cc @pelson

CC @mdboom based on the comment in #3972.

@blink1073 - the hardest part of this is figuring out where to go to download content.

If this is really a route we want to take (I personally don't see much of a use for a save button inside the notebook UI), then perhaps downloadify may be of interest.

@pelson, my use case is when the figure is part of a larger notebook UI. We could have a "Save" button in the UI, but this is more elegant and in keeping with the rest of the backends' toolbars.

New method: use the HTML5 input tag, so that when the user clicks the button it brings up a file selector dialog. This could be used for the webagg backend as well.

We can then push the save event back to the Python side, and use canvas.print_figure.

Cool. At least speaking for myself, we're always looking for more web/HTML5 development expertise, so feel free to fix up anything else you find that could be done better.

@mdboom, I'm a beginner at JS, but thanks for the vote of confidence!

How do we deal with converting the figure to a png when the figure gets killed? Presumably we could just use that method to serve up a png? [disclaimer, making this up as I go]

Shucks, the input tag can only select files on the user's file system. I'm still puzzling out how select a new file.

Success!

When you click the download button, it opens a new window with the image in the desired type, and you can save it from there.

@tacaswell, I don't follow your question.

Hmm, the only two formats that work are PNG and JPG, and the JPG doesn't look that great (it only shows a white background and the plot line itself). Should we just output PNG and remove the selector?

This needs a re-base.

Done.

Is there some reason the drop-down didn't get ported over to the nbagg ui? It seems useful to be able save vector graphics from the notebook.

On the other hand, users have access to *.savefig('...pdf') so not sure how compelling that use-case is as compared to the web-agg use-case.

and everything here works as advertised for me.

At first I tried using toDataURL, but for all but JPEG it ended up falling back on PNG. With JPEG, it rendered the transparent background as black, obscuring the spines.

@tacaswell - do you have some git foo which allows you to cherry-pick a branch onto v1.4.x?

Git co v1.4.x
Git cherry-pick hash_of_merge -m 1

Take a look that it added just one commit past upstream v1.4.x with all of
the changes squashed and then push to upstream.

That should probably be documented some place

On Mon, Jan 12, 2015, 08:13 Phil Elson notifications@github.com wrote:

@tacaswell https://github.com/tacaswell - do you have some git foo
which allows you to cherry-pick a branch onto v1.4.x?

—
Reply to this email directly or view it on GitHub
#3974 (comment)
.

back-ported as 408fed6

Seems to have been merged into v1.4.x with 408fed6. (doh - too early in the morning - I should have refreshed my browser)

Yes

On Tue, Jan 13, 2015, 05:08 Phil Elson notifications@github.com wrote:

Seems to have been merged into v1.4.x with 408fed6
408fed6.
Wasn't me though... @tacaswell https://github.com/tacaswell - did you
do it?

—
Reply to this email directly or view it on GitHub
#3974 (comment)
.