OK, but both problems are "realigning your notes".
It is possible to render an element to the side of the main text in a flowing epub; I did this when I wanted to include line numbers in a text. You could use that idea to keep visible notes near their original location while reflowing the epub. But it wouldn't work at all with notes that appear over the main text.
It's also possible to just print the notes within the text; this is the approach taken by this recent edition of a selection of the 太平广记 ( https://www.amazon.com/dp/7540351934/ ). Rather than being reproduced images of older printings that include notes, it's all flowing text and marginalia is reproduced inline, within brackets and in a smaller font. (This goes so far as to indicate which part of the page the marginalia originally appeared in, though I think this is more a matter of there being different words for marginalia from different locations.)
That approach, of course, will not handle non-textual notes well.
I think OP could have used a more specific description for the sake of clarity.
The idea is that using a reMarkable is just like using a pad of paper, so you can make arbitrary handwritten notes on the text. It's hard to imagine how arbitrary notes like that would be displayed on different epub renderings, so I think it's understandable why they use their approach. I also think it's a different problem than kindles have solved.
That is exactly what I meant. The simplest UX you can have for arbitrary notes of any kind, scribbles, drawings, highlights, is to not encourage page reflow. You can come up with a bunch of other funky weird technical solutions, but once the user hits them there's a strong chance of confusion.
For tablets which encourage writing on your document, the rm2 approach of "just render epub as you would a pdf" feels by far like the best strategy.
The original epub is still stored within their document format btw. You could technically still do things with that file if you wanted.
Marginalia does not generally appear over the text on which it comments, because that would make both the text and the marginalia difficult to read. (Just look at the word - it's text that appears in the margins.)
So for practically all purposes, treating each note as an image which should be rendered to the side of a particular part of the dynamically-flowed text will solve the problem. This isn't that hard to do.
If someone is underlining parts of the text itself, that isn't independent of the flow of the text, and so it's harder to reflow. But I'm taking "notes" to mean commentary.
In the context of the reMarkable / this conversation, “notes” means whatever the user draws on the device with the stylus: underlining or circling or crossing-out words or drawing lines between them, drawing pictures or hand-writing in the margin or between lines or on the words, etc: whatever you would or could do with a physical book and a pen/pencil.
(A primary design goal of the reMarkable tablet is to be as similar to paper as possible, so (if I understood your suggestion correctly) telling users that their "notes" will be treated as images to be re-rendered to the side of the text, instead of where they put it—anywhere—would break the similarity, and apply only to a small subset of possible "notes", namely "commentary", as you said.)
> So for practically all purposes, treating each note as an image which should be rendered to the side of a particular part of the dynamically-flowed text will solve the problem.
I don't think this is true. Even without underlines, which are common in notes, people place their notes based on a combination of the layout of the page and their personal preferences. I don't think it's possible to know, given one layout and note placement, how to place the note in a new layout.
It is also worth saying that many historic uses of marginalia (for example, commentary on the Tanakh) are also associated with particular parts of the text and could not be laid out as you are describing without losing intended meaning.
> It is also worth saying that many historic uses of marginalia (for example, commentary on the Tanakh) are also associated with particular parts of the text and could not be laid out as you are describing without losing intended meaning.
I don't understand this. What part of the intended meaning would be lost? I have a printing plate of part of a Chinese Buddhist text with traditional commentary attached, and it works almost identically to this modern HTML from Harvard Law Review ( https://harvardlawreview.org/2021/06/commonwealth-v-mccarthy... ) - there's an area on the page for the text, another area for the commentary, and symbols identify which comments apply to which parts of the text.
But the Harvard Law Review piece is exactly what I just described, except that the notes do not appear near the text to which they apply.
Honestly, for marginalia I would love it if you could just insert blank notes pages within pdfs and write on them! Also that you could read the pdf with those notes pages enabled/disabled, and easily copy/move them to a separate document later.