These are chat archives for ipython/ipython

16th
Oct 2014
Min RK
@minrk
Oct 16 2014 00:05 UTC
@jdfreder 426 means old web sockets. This happens with tornado 4 + phantomjs
that's why we pin travis to tornado < 4
Jonathan Frederic
@jdfreder
Oct 16 2014 00:14 UTC
Thanks @minrk , I'm only seeing the 426 locally, so I must have a newer tornado installed
Min RK
@minrk
Oct 16 2014 00:26 UTC
it's one advantage of slimer - since it just uses your Firefox, it's not stuck on an outdated pre-release version of web sockets like phantom is
Min RK
@minrk
Oct 16 2014 00:40 UTC
@jdfreder #6712 should make it slightly less confusing
Jonathan Frederic
@jdfreder
Oct 16 2014 04:08 UTC
SVG. broken. in. master. ?.
Matthias Bussonnier
@Carreau
Oct 16 2014 16:22 UTC
handout in 10 or 40 today ?
Jonathan Frederic
@jdfreder
Oct 16 2014 16:23 UTC
I thought today was still 10:00am, I could be wrong though. @minrk ?
Remi Rampin
@remram44
Oct 16 2014 16:23 UTC
What is this "deferred_type_pprinters" in lib/pretty.py?
It says #: printers for types specified by name but it doesn't look like it resolves names at any point (so it'd have to contain callable to begin with)
Matthias Bussonnier
@Carreau
Oct 16 2014 16:25 UTC
@remram44 , not sure havent touched pprint in loooooooonnng time.
Remi Rampin
@remram44
Oct 16 2014 16:25 UTC
look here
Matthias Bussonnier
@Carreau
Oct 16 2014 16:27 UTC
used by SymPy extension apparently.
Remi Rampin
@remram44
Oct 16 2014 16:27 UTC
Do you have a link?
Do they actually use the fact that it's deferred or did they randomly put it there instead of the other dict?
Matthias Bussonnier
@Carreau
Oct 16 2014 16:28 UTC
No, I just grepped IPython source.
Thomas Kluyver
@takluyver
Oct 16 2014 16:28 UTC
I think it's designed so you can register a pretty printer without having to import the module for that type
Min RK
@minrk
Oct 16 2014 16:28 UTC
I thought we had moved to 9:30
Jonathan Frederic
@jdfreder
Oct 16 2014 16:29 UTC
oh ok :)
well I'm ready
Remi Rampin
@remram44
Oct 16 2014 16:29 UTC
@takluyver yes but for that there should be code somewhere to do the deferred import
Matthias Bussonnier
@Carreau
Oct 16 2014 16:29 UTC
I'm ready, but just to know if everyone knows
Min RK
@minrk
Oct 16 2014 16:29 UTC
I assume @fperez and @ellisonbg won't be coming due to Strata
Matthias Bussonnier
@Carreau
Oct 16 2014 16:29 UTC
Oh, ok.
Kyle Kelley
@rgbkrk
Oct 16 2014 16:30 UTC
nor me
I'm also at Strata
Thomas Kluyver
@takluyver
Oct 16 2014 16:30 UTC
@remram44 There doesn't need to be. You can get the module name and class name from an object, and look up the pprinter by that.
Min RK
@minrk
Oct 16 2014 16:30 UTC
probably doesn't need to be a long one this week
Remi Rampin
@remram44
Oct 16 2014 16:31 UTC
Oh, so the deferred import is not the printer's it's the printed object's class's
This makes sense. Also, this looks like what Python's warnings module should be doing.
Thanks!
Thomas Kluyver
@takluyver
Oct 16 2014 16:31 UTC
yep
you're welcome
Min RK
@minrk
Oct 16 2014 16:32 UTC
invites went out
Matthias Bussonnier
@Carreau
Oct 16 2014 16:34 UTC
no sound ....
ok, can't here and can't speek... love hangout
Thomas Kluyver
@takluyver
Oct 16 2014 16:36 UTC
@jhamrick , if you're around, we're starting the meeting now (we agreed to move it half an hour forwards)
Min RK
@minrk
Oct 16 2014 16:36 UTC
@Carreau want to keep trying, or just chime in on chat and watch on YouTube?
Jessica B. Hamrick
@jhamrick
Oct 16 2014 16:54 UTC
Oh, didn’t realize thursday meetings had also moved to 9:30
Jason Grout
@jasongrout
Oct 16 2014 17:16 UTC
is the dev meeting done?
that was quick!
Thomas Kluyver
@takluyver
Oct 16 2014 17:18 UTC
we started at 9.30
and it tends to go quicker with fewer people ;-)
Jason Grout
@jasongrout
Oct 16 2014 17:19 UTC
one person making decisions goes even quicker :)
Damian Avila
@damianavila
Oct 16 2014 17:20 UTC
jajaj
Remi Rampin
@remram44
Oct 16 2014 17:35 UTC
What does "synchronize" mean in gitter's "activity" panel on the right?
Jessica B. Hamrick
@jhamrick
Oct 16 2014 17:36 UTC
That someone pushed more commits to a pull request branch
Damian Avila
@damianavila
Oct 16 2014 17:37 UTC
that's nice... github does not notify you if there is a new push into a PR...
Jessica B. Hamrick
@jhamrick
Oct 16 2014 17:38 UTC
Yeah
It might be a little overwhelming email-wise if it did though
Is there a recommended way for writing javascript tests for ipython extensions?
It would be nice to use the existing tools that are in ipython already
Min RK
@minrk
Oct 16 2014 17:42 UTC
I think someone has done that
it shouldn't be too hard to allow an absolute path under iptest js
(a symlink from IPython/html/tests/yourtests to your test files might work with iptest js/yourtests right now)
Jessica B. Hamrick
@jhamrick
Oct 16 2014 17:44 UTC
Awesome I’ll try that out
Thanks!
Min RK
@minrk
Oct 16 2014 17:45 UTC
I wonder if iptest js//abs/path would work with what we have now...
Jessica B. Hamrick
@jhamrick
Oct 16 2014 17:46 UTC
I think it might actually
Let me try
Hah yeah it totally does
Jason Grout
@jasongrout
Oct 16 2014 17:48 UTC
@jdfreder?
What do you think of making a default behavior of the base view class in its render method to tag the top-most view div with the view id and the view class name?
this.el.dataset.viewName = "view name"
this.el.dataset.viewID = "view id"
Jason Grout
@jasongrout
Oct 16 2014 19:47 UTC
eh, maybe that's a bit too much. It would be nice for styling, I guess.
(so I don't know what I think about it, so never mind)
Jonathan Frederic
@jdfreder
Oct 16 2014 21:12 UTC
@jasongrout is that addressable with selectors?
At one point, early in the design, I thought about exposing the model ids through the DOM views via the el's id, but quickly realized that wouldn't work for multiple views.
I then thought about doing the same using classes
i.e. widget-3249872349623984
but thought that was a kludge
Jason Grout
@jasongrout
Oct 16 2014 22:16 UTC
this is view ids, not model ids
(and I guess it would be done in the DOMWidgetView, not the base widget view, of course)
Jonathan Frederic
@jdfreder
Oct 16 2014 22:19 UTC
I think setting this.$el.attr('id', this.id) would be okay to do in the render method of the DOMViewBase
Jason Grout
@jasongrout
Oct 16 2014 22:21 UTC
It would be useful in cases where you have some external javascript library doing things to the DOM, and then you want to reflect those changes back into the widget infrastructure.
Jason Grout
@jasongrout
Oct 16 2014 22:31 UTC
Or if you wanted more references flying around, I suppose you could associate the actual view with the DOM element.
Jonathan Frederic
@jdfreder
Oct 16 2014 22:35 UTC
Well view ids are references, they just don't have the negative side effect of leaking memory if we forget to delete them from the DOM elements when the views are destroyed.