'reshare' or 'republish' button remains when notebook content returns to the shared/published state

[jrus]

Summary: When a change is made to a shared or published notebook, but then undone (e.g. a constant is changed and then changed back, or a cell is created and then deleted again) the notebook UI still displays a 'reshare' or 'republish' button, even though resharing or republishing has no real effect on content (it only changes the notebook modification date and shared/published version number).

To Reproduce: Go to any shared notebook where you are an author. Create a new cell. Delete that cell.

Expected behavior: The 'reshare' button should disappear

Actual behavior: A 'reshare' button floats at the top of the screen.

Proposed remedy: Hash the notebook content, and don't display the 'reshare' or 'republish' button when the current notebook state matches the currently shared/published version.

Current workaround: Go to the history view and make an explicit diff.

Discussion: This causes confusion and anxiety for authors who aren't sure if they made a meaningful change or not.

Original request at talk/3518

[mootari]

This looks like a duplicate of #24

Edit: see below

[mbostock]

@mootari This seems to be a slightly different issue, since it pertains to edits made it a single window, and asks for the reshare/republish promotion to be based on the content of the notebook changing.

[mootari]

@mbostock Oops, you're right! Sorry for the noise!

[mootari]

Note that the correct workaround would be to select "compare edits" from the notebook menu (currently missing in Next).

So, there's a gotcha with the suggested solution: While the published content matches the current version, the most recent version would still be marked as unpublished. Removing the Publish / Reshare button in this case would prevent me from marking the most recent version as published.

Technically you can't "undo" anything in Observable, because even a revert will log a new version, and hiding the button might make the situation even more confusing because the history would still show more recent versions than the last published one.

I would instead suggest to find some way to display the "effective" changes (as number of lines, modified cells, or simply a boolean state) along with the republish/reshare button.