CW's Markdown engine gives inconsistent results between preview and actual result

[Voileexperiments]

An extension to #1441.

I'm not sure how is this even possible, but CW's Markdown engine apparently gives different results in "preview my comment" mode and in actual render, which I've encountered at least several times. One of the simplest examples is this:

~~should be striked~~next to correct text

In Github:

should be strikednext to correct text

In CW preview mode (correct):

In CW actual render (incorrect):

[Voileexperiments]

Related: #1617

[kazk]

I'm not sure how is this even possible

I think the preview uses a client side engine Marked and the actual one is by a server side engine.
Markdown implementation was all over the place before CommonMark and it's from those days.

[kazk]

This will be fixed with the huge updates I've been working on for the past month. The Markdown engine on the server side will be Common Mark.

#1441 is not fixed and seems to be a styling issue Never mind, it produces <ul> and <ol> as siblings.

[kazk]

How many spaces are there for the second level?

In this case, I think the preview is wrong.
The following

1. 1
2. 2
3. 3
  1. 3.a
  2. 3.b

is

on GitHub comment.

If you indent the second level with 4 spaces, you should get

---

Markdown preview uses a different engine on the client side so the best I can do is to find options to make it agree with the server side. And I haven't looked into this yet.

[Blind4Basics]

I used indentations with 2 whitespaces

[kazk]

I've updated to the latest version of marked. They're still working on becoming fully compatible with CommonMark spec, but it should be much better than the version we were using.