Thank you very much for your encouragement. Honestly, I totally second your idea that using the same matplotlib would have probably been the best choice from the very first moment. Please also keep in mind that I am not a software developer, and that the time spent on this should be like an investment to speed up my future work, but unfortunately I can't really work on this all day ;)
Before starting to write down some lines of code, we had a look at the gnuplot Canvas page (http://skuld.bmsc.washington.edu/~merritt/gnuplot/ canvas_demos/ ) , and I was disappointed by not seeing a single example (there are more than 20, I certainly not looked at them all), about fast and useful interaction with the plots. I mean, apart from enabling a grid, what's the big deal (from my point of view, as a user) in having such a Canvas thing, if the result is having a static image? What do I know that the interaction is still possible, if they don't show in any way its possibilities? Panning, zooming, turning 3d objects, and other stuff should be the real target in my opinion! Have a look at the FLOT example page: http://people.iola.dk/olau/flot/examples/ . Apart from the very basic ones, the selection tool, zooming tool, online enabling of series, etc, looks very attractive to my eyes! Moreover, if it hadn't been for the fact of semilog or loglog plots, the work of including the plot in the sage notebook was done by this morning!! So at the end of this long day, my thoughts are: - FLOT is not so powerful, and the other derivatives (FLOTR, Protochart, etc) don't seem to be any better ( I can hardly recognize the differences), but still it is very fast to develop with in simple cases - matplotlib is obviously the most powerful and well integrated (my previous bode plot function relied on matplotlib), but I have some doubts about whether the rendering time would be small enough to provide a sufficiently satisfying user experience I hope you will hear from us pretty soon Regards Maurizio > As I understand it, matplotlib can do those sorts of things with an > interactive backend (i.e., wxwindows, gtk, qt, etc.). However, none of > those work with the web-based notebook, but the pieces are there. It > would be *cool* if someone would write the necessary > javascript/canvas/html code to make those tools available in a "canvas" > backend. Again, something like the gnuplot canvas backend. You might > be able to even reuse code from the gnuplot canvas backend. > > That said, it may be less effort to make an optional flot spkg. > However, for advanced plotting and deep integration with Sage, as you > are seeing, leveraging the power of matplotlib by just writing another > backend may be the best way to go in the end. > > I've looked at writing a matplotlib backend, and it didn't look > extremely difficult. There are examples in the backends/ directory, and > it looks like you can get a minimal backend up and running by just > writing a few functions. > > You're doing some great work on getting Flot working too! I think > exploring the options is what is most valuable at this point. I just > think that the matplotlib backend is one of the most valuable options > that can be explored. > > > > > Today we (Kenny and me) have a working alpha. It is unfinished, but we > > think that these added capabilities show the possible advantages, if > [snip] > > > We hope to send you some example ASAP, so that you can judge yourself > > and the decide what to do. > > I look forward to seeing what you have! > > Jason --~--~---------~--~----~------------~-------~--~----~ To post to this group, send email to sage-devel@googlegroups.com To unsubscribe from this group, send email to sage-devel-unsubscr...@googlegroups.com For more options, visit this group at http://groups.google.com/group/sage-devel URLs: http://www.sagemath.org -~----------~----~----~----~------~----~------~--~---
