These are chat archives for ipython/ipython

16th
Jul 2015
Scott Sanderson
@ssanderson
Jul 16 2015 15:10
just hit a bizarre bug in the IPython terminal; I ran something, then tried to use the Up Arrow to invoke it again, but instead of a history entry for that command there was just an empty line
In [4]: *** Error in `/home/ssanderson/.virtualenvs/qexec/bin/python': double free or corruption (out): 0x00000000031ed770 *** Aborted
when I tried to execute the blank line, I got this:
are there C extensions hidden away somewhere in the codebase?
Dave Willmer
@dwillmer
Jul 16 2015 15:12
@ssanderson can you reproduce? can you give us instructions on what you actually entered? do you have a core dump?
Scott Sanderson
@ssanderson
Jul 16 2015 15:13
haven't been able to reproduce it
my first guess is there's some odd interaction with tmux and/or the terminal
but that's entirely speculation
if I get into the state with the bad entry I'll try to poke at it some more before executing
Dave Willmer
@dwillmer
Jul 16 2015 15:14
what version of python/ipython are you running?
thanks - are you familiar with gdb?
Scott Sanderson
@ssanderson
Jul 16 2015 15:14
IPython 3.1.0, Python 2.7.6; it's entirely possible that this is fixed in 2.7.10 as well
if it's a Python issue rather than an IPython issue
familiar enough with gdb to follow instructions if there's something straightforward I can do
Dave Willmer
@dwillmer
Jul 16 2015 15:16
well, if you can attach gdb to the python process, that would enable us to find out where it's core dumping, but not necessarily where the corruption is taking place
Scott Sanderson
@ssanderson
Jul 16 2015 15:16
gotcha
Dave Willmer
@dwillmer
Jul 16 2015 15:17
import os; os.getpid() # gives you the python process id
Scott Sanderson
@ssanderson
Jul 16 2015 15:17
yeah, memory corruption bugs are no fun :/
Dave Willmer
@dwillmer
Jul 16 2015 15:17
gdb -p <process_id> will attach to that process
yea agreed! if there's any lines of input that we can try in order to repro that would be helpful as well
Scott Sanderson
@ssanderson
Jul 16 2015 15:18
it doesn't seem to have been tied to the particular lines I was entering
which is partly why I suspect terminal interaction
Dave Willmer
@dwillmer
Jul 16 2015 15:18
yea, fair enough
Min RK
@minrk
Jul 16 2015 16:27
given the description, readline seems most likely to me.
Sylvain Corlay
@SylvainCorlay
Jul 16 2015 16:57
hey anyone aware of a web ui to hdf5 ?
Jason Grout
@jasongrout
Jul 16 2015 16:59
do I hear another use-case for widgets? :)
Sylvain Corlay
@SylvainCorlay
Jul 16 2015 17:01
yeah, this is the type of thing that would require having a light viewmodel compared to the actual dataset
Min RK
@minrk
Jul 16 2015 17:01
There is the hdf5 rest api, but I don't think they have UI yet
Sylvain Corlay
@SylvainCorlay
Jul 16 2015 17:03
yeah, I came across that one.
Jason Grout
@jasongrout
Jul 16 2015 17:04
well, chris's new scrolling widget supports 8.5 quadrillion rows. Who needs more than that?
S. Chris Colbert
@sccolbert
Jul 16 2015 17:10
If printed out at 20px per row and 96dpi, that would stretch to Jupiter and back 38 times
/ididthemath
Dave Willmer
@dwillmer
Jul 16 2015 17:11

toomuchfreetime

...that was meant to be a hashtag, clearly i haven't read the gitter manual
S. Chris Colbert
@sccolbert
Jul 16 2015 17:14
I actually prefer you shouting at me. You should leave it as-is.
Jason Grout
@jasongrout
Jul 16 2015 17:20
pluto is the fashionable target these days...
(is that heresy on the jupyter channels??? :)
S. Chris Colbert
@sccolbert
Jul 16 2015 17:22
= almost 4 round trips to pluto
Min RK
@minrk
Jul 16 2015 17:24
I think you mean plyto
S. Chris Colbert
@sccolbert
Jul 16 2015 17:24
if printed on standard 8.5 x 11 paper, it would weigh 805 megatons
Jason Grout
@jasongrout
Jul 16 2015 17:25
Chris has to use Knuth paper-stack notation...
S. Chris Colbert
@sccolbert
Jul 16 2015 17:26
if the sheets were stacked vertically, they would round-trip the moon 21 times
Jason Grout
@jasongrout
Jul 16 2015 17:27
there should be a website that takes a distance and gives it back in all sorts of interesting measurements.
S. Chris Colbert
@sccolbert
Jul 16 2015 17:27
relevant xkcd :)