Fixes #24: representation extensions; provides a plotting extension: scatterplots, line plots, decision boundaries etc. #47
Conversation
retrieval of handlers to avoid depending on module loading order.
This works sufficiently well for scatterplots, but insufficiently well for line plots and horribly for other plots
…-bounds crashes The out-of-bounds happen now due to still-unfixed bugs. Maybe it's better to keep crashing to iron out the bugs?
Also just one fallback point when there is a clash on a Boundary_map or Map.
|
All should be finished and tested now, ready for review. |
|
It's simple to make |
|
I just fixed HTML overlap ordering (z-index) and made framed=opaque. Now the the overlap of nested boxes output on HTML is as almost as good as in text -- only missing crop-to-canvas, and misplaced endpoints of axis labels (which is outside the scope of the plotting extension, it's
|
There was a problem hiding this comment.
I think this is a great feature, and the plotting part is excellent.
I have a concern or two about the extensibility story. type ext = .. is good; however I think the other types and tables are not at the right place.
Ideally, each backend (text, html, etc.) would have its own extension handlers table (e.g. (string, (ext -> html)) hashtbl or (string, (ext -> PrintBox_text.B.t)) Hashtbl.t). Then each extension (e.g plotting), extends ext, and registers a corresponding extension handler to PrintBox_text and PrintBox_html's respective tables.
What do you think?
|
Alright, it will be simpler indeed. |
|
Done! |
…ntbox-text` is spurious
|
Thank you!! |
There's no interpolation for line plots, and no nice graphics in the HTML backend -- it simply follows the text backend by placing boxes at the right coordinates.