Bold/italic flanking issue

[studyingegret]

Yesterday I tried out Quarkdown for the first time and I specifically tested this code kinda well-known to have problem in Markdown

**人类的本质：**鸽子和复读机。

This renders to "**人类的本质：**鸽子和复读机。" instead of making bold. This is a known problem in CommonMark (see e.g. this thread), and as far as I know CommonMark has not made a solution for it, and people have made their own solutions like this and this(?).

I'm interested what Quarkdown feels about this issue.

---

My personal feel about it is an escape sequence like \! or \ that acts as punctuation character for the purpose of left/right-flanking computation _{(see section 6.2 of the spec)} but expands to nothing:

**人类的本质：**\!鸽子和复读机。

The big advantage is KISS:

• requires small changes to the parser (compared to a full spec for CJK)
  • in fact, flanking semantics is already present for \! and the only change needed is what it expands to: **人类的本质：**\!鸽子和复读机。: 人类的本质：!鸽子和复读机。
  • Edit: On second thoughts, it may be the AST conversion stages instead of the parsing stages that need to deal with \!
• no need to wrangle with a new set of rules (that never fulfill real-world text perfectly anyway)
• language independent!
• consistent with the trick of adding zero-width space or <span></span>

注意：**不要向任何人发送你的秘钥。**\!如果别人知道了你的秘钥，将能解密用公钥加密的信息，公钥会失去意义。

Luckily, equation 6 can be written in the form of a function of K+L (observe 16J\!**+3K+3L**=26J\!**-4K-4L**-18A), so we can express K+L with J and A and call it a day.

*std::box::*\!Box\<i32>

No, it *un-*\!locks the door instead of locking it.

Expected results

注意：不要向任何人发送你的秘钥。如果别人知道了你的秘钥，将能解密用公钥加密的信息，公钥会失去意义。

Luckily, equation 6 can be written in the form of a function of K+L (observe 16J+3K+3L=26J-4K-4L-18A), so we can express K+L with J and A and call it a day.

std::box::Box<i32>

No, it un-locks the door instead of locking it.

Of course, \! is already taken in CommonMark and I'll expect it to be used to escape images \![Not an image](https://example.com/dont-render-as-image), and it may look a bit cluttery, so they're tradeoffs to consider.

(\ seems not taken: a\ b: a\ b, but I dislike it because it looks spacey in document contexts not wanting a space, and expands to \ and a real space in Markdown.)

_{This might be a better thing to discuss at the CommonMark forum, but I am uncomfortable with posting there due to an unpleasant moderator unresponsiveness I have experienced.}

[iamgio]

Hi @studyingegret, thank you for the cool insights!

I feel an ad-hoc escape would be too overkill for such a niche use case. Luckily, you can get rid of ambiguity thanks to Quarkdown's .text function:

.text {人类的本质：} weight:{bold}鸽子和复读机。

I hope this is enough for your case!

[studyingegret]

Actually I disagree that it's a niche case. Search the regex /：\*/ and you'll see the pattern punct*nonpunct：*nonpunct used a lot. Some results in the search choose not to format the colon or add a space after the bold/italic mark to work around it. I can find more examples if you're interested

_{and notice how the full-width colon ： instead of the half-width one : brings out Chinese results :)}

[iamgio]

It makes sense, although as the sole maintainer I do not currently have the bandwidth to tweak the parser or add support for it. I don't mean to say that as an excuse, but Quarkdown has been parting away from CommonMark on so many small features (e.g. image titles now support formatting) that I'd say it's acceptable to leave it like that and use the function call where needed.

[studyingegret]

That's a pity, but I'll definitely try to contribute this change if I have the time later on!