These are chat archives for ipython/ipython

5th
Nov 2014
Scott Sanderson
@ssanderson
Nov 05 2014 00:00
err, rather
ipython/ipython#6858
epifanio
@epifanio
Nov 05 2014 03:55
hi, is it possible to change the widget style when they are generated directly by interact ?
i’m looking at Widget Styling.ipynb
but in my case the widgets are “autogenerated” by : w = interactive(solve_ffbutter, freq1=(0.005,0.3,0.00001), .. .. )
i’ve to many parameters .. so they fill the page and i can’t see the graph anymore
so i was looking for a way to group them in a sort of layout
epifanio
@epifanio
Nov 05 2014 04:13
at the moment the widgets examples are broken on master
Álvaro Justen
@turicas
Nov 05 2014 13:40
@fperez: about debuggers - there is one created by @rodsenra (he'll be on PythonBrasil tomorrow), it may be integrated to IPython notebook debugger: https://github.com/rodsenra/pode
Álvaro Justen
@turicas
Nov 05 2014 14:04
@fperez, (I'm in the tutorial) yes, you can rewrite text console history (at least on UNIX) - try: import sys; sys.stdout.write('\b' * 500); sys.stdout.flush()
Jason Grout
@jasongrout
Nov 05 2014 16:59
@ssanderson - thanks! quick question---how do you override something like the notebook.html template, but then fall back to the IPython notebook.html template?
Jason Grout
@jasongrout
Nov 05 2014 17:11
ah, I see @Carreau posted about this in https://groups.google.com/forum/#!topic/sage-devel/j1UUQADct3s
Jonathan Frederic
@jdfreder
Nov 05 2014 17:16
Hey @/all , are we having a supplemental meeting today?
Thomas Kluyver
@takluyver
Nov 05 2014 17:17
I don't think so - it's usually Tuesday, isn't it?
Kyle Kelley
@rgbkrk
Nov 05 2014 17:17
We can supplementally high five
Thomas Kluyver
@takluyver
Nov 05 2014 17:18
:hand:
Jonathan Frederic
@jdfreder
Nov 05 2014 17:18
Oh whoops
looks like I got my days scrambled
was there a meeting yesterday? I don't remember anyone pinging about it so I didn't attend if there was
Thomas Kluyver
@takluyver
Nov 05 2014 17:18
no, there wasn't
at least, not that I knew about
Jonathan Frederic
@jdfreder
Nov 05 2014 17:20
@epifanio you can child them to another parent, one that does layout differently.
With the widgets, the parents know about the children, generally not the other way around
Also make sure you are using the correct examples
if you're running master, use the examples in master. If you're running 2.x, use the examples in 2.x
Also if you are running on a commit more than a month old, there's a chance you have a copy where the examples fell out of sync with the widgets themselves
Thomas Kluyver
@takluyver
Nov 05 2014 17:49
awesome
let's start tweeting
Kyle Kelley
@rgbkrk
Nov 05 2014 17:50
I think we should wait a teensy bit
Thomas Kluyver
@takluyver
Nov 05 2014 17:50
spoilsport ;-)
Kyle Kelley
@rgbkrk
Nov 05 2014 17:50
To hit various time periods
Otherwise its all of us at once
Thomas Kluyver
@takluyver
Nov 05 2014 17:51
OK, well, let us know when's best to tweet
Kyle Kelley
@rgbkrk
Nov 05 2014 17:52
Let me consult my public relations department
(I'm kidding)
You can tweet anytime you want. ;)
Thomas Kluyver
@takluyver
Nov 05 2014 17:54
I think it's worth co-ordinating a bit
this is significant enough that we should try to ensure it picks up moment on the 'twittersphere'
I've just had a play around with it, it's pretty nice
Thomas Kluyver
@takluyver
Nov 05 2014 18:44
@rgbkrk Min just started a 24-node cluster on the nature demo server
Min RK
@minrk
Nov 05 2014 18:46
didn't seem to start. Hopefully that's not too bad
Kyle Kelley
@rgbkrk
Nov 05 2014 19:01
Gee thanks min
Min RK
@minrk
Nov 05 2014 19:02
I didn't mean to, I just clicked the 'start' button on the cluster, which lets the controller pick the default number of engines
apparently every container can see all 24 CPUs
Kyle Kelley
@rgbkrk
Nov 05 2014 19:02
yeah
Min RK
@minrk
Nov 05 2014 19:02
I didn't expect that
Kyle Kelley
@rgbkrk
Nov 05 2014 19:02
I don't think I turned on resource limiting for CPUs
...
Min RK
@minrk
Nov 05 2014 19:03
how many tmpnb instances?
Kyle Kelley
@rgbkrk
Nov 05 2014 19:03
Even if I did though it would think there were 24 CPUs
only one
Min RK
@minrk
Nov 05 2014 19:03
ok
Kyle Kelley
@rgbkrk
Nov 05 2014 19:03
512 containers
Min RK
@minrk
Nov 05 2014 19:03
That's what I'm watching
Kyle Kelley
@rgbkrk
Nov 05 2014 19:03
The memory is limited though
Min RK
@minrk
Nov 05 2014 19:03
is the full container stats thing running?
Kyle Kelley
@rgbkrk
Nov 05 2014 19:03
No
I didn't put cadvisor on it
Min RK
@minrk
Nov 05 2014 19:03
ok
It'd be interesting to see the culling time
What's the idle value?
Do you have an LB, so you can drop in another instance if you need to?
Jason Grout
@jasongrout
Nov 05 2014 19:04
so the nature article is live?
Kyle Kelley
@rgbkrk
Nov 05 2014 19:05
cull timeout is 120 seconds
Oh you want the cull timeout in the stats endpoint?
I think that'd be good too
Found out why tmpnb would fall flat with larger numbers of containers
The hokey async client was choking Docker
Min RK
@minrk
Nov 05 2014 19:06
weird
Kyle Kelley
@rgbkrk
Nov 05 2014 19:06
Switching to a real async client wouldn't help anything
Min RK
@minrk
Nov 05 2014 19:06
idle timeout is 2 minutes?
Kyle Kelley
@rgbkrk
Nov 05 2014 19:06
I took the workers down to 8 and now I can boot 1000 instances for a tmpnb
yeah
DM has details
I'd prefer not to put cadvisor up publicly though
since it runs somewhat privileged
Scott Sanderson
@ssanderson
Nov 05 2014 20:21
@rgbkrk that's pretty awesome; I need to take a closer look through the tmpnb codebase