These are chat archives for ipython/ipython

23rd
Jan 2015
Zachary Jones
@bigzachattack
Jan 23 2015 02:37
I started using the terminal app in in notebook mode of master today and noticed some weird behaviors. I was wondering what is the expected functionality of the terminal for the 3.0 release?
I noticed that...
  • When the terminal first starts, the shell prompt is not there. Any key stroke causes it to be shown
  • Curses programs (or at least htop) cause the terminal to become complete garbled
  • Ctrl+<keys> were not being passed into the shell.
Not big deal for me, if these are out of scope for 3.0 or limitations of the libraries being used. I just wanted to make sure they are not bugs.
Zachary Jones
@bigzachattack
Jan 23 2015 02:46
Well, just tried a different browser on a different machine and looks like they are bugs in the browser. Though I still can't get Ctrl+A + <key> in screen to work.
Min RK
@minrk
Jan 23 2015 02:48
Seems to be specific to ctrl-A
ctrl-B for tmux works fine
There are definitely lots of issues in the terminal, but we aren't going to let them block the release.
Zachary Jones
@bigzachattack
Jan 23 2015 02:48
Ctrl+A is probably being caught by the browser for select all
Min RK
@minrk
Jan 23 2015 02:49
I wouldn't think so
but something's definitely amiss
The missing prompt seems to happen on terminals that are already running, since a new prompt hasn't been drawn. It should work, but doesn't seem to trigger the initial sync properly.
ctrl-everything else I can think of works, just not ctrl-A
and I'm on OS X, where normal ctrl-A behavior is go to beginning of line, not select all
Zachary Jones
@bigzachattack
Jan 23 2015 02:52
I just RTFM term.js and Ctrl+A is captured for its own purposes.
Min RK
@minrk
Jan 23 2015 02:52
ah
We might be able to disable that.
Zachary Jones
@bigzachattack
Jan 23 2015 02:55
Well maybe not, that looks like Ctrl+A, not Ctrl+a, but Ctrl+a is going somewhere too
The Ctrl+A followed by Ctrl+V lets you paste.
Min RK
@minrk
Jan 23 2015 02:56
keyDown is not case sensitive
a == A
looks like the keycode stuff is super hard-coded
Zachary Jones
@bigzachattack
Jan 23 2015 02:58
I was just looking
Min RK
@minrk
Jan 23 2015 02:58
that's unfortunate
Zachary Jones
@bigzachattack
Jan 23 2015 02:58
and I have to stop using chrome, Ctrl+a is disappearing into something
Min RK
@minrk
Jan 23 2015 02:59
The garbled curses is weird, htop runs fine for me in Chrome
But things like ctrl-A we won't accept as IPython issues, but upstream problems.
Zachary Jones
@bigzachattack
Jan 23 2015 03:00
It looks like the garbled screen is just my work computer running chrome
Firefox is fine and at home chrome and firefox are fine
Min RK
@minrk
Jan 23 2015 03:57
might be an extension
Jessica B. Hamrick
@jhamrick
Jan 23 2015 07:26
@bigzachattack @minrk Ah, I’ve also been encountering the issue where the prompt is not shown initially until you press enter or something (and I’ve had this happen even with new terminals as well as ones that are already running).
Jason Grout
@jasongrout
Jan 23 2015 15:55
@olgabot: is there a separate help channel that you are talking about?
@olgabot: the .map files are for debugging javascript. Can you find the .map files in the directory tree? They should be in IPython/html/static/components/backbone/backbone-min.map and
IPython/html/static/components/underscore/underscore-min.map
Jason Grout
@jasongrout
Jan 23 2015 16:01
oh, hey, I just found the help channel. I guess @olgabot has been continuing the conversation there...
Jason Grout
@jasongrout
Jan 23 2015 17:46
@minrk: I'm working on updating your gist that doctests a notebook. I have some questions about changes in the message spec and the notebook format since you wrote it. Can you ping me when you have a minute?
Min RK
@minrk
Jan 23 2015 17:52
@jasongrout will do, just relocating
Jason Grout
@jasongrout
Jan 23 2015 17:57
I think I figured it out. I should treat the new execute_result as a pyout message for the purposes of comparing, since it appears that the notebook format doesn't store the execute_result message as an execute_result message.
Min RK
@minrk
Jan 23 2015 18:15
msgspec and nbformat have both changed, so it depends on which of each you are using
if both are latest, msgspec matches nbformat more closely than it used to.
Jason Grout
@jasongrout
Jan 23 2015 18:38
Let me upload the changes to your ipnbdoctest.py and see what you think. I am using the latest of everything.
Min RK
@minrk
Jan 23 2015 18:47
That means there's no pyout anywhere
or pyerr
Jason Grout
@jasongrout
Jan 23 2015 19:02
I just realized that when I used reads(), I was asking it to convert to nbformat 3. oops.
Jason Grout
@jasongrout
Jan 23 2015 22:03
@minrk: question about msg spec vs nbformat. It seems that the message spec, for msg_type=='stream', has 'name' and 'text' fields, while the nbformat has 'name' and 'data' fields. Is that right?
Min RK
@minrk
Jan 23 2015 22:12
nbformat doc is out of date. It's name, text in both.
Jason Grout
@jasongrout
Jan 23 2015 22:12
ah, okay, thanks.
Jason Grout
@jasongrout
Jan 23 2015 22:27
I think before, the notebook metadata had a name attribute. Does the new nbformat have a name stored anywhere?
Thomas Kluyver
@takluyver
Jan 23 2015 22:29
No. The name was redundant with the actual filename, which caused strange effects when they got out of sync.
Jason Grout
@jasongrout
Jan 23 2015 22:45
@minrk: do you mind putting a license on https://gist.github.com/minrk/2620735?
Min RK
@minrk
Jan 23 2015 22:46
@jasongrout sure. In general, I consider my gists "Public Domain, but credit is nice"
Jason Grout
@jasongrout
Jan 23 2015 22:47
that makes sense. it's easier for us if the license is explicit, of course.
Min RK
@minrk
Jan 23 2015 23:00
yup, I'll add it.
license added. I'll try to remember to do that in the future.
Jason Grout
@jasongrout
Jan 23 2015 23:06
thanks! Also, do you think that there might be a place for an updated version of that script in the utils folder or something?
It really is kind of handy.
I have an updated version too...
it probably wouldn't hurt to add the IPython example notebooks to the testing infrastructure with that script either...
Min RK
@minrk
Jan 23 2015 23:09
I'd rather keep it out of the IPython repo
since it is decidedly incomplete
Jason Grout
@jasongrout
Jan 23 2015 23:09
considering how easy it is to have those notebooks out of date.
Min RK
@minrk
Jan 23 2015 23:09
but I can move it from a gist to a proper repo
yeah
Jason Grout
@jasongrout
Jan 23 2015 23:09
in fact, I think I'll run it on the existing notebooks, just to see.
Min RK
@minrk
Jan 23 2015 23:10
I wouldn't add the script to the repo before we start doing that, though