Change em and strong to i and b

[snan]

Normally Markdown's default of using em for emphasis makes sense.
Emphasis is what you use italics for most of the time. When you need
cite, you can type and when you need a more generic italic, for
example for foreign language words or Linnaean names, you can type
. But on user-writable forums that disable fallthrough HTML, i is
the better default. It's never "wrong", per se (em is just more
specific) and it matches how many people sometimes use * and _ on
these forums, when they think "I want to make this look cursive"
rather than "I need to semantically emphasize this".

[kivikakk]

Hi there! Unfortunately this change would break CommonMark compatibility, so it's not one I can accept. Comrak is first and foremost a CommonMark library, and spec compliance takes priority.

Additionally, I'm a little bit confused by the rationale -- I'm not sure why one would convert Markdown to HTML only to then feed that HTML into a forum's user interface.

[snan]

I'm not sure why one would convert Markdown to HTML only to then feed that HTML into a forum's user interface.

Ah, no, that's not what I was saying at all. What I meant was that forum user interfaces use Comrak in safe mode (which is appropriate), people type in * and _ in there, and they get the wrong thing, which is em when they meant i.

With CommonMark, it's possible to write:

*this* to get <em>this</em>
<cite>this</cite> to get <cite>this</cite>
<i>this</i> to get <i>this</i>

<i> is the more general one. It is a superset of the other ones. People choose <em> or <cite> only when they want to be more specific.

Emphasis is for when you <em>really</em> mean something, cite is for when you mention name of a book, like <cite>Structure and Interpretation of Computer Programs</cite>, and the plain old italics element can handle those cases as well as cases not covered by them, like names of trees like <i>Gingko biloba</i>.

When it's possible to write all three, which CommonMark can (by being able to enter HTML directly), it's fine that the "shortcut" (using * or _) is for the most common one, which is em.

When it is not possible to write all three, and Comrak cannot (unless people turn on --unsafe, which, for web forums is a bad idea), the "shortcut" (using * or _) should emit the most general one, which is i. <i>This</i> is not wrong, but this is: <em>Gingko biloba</em>.

Comrak already is incomaptible with CommonMark unless you use --unsafe. (If you wanna make the "shortcut" (using * or _) emit em in --unsafe mode and i in safe mode, that'd be fine by me.)

[kivikakk]

Alright, I see what you mean!

I'm not inclined to change the default and change the output from that suggested by the spec (e.g. all of these), surprised downstream users, etc. The majority of implementations will follow the suggested output in the spec, given its double status as a conformance test.

But I'm happy to take a PR which introduces an option or options to switch these in the output.