Hi all,
I haven’t been able to find a more official place to report potential Matplotlib bugs, so I’m going to describe the issue I’m seeing here. Sorry if this is the wrong forum.
On my system, it takes matplotlib a very very long time to close plots. Sometimes, up to 20 minutes to close a simple figure. Creating new figures remains fast. The problem seems to occur only when I’ve loaded a large amount of data in to python ( on the order of 1GB ). I am using the current version of Ubuntu and running “ipython --pylab”. To reproduce on my system, it is sufficient to load a large amount of data, create a plot… any plot, and then try to close it using the little “x” at the top right corner of the window. The whole session will freeze for an extended period of time. The plot does not have to be complex: a hundred datapoints, a thousand, it makes no difference. Since the problem only occurs when a large amount of data has been loaded, my guess is that there is a problem with how Matplotlib/Pylab/Python is trying to free the memory associated with the figure?
So… I just though I’d put this out there in case anyone else sees the same issue, or in case a developer who knows why this might be happening reads this. The workaround for me is… to simply wait for the figures to close, however long that may take, or restart the whole session.
Best,
michael.
Hi Michael,
I don’t have an answer about your bug. But the official place to report possible bugs is github.
https://github.com/matplotlib/matplotlib/issues?state=open
Cheers,
Fra
···
2014-06-12 18:07 GMT+02:00 M.Rule <mrule7404@…287…>:
Hi all,
I haven’t been able to find a more official place to report potential Matplotlib bugs, so I’m going to describe the issue I’m seeing here. Sorry if this is the wrong forum.
On my system, it takes matplotlib a very very long time to close plots. Sometimes, up to 20 minutes to close a simple figure. Creating new figures remains fast. The problem seems to occur only when I’ve loaded a large amount of data in to python ( on the order of 1GB ). I am using the current version of Ubuntu and running “ipython --pylab”. To reproduce on my system, it is sufficient to load a large amount of data, create a plot… any plot, and then try to close it using the little “x” at the top right corner of the window. The whole session will freeze for an extended period of time. The plot does not have to be complex: a hundred datapoints, a thousand, it makes no difference. Since the problem only occurs when a large amount of data has been loaded, my guess is that there is a problem with how Matplotlib/Pylab/Python is trying to free the memory associated with the figure?
So… I just though I’d put this out there in case anyone else sees the same issue, or in case a developer who knows why this might be happening reads this. The workaround for me is… to simply wait for the figures to close, however long that may take, or restart the whole session.
Best,
michael.
HPCC Systems Open Source Big Data Platform from LexisNexis Risk Solutions
Find What Matters Most in Your Big Data with HPCC Systems
Open Source. Fast. Scalable. Simple. Ideal for Dirty Data.
Leverages Graph Analysis for Fast Processing & Easy Data Exploration
http://p.sf.net/sfu/hpccsystems
Matplotlib-users mailing list
Matplotlib-users@lists.sourceforge.net
https://lists.sourceforge.net/lists/listinfo/matplotlib-users
Hi all,
I haven't been able to find a more official place to report potential
Matplotlib bugs, so I'm going to describe the issue I'm seeing here.
Sorry if this is the wrong forum.
This is *exactly* the right place to make a report like this. It is better to start with a message to a wide audience (at least, we hope it is a wide audience) such as this list, or maybe stack overflow, to see if someone else recognizes the problem. In many cases, it is not a matplotlib bug. If the response to an email like this does not lead to a solution, and the consensus is that it looks like you have hit a real bug, *then* file an issue on github.
On my system, it takes matplotlib a very very long time to close plots.
Sometimes, up to 20 minutes to close a simple figure. Creating new
figures remains fast. The problem seems to occur only when I've loaded a
large amount of data in to python ( on the order of 1GB ). I am using
the current version of Ubuntu and running "ipython --pylab". To
reproduce on my system, it is sufficient to load a large amount of data,
create a plot.. any plot, and then try to close it using the little "x"
at the top right corner of the window. The whole session will freeze for
an extended period of time. The plot does not have to be complex: a
hundred datapoints, a thousand, it makes no difference. Since the
problem only occurs when a large amount of data has been loaded, my
guess is that there is a problem with how Matplotlib/Pylab/Python is
trying to free the memory associated with the figure?
This sounds like the problem that prompted another user to propose Use less aggressive garbage collection by kmike · Pull Request #3045 · matplotlib/matplotlib · GitHub, except that your case sounds *much* more severe.
Can you reproduce the data loading and plot generation in a script that does not have to be run via ipython? And if so, is it still slow?
Is the large amount of data in the form of a very large number of python objects? Is there something odd about the data structure--extreme complexity that would make garbage collection take an absurd amount of time?
Since this does sound like the problem addressed by the pull request ("PR") noted above, please add your report (and any answers to my questions) as a comment on that PR.
Eric
···
On 2014/06/12, 6:07 AM, M.Rule wrote:
So... I just though I'd put this out there in case anyone else sees the
same issue, or in case a developer who knows why this might be happening
reads this. The workaround for me is... to simply wait for the figures
to close, however long that may take, or restart the whole session.
Best,
michael.
------------------------------------------------------------------------------
HPCC Systems Open Source Big Data Platform from LexisNexis Risk Solutions
Find What Matters Most in Your Big Data with HPCC Systems
Open Source. Fast. Scalable. Simple. Ideal for Dirty Data.
Leverages Graph Analysis for Fast Processing & Easy Data Exploration
http://p.sf.net/sfu/hpccsystems
_______________________________________________
Matplotlib-users mailing list
Matplotlib-users@lists.sourceforge.net
matplotlib-users List Signup and Options