I am especially distrustful of copy/paste
As I've probably written somewhere above, this is something that I consider an issue across all program pairs. Between any given pair, the behaviour is usually stable.
But between different pairs, that's not true:
I'm just testing Logseq. One page, in 'document' mode, I have typed four new paragraphs and a total of six new lines. I indented one paragraph. Copied (Ctrl-C) the page.
- Paste into Word produces 10 paragraphs, but correctly picks up the indented lines.
- Paste into FocusWriter, MarkText correctly identifies the paragraphs and lines and the indents. But also sticks bullets in for each new paragraph.
- Typora gets the bullets, but doesn't identify all the paragraphs correctly.
- I think Obsidian is right.
- Workflowy is correct for bullets in its note, but has no paragraphs. Paste into the outline and it has too many indents
- Dynalist is similar but not identical.
OK, that's all from Logseq which is basically an outliner, even in document mode.
Or paste a small piece from Obsidian with two paragraphs and two new lines added to the first.
- Workflowy outline counts them as four undifferentiated bullets. The note is accurate.
- Dynalist outline has an extra empty bullet for the empty line signifying a markdown paragraph. Again, note is accurate.
- Word and Atlantis have four paragraphs plus an empty one.
- Logseq is accurate, as are FocusWriter, MarkText and Typora.
They were nearly all correct pasting a number of paragraphs from Word.
But Dynalist pasted them as lines in the note, whereas Workflowy correctly identified the paragraphs.
The whole copy/paste thing is unreliable unless you know the detailed circumstances, and is more likely to misfire when switching between RTF and plaintext programs. And programs of different types such as outliners and editors or word processors. (I suspect word processors are better sources because their paragraph and line markers are more explicit.)
And that's ignoring HTML and browsers.