I'm guessing that this is a side-effect from rPHU0db4b149c4e0. Strange though. :/

Ah, no - I was wrong. It's because of the ---- line from the quoted section (- indicates a list item too). Observe:

> --
> Hello
> 1. One
> 2. Two

Hello

1. One
2. Two

Versus:

> Hello
> 1. One
> 2. Two

Hello

1. One
2. Two

A partial "fix" for this is to allow space before the divider rule, which turns --- into this:

Then the line of "------" in the mail would render as a divider instead of an empty list item. This would also fix quoting dividers in general, since this:

> ---

..becomes this right now:

...which is clearly unexpected. I think this is worth pursuing, and making sure the divider rule runs before the list rule so it wins in these cases (if it doesn't already) to pick up the side effect "fix" is probably worthwhile.

We also should possibly decline to begin lists when the first item is empty. It's sort of nice right now that you can type - and get a bullet in the preview without needing to enter text yet (and, e.g., check that # or 1. or other list markers are doing what you expect them to do), but it produces very odd results in this case.

I'm not sure if the slightly nicer preview behavior is worth more or less than the unusual result here.

Another approach would be to use a different set of divider characters (i.e., not hyphens).

My biggest objection is actually to the renumbering — I numbered my bullets precisely so that I could refer to them by number later. So maybe there's a way to say, "hey, I'm about to combine two lists in a way that causes a numbered list to be renumbered" and just not do that?

It's hard for me to imagine a case where somebody would really want:

1. a
2. b
3. c
4. d
5. e
6. f

to turn into 1, 2, 3, 4, 5, 6.

...yes, see, like that. That seems broken. That is not what I wrote.

(To clarify, in case you can't see the original, I wrote that as two sets, separated by a blank line, one 1,2,3 and then another 1,2,3 again.)

rPHU31e0327 partially fixed some of this. Specifically, > --- is no longer interpreted as a list item: