[IPython-dev] Interactive Matplotlib in the browser

Michael Droettboom mdroe at stsci.edu
Fri Oct 12 13:04:49 EDT 2012

On 10/12/2012 10:28 AM, Michael Droettboom wrote:
> On 10/11/2012 10:06 PM, Jason Grout wrote:
>> On 10/11/12 5:50 PM, Jason Grout wrote:
>>> On 10/11/12 4:49 PM, Michael Droettboom wrote:
>>>> I have a proof-of-concept way to make interactive plots in the browser
>>>> work using transparent PNGs described here:
>>>> http://mdboom.github.com/blog/2012/10/11/matplotlib-in-the-browser-its-coming/ 
>>>> No PRs yet, because this is miles from ready for that, but it would be
>>>> helpful to get some feedback about how this works in different
>>>> browsers/platforms/network environments etc.
>>> A sample implementation using websockets instead of polling is here:
>>> https://gist.github.com/3875846
>>> It still requests the file, which causes a delay.  I think doing a png
>>> diff sounds like a great idea.  What if we also transfer the png diff
>>> over the websocket connection (maybe in a binary frame)?
>> I just updated it to transfer the image over websocket too. More timing
>> tests need to be done to see which approach is better, especially tests
>> across a network.
> Thanks.  That is very helpful
>>    The current version of
>> https://gist.github.com/3875846 also transfers over a base64 version of
>> the data; it seems like it would be better to transfer it as a binary
>> message.
> I see -- just so I'm clear: that's because we send it over a 
> websocket, and it's not obvious how else to put the binary data into 
> the Canvas Image object on the browser end?  I see that that's a bit 
> suboptimal.  I wonder if it makes sense to use a socket for the event 
> communication, but a regular request to get the image data. Then then 
> "push" would come from the event websocket as in my original example.

I found a better way -- recent versions of the websockets API do support 
binary data.  The magic incantation to put load binary data into an 
image object is here:


I've created a new github repo for this -- if we're going to collaborate 
on this before it develops into a real patch against matplotlib and/or 
ipython and/or sage that should be more convenient than using gists.


-------------- next part --------------
An HTML attachment was scrubbed...
URL: <http://mail.python.org/pipermail/ipython-dev/attachments/20121012/9bc47595/attachment.html>

More information about the IPython-dev mailing list