These are chat archives for ipython/ipython

29th
Jan 2015
Jason Grout
@jasongrout
Jan 29 2015 00:08
Travis likes ipython/ipython#7625 :)
Thanks @minrk. I can finish my last project and go home now :)
Min RK
@minrk
Jan 29 2015 00:13
@jasongrout thanks for the fix!
Jason Grout
@jasongrout
Jan 29 2015 00:38
did something change recently with the styling for the menubar and menu header on a notebook?
Thomas Kluyver
@takluyver
Jan 29 2015 00:39
I think there was a minor adjustment to make the shadow appear below it
I don't remember anything else
but there have been a lot of changes recently
Jason Grout
@jasongrout
Jan 29 2015 00:48
thanks; I guess I'll bisect tomorrow to find where my issue is.
Jason Grout
@jasongrout
Jan 29 2015 01:04
well, it looks like minrk/ipython@fb558f1 causes us some problems. I'll look into it more tomorrow.
it does look like using translateZ(0) is something of a hack, though.
Min RK
@minrk
Jan 29 2015 01:10
Arg, sorry.
It's a common hack that solves a real problem, so removing it would require an alternative fix.
What does it cause a problem for?
Jason Grout
@jasongrout
Jan 29 2015 16:11
fixed position divs that try to overlay it.
Jason Grout
@jasongrout
Jan 29 2015 16:28
ah, I see better what's happening. Putting translateZ(0) on the #site makes #site act "a containing block for fixed positioned descendants." (http://www.w3.org/TR/css3-transforms/#transform-property). This means that fixed-position elements inside of #site no longer are positioned relative to the viewport, but instead relative to the #site element.
Jason Grout
@jasongrout
Jan 29 2015 16:43
@minrk - I've been playing with the translateZ(0) on #site being turned on and off, and looking at the paint rectangles in Chrome 40 and Firefox 33, and I don't see any difference in the repaints with having the translateZ(0) on or off.
Jason Grout
@jasongrout
Jan 29 2015 16:51
(both having translateZ(0) on and off seem fine to me on the commit just before you added it). Maybe browsers have gotten smarter now?
Jason Grout
@jasongrout
Jan 29 2015 16:57
oh, I guess that update was 6 days ago. I didn't test on Safari; was that the main problematic browser?
Min RK
@minrk
Jan 29 2015 17:09
Chrome was where I tested it.
Jason Grout
@jasongrout
Jan 29 2015 17:12
weird. I'll test again.
Jason Grout
@jasongrout
Jan 29 2015 17:21
yep; with stock ipython master, chrome 40 on windows, I see the same repainting rectangles scrolling a notebook page with or without the transform applied to #site
perhaps it has something to do with video cards and gpu acceleration; I don't know
Min RK
@minrk
Jan 29 2015 17:26
could be. I'll double check, and keep an eye on versions. That's super frustrating.
Jonathan Frederic
@jdfreder
Jan 29 2015 17:29
/@all are both meetings at 9:30 now? Are we having a supplementary meeting today?
@/all
Kyle Kelley
@rgbkrk
Jan 29 2015 17:31
afaict, yes
I arranged for us to chat with the PyImageSearch author
Jason Grout
@jasongrout
Jan 29 2015 17:32
hehe...you trying to keep up with upstream Chrome caused problems for us trying to keep up with upstream you, or something like that. And the world goes around...
Matthias Bussonnier
@Carreau
Jan 29 2015 17:32
Sont be There (in a train) and computer dead. Have fun.
Min RK
@minrk
Jan 29 2015 17:40
@jasongrout I can reproduce with Chrome 42 on a low-DPI screen, not on 41 with a high-DPI screen.
Jessica B. Hamrick
@jhamrick
Jan 29 2015 17:45
Sorry, won’t be able to join again today, have fun!
Jonathan Frederic
@jdfreder
Jan 29 2015 18:15
@/all https://developer.mozilla.org/en-US/docs/Web/API/Window.open can be used with a 'target' string.
Should be easy enough to use that
for the new notebook button
Min RK
@minrk
Jan 29 2015 18:16
@jdfreder that should do it, thanks
Thomas Kluyver
@takluyver
Jan 29 2015 18:18
not sure that it will - I think window.open() opens a new window unconditionally. But it shouldn't be too hard to have a branch in the JS that loads the URL in the same window.
Jonathan Frederic
@jdfreder
Jan 29 2015 18:21
seems to work on FF too
window.open('http://google.com', '_self');
Jason Grout
@jasongrout
Jan 29 2015 18:27
@minrk: do you see a noticeable difference on the chrome 42 with regards to lagging, etc? For now, I've just overridden that one bit of css; I'm curious what effects that may have
Min RK
@minrk
Jan 29 2015 18:27
@jasongrout it depends on the content of the notebook
Jason Grout
@jasongrout
Jan 29 2015 18:28
okayu
Jason Grout
@jasongrout
Jan 29 2015 20:56
@rgbkrk - did you ever track down what was going wrong with installing rzmq straight from github on Ubuntu 14.04, or are you still using your fork? I'm referring to the discussion at armstrtw/rzmq#12
Kyle Kelley
@rgbkrk
Jan 29 2015 20:57
@jasongrout I'm still just using my fork
I didn't want to track down how Ubuntu does its R build
It sounded like he was saying that the problem was in Ubuntu and that my patch would never be accepted
Jason Grout
@jasongrout
Jan 29 2015 21:01
yes, that's what it sounded like. Have you tried with 14.10?
Kyle Kelley
@rgbkrk
Jan 29 2015 21:02
Wasn't in the mood to solve a "political problem" at that time (nor now)
nope
the ipython/ipython Dockerfile is using 14.04
Unless I make a PR to change that, that's where things are
Jason Grout
@jasongrout
Jan 29 2015 21:11
sure.
Kyle Kelley
@rgbkrk
Jan 29 2015 21:12
Are you going to set it up now with 14.10?
Jason Grout
@jasongrout
Jan 29 2015 21:14
well, first I'm seeing if I can understand the 14.04 issue a bit better
are you sure that the 14.04 package is compiled with CXX_STD set?
Jonathan Frederic
@jdfreder
Jan 29 2015 23:40
@/all I just found a gross bug
I've somehow managed to break a cell
I've used 0 Javascript in my notebook
I have no extensions installed
and I'm on master
I'll post a YouTube video, it's really weird!
Min RK
@minrk
Jan 29 2015 23:43
what do you mean broken?