Interdependent change tracking, ODF proposal

[Miklos Vajna]

Hi Regina,

On Thu, May 29, 2025 at 12:28:22AM +0200, Regina Henschel <rb.henschel at t-online.de> wrote:
> How is the hierarchy of the now two child elements of <text:changed-region>
> determined? Is it the element order in markup or the date in their
> <office:change-info> child element?

It's the order. So if you e.g. create an insert and then you have a
format on top of it, then format is listed first, then the insert.

The thinking is that ordering by date is not a good idea when you should
not trust the date of the entries, since it's simple user input.

[ Note that this is not something I invented, it's like this in
documents even from OOo times for "delete on top of insert", I just
noticed that this is not in ODF. ]

> What should happen if the <dc:date> of the <office:change-info> elements are
> in contrast to the type of elements? For example there is a delete of
> content that is inserted later than the delete was made? Or there is a
> formatting on a content that was already deleted according to the <dc:date>?

The properties of the change (author name, date, comment) is all
informational. The type and order of the changes is what's important.

The following combinations are what I handle so far in tdf#166319:

- delete on insert: this means reducing the result if a proposed insert
- format on insert: this results in a plain format on accepting the
  insert
- format on delete: this results in a plain format on rejecting the
  delete

These have a well-defined meaning, also in other formats like DOCX.

Thanks,

Miklos