 15:50 UTC
Is the nb variable in scope (available) in that template?
Weird, didn't know posting a gist posts a picture of the account's gravatar...
Anyways, that's a link to a notebook file, with the appropriate line highlighted
Spencer Lyon
@sglyon
Jun 23 2015 15:52 UTC
haha that is weird. I can’t quite see (more accurate, can’t quite remember how scoping works in jinja2 templates) from null.tpl if nb will still be available in markdown.tpl.
Jonathan Frederic
@jdfreder
Jun 23 2015 15:53 UTC
It should be because of the trailing scoped keywords.
If it isn't, chances are a scoped keyword is missing somewhere.
Spencer Lyon
@sglyon
Jun 23 2015 15:54 UTC
Well, there’s one way to find out :)
Jonathan Frederic
@jdfreder
Jun 23 2015 15:54 UTC
At least, that's what I remember. It's been a year or two..
Spencer Lyon
@sglyon
Jun 23 2015 15:56 UTC
Now time for me to show my ignorance…. I haven’t hacked on ipython stuff since the big split. How do I install the jupyter command so I can run jupyter nbconvert …?
Looking here right now
Jason Grout
@jasongrout
Jun 23 2015 16:00 UTC
hey, Spencer, good to see you!
Spencer Lyon
@sglyon
Jun 23 2015 16:00 UTC
@jasongrout thanks. Crazy how quick it is to get feedback on here!
Sylvain Corlay
@SylvainCorlay
Jun 23 2015 16:01 UTC
Jason was asking is it The Spencer Lyon.
Spencer Lyon
@sglyon
Jun 23 2015 16:03 UTC
Haha… not sure I deserve the article in front of my name (captialized nontheless)!
Jason Grout
@jasongrout
Jun 23 2015 16:04 UTC
Yeah, you should see what they did with my name the other day
Spencer Lyon
@sglyon
Jun 23 2015 16:05 UTC
Are there instructions somewhere for installing the post-split jupyter toolchain.
Jason Grout
@jasongrout
Jun 23 2015 16:05 UTC
you wouldn't complain about the capital letters for your name, then :)
Jonathan Frederic
@jdfreder
Jun 23 2015 16:05 UTC
@spencerlyon2 I've been using this:
Jason Grout
@jasongrout
Jun 23 2015 16:05 UTC
that looks roughly like my installation procedure.
Jonathan Frederic
@jdfreder
Jun 23 2015 16:05 UTC
It's @minrk 's setup script, it works well for me.
Of course I have a modified one under my name too, but it's personalized to my strange tendencies. So I don't recommend starting from it.
Spencer Lyon
@sglyon
Jun 23 2015 16:08 UTC
Sounds good, I’ll start with that. Cloning away now
Jason Grout
@jasongrout
Jun 23 2015 16:08 UTC
hey, where are the widgets in that gist?
:)
Jonathan Frederic
@jdfreder
Jun 23 2015 16:09 UTC
They are too special to be cloned automatically
Sylvain Corlay
@SylvainCorlay
Jun 23 2015 16:10 UTC
thanks for the script
I did this manually a couple of times
Jonathan Frederic
@jdfreder
Jun 23 2015 16:10 UTC
Ouch
Spencer Lyon
@sglyon
Jun 23 2015 16:16 UTC
Wow, @jdfreder that was easy. It worked first try :)
Jason Grout
@jasongrout
Jun 23 2015 16:16 UTC
@spencerlyon2, add ipython/ipywidgets to the end of that list to get the widgets too
they were split out after the script was made, probably
and then you'll need to install the nbextension for the widgets
Spencer Lyon
@sglyon
Jun 23 2015 16:18 UTC
Ok will do, thanks. Where is the extension?
Jason Grout
@jasongrout
Jun 23 2015 16:19 UTC
python -m ipywidgets.install --enable --user --symlink is what we use
Spencer Lyon
@sglyon
Jun 23 2015 16:21 UTC
Excellent, added to the bottom of the script
nbconvert|fenced-input⚡ ⇒ jupyter console
[ZMQTerminalIPythonApp] CRITICAL | Could not find kernel python
Do I need to have the python kernelspec in /usr/local/share/jupyter?
Jonathan Frederic
@jdfreder
Jun 23 2015 16:38 UTC
Hmm, strange the Python kernel spec should install for you. Did the output of the installation script have any errors around the ipython\ipython install?
Min RK
@minrk
Jun 23 2015 16:38 UTC
It shouldn't be a 'python' kernel, but a 'python2' or 'python3' kernel.
Spencer Lyon
@sglyon
Jun 23 2015 16:42 UTC
Thanks for the help. Sumbitted jupyter/nbconvert#49 with tests.
Fernando Perez
@fperez
Jun 23 2015 17:02 UTC
Dev meeting URL?
Cameron Oelsen
@cameronoelsen
Jun 23 2015 17:05 UTC
We are waiting on the URL as well :shipit:
Jason Grout
@jasongrout
Jun 23 2015 17:07 UTC
Do you have a youtube URL?
Min RK
@minrk
Jun 23 2015 17:09 UTC
Someone else can paste it. Hangouts won't let me copy it.
It should be on the YouTube page.
Jason Grout
@jasongrout
Jun 23 2015 17:09 UTC
Thanks!
Fernando's ghost is echoing
Jason Grout
@jasongrout
Jun 23 2015 17:35 UTC
I think the idea is that you ask once, when you refresh a page, what comms are valid on the kernel-side
perhaps a more general "I just reloaded, what is your state" request/reply message is appropriate
that would include valid comms, but may include other things too
it might include things like widget model state, etc?
Min RK
@minrk
Jun 23 2015 17:40 UTC
@jasongrout you've convinced me we shouldn't merge it for 4.0.
Matthias Bussonnier
@Carreau
Jun 23 2015 17:44 UTC
Btw BIDS, much better audio now.
Jonathan Frederic
@jdfreder
Jun 23 2015 17:45 UTC
Correction, 4 PRs
Brian E. Granger
@ellisonbg
Jun 23 2015 17:50 UTC
Sorry, G+ crashed my Mac :(
Fernando Perez
@fperez
Jun 23 2015 17:50 UTC
ouch
Brian E. Granger
@ellisonbg
Jun 23 2015 17:50 UTC
I think I am going to boycot our dev meetings until we move away from G+
So sick of it...
Rejoining...
Jason Grout
@jasongrout
Jun 23 2015 17:50 UTC
how bad are the bugs that we are fixing with the new message?
I'm +0.5 for holding off until 4.0, unless it really fixes a serious bug that shows up in practice
(as in, shows up in other widgets out there too...)
(blame those guys that gave me commit bits, so now I am a lot more conservative about stuff going in :)
Min RK
@minrk
Jun 23 2015 17:54 UTC
Our secret plan all along!
Brian E. Granger
@ellisonbg
Jun 23 2015 17:55 UTC
just for kicks - I am probably +0.1 on moving forward with this - but I agree with Fernando that patience typically is a good idea
Jason Grout
@jasongrout
Jun 23 2015 17:55 UTC
@SylvainCorlay - yes, it's called a PR :)
can't the widget manager send a message to the other widget manager asking about valid comms?
or you could open a new comm specifically to request available comms
Fernando Perez
@fperez
Jun 23 2015 17:59 UTC
xmsg_
Matthias Bussonnier
@Carreau
Jun 23 2015 18:03 UTC
udp on com on tcp/ip on udp yeahhhh turtles !
Jason Grout
@jasongrout
Jun 23 2015 18:09 UTC
when is the eta for releasing 4.0? Next week?
perhaps in the future, we should establish beforehand hard freeze dates
Sylvain Corlay
@SylvainCorlay
Jun 23 2015 18:10 UTC
the comm_list could have been documented as experimental
Thomas Kluyver
@takluyver
Jun 23 2015 18:10 UTC
we've roughly said we'll aim for 4.0 releases by scipy
that will inevitably slip a bit, but probably not by too much, I think
Jason Grout
@jasongrout
Jun 23 2015 18:11 UTC
not just a larger team, but many more repos
Thomas Kluyver
@takluyver
Jun 23 2015 18:11 UTC
the next point on the agenda is about the release status... if we ever get to it ;-)
Jason Grout
@jasongrout
Jun 23 2015 18:11 UTC
:)
Sylvain Corlay
@SylvainCorlay
Jun 23 2015 18:12 UTC
The split was supposed to help things go a bit faster by separating concerns...
Fernando Perez
@fperez
Jun 23 2015 18:12 UTC
We're just disentangling the entire hairball, hopefully once it's many smaller hairballs it will be true...
Sylvain Corlay
@SylvainCorlay
Jun 23 2015 18:17 UTC
Being backward compatible with the hack of having a special comm with a fixed comm id might prove difficult...
While there were other major changes in the message spec for existing messages.
Jason Grout
@jasongrout
Jun 23 2015 18:19 UTC
Please, not dozens of more repos!
Thomas Kluyver
@takluyver
Jun 23 2015 18:19 UTC
yeah, that does not sound like fun
atom as an example: atom was designed from the ground up to be a collection of tiny packages. IPython wasn't, and there's a significant cost to re-organising the code
Jason Grout
@jasongrout
Jun 23 2015 18:20 UTC
I do think that splitting off the frontend js/html of the repo is a good idea. I'm doing that right now with my phosphor-notebook repo
Thomas Kluyver
@takluyver
Jun 23 2015 18:20 UTC
I think we should be focussing on user focussed features for a while after 4.0, not more reorganisation
Kyle Kelley
@rgbkrk
Jun 23 2015 18:20 UTC
It does make those experiments easier
Thomas Kluyver
@takluyver
Jun 23 2015 18:22 UTC
I feel like we're getting very navel-gazing about things that might benefit us as developers, and neglecting things that actually improve what users see
Jason Grout
@jasongrout
Jun 23 2015 18:22 UTC
Think of the children^Wusers!
Thomas Kluyver
@takluyver
Jun 23 2015 18:22 UTC
like selecting multiple cells - we've been talking about that for ages, but still not actually implemented it
and it's a real pain point
Fernando Perez
@fperez
Jun 23 2015 18:23 UTC
You have a point!
Thomas Kluyver
@takluyver
Jun 23 2015 18:26 UTC
If we can actually solve packaging, that would certainly be amazing. But it's ambitious, to say the least.
Jason Grout
@jasongrout
Jun 23 2015 18:26 UTC
let's do it before 4.0! :)
hehe...design by snapchat
Jason Grout
@jasongrout
Jun 23 2015 18:32 UTC
Let's definitely talk also at scipy
about frontend stuff
Thomas Kluyver
@takluyver
Jun 23 2015 18:36 UTC
I would at least demote the chat link below Stackoverflow
Jason Grout
@jasongrout
Jun 23 2015 18:37 UTC
+1 for smaller numbers of rooms
+1 for smaller rooms
can we register a bot on the "inactive" gitter channels to notify people to move the conversation?
Kyle Kelley
@rgbkrk
Jun 23 2015 18:39 UTC
vidyo
We swear by it at Rackspace
Matthias Bussonnier
@Carreau
Jun 23 2015 18:40 UTC
Jason Grout
@jasongrout
Jun 23 2015 18:43 UTC
do we need videoconferencing?
or are phone conversations good enough?
Jason Grout
@jasongrout
Jun 23 2015 18:54 UTC
By the way, if we don't really release next week, then I change my comm-list message vote to +1 merge it. :)
Because if we don't release next week, likely it'll be another month or two - plenty of time to test drive and iterate this! :)
Sylvain Corlay
@SylvainCorlay
Jun 23 2015 18:56 UTC
Saying that we release next week is IPython's way of calling a feature freeze
cf last january and 3.0
Jason Grout
@jasongrout
Jun 23 2015 18:56 UTC
(and the two months or so leading up to 3.0 :)
Cameron Oelsen
@cameronoelsen
Jun 23 2015 18:57 UTC
What was the website that we used to print for the previous IPython 1x1 stickers?
Cameron Oelsen
@cameronoelsen
Jun 23 2015 19:03 UTC
@Carreau @minrk
if you want to use them, let me know, I have a discount code
Cameron Oelsen
@cameronoelsen
Jun 23 2015 19:04 UTC
Thanks will do @takluyver
Thomas Kluyver
@takluyver
Jun 23 2015 19:05 UTC
btw, all info on previous stickers etc. we've ordered is here: https://github.com/ipython/marketing/blob/master/past_orders.rst
it would be good if you can update that when you've ordered this batch.
Brian E. Granger
@ellisonbg
Jun 23 2015 23:32 UTC
@takluyver we are using the jupyter/design repo for this, but I will definitely put theinformation there now