reopening #27605 (Math blocks can't be directly preceded or followed by alphanumeric character)

[nschloe]

Description

Unless I'm missing something, the issue #27605 has reappeared despite its fix in #30175.

MWE:

a$x$ -$x$ 1$x$

Screenshot:

$a$1

$a$-

$a$x

This also concerns the \(...\) and \[...\] syntax.

Gitea Version

1.23.0+dev-753-gc5422fae9a

Can you reproduce the bug on the Gitea demo site?

Yes

Log Gist

No response

Screenshots

No response

Git Version

No response

Operating System

No response

How are you running Gitea?

demo.gitea.com

Database

None

[wxiaoguang]

a$x$ -$x$ 1$x$

What's the problem? If you'd like to render it, please add some spaces.

Otherwise, it's impossible to distinguish from This is $1 and that is 20 US$, we shouldn't render everything between $, say here it is 1 and that is 20US.

[wxiaoguang]

Unless I'm missing something, the issue #27605 has reappeared despite its fix in #30175.

The issue doesn't "reappear", because there are enough tests in #30175, a$x$ isn't rendered as "math".

[wxiaoguang]

By the way, maybe you might be interested in my recent refactoring PRs like #32728

[wxiaoguang]

Hmm, I see the misconception now. Actually #30175's title doesn't really match its test cases.

So, if a$x$ -$x$ 1$x$ is a must to support, we need to clarify its usage first and design the cases to avoid interfering other daily sentences like US$1 and 20US$

• the question is: why not write $ax$ $-x$ $1x$?

---

The original discussion is here: #30175 (comment)

And this: #30175 (comment)

I looked at some websites and math research papers and it seems that the math expressions are never directly preceeded by a character, they are either a whole mathematical block or have a space in between like a $x$.

[nschloe]

the question is: why not write $ax$ $-x$ $1x$?

There are various cases where you don't want spaces before or after math, for example:

• $n$-dimensional
• The answer is $x$. (any punctuation mark)
• ($a$ replaced by $b$) (any bracket, before and after)
• the $i$th iterate

There are more examples. None of this works in Gitea now; this also concerns the the $$...$$-, \(...\)-, and \[...\] syntaxes.

Opinion

Dealing with $-math is hard, so hard even that Donald Knuth himself suggested adding it to TeX was a mistake, later remedied by adding \(...\) to LaTeX. In Markdown, as opposed to TeX and as @wxiaoguang mentiones above, it gets even more tricky since the single $ already has a meaning that must be preserved.

Both GitLab and GitHub added a syntax that ties in beautifully with Markdown, namely

$`a + b = c`$ for inline math and 

```math
a + b = c
```

for display math.

None of the problems mentioned in this and other reports exist here because it makes use of the backticks, a well-established syntax in Markdown.

I've suggested adopting this syntax in Gitea as well.

[wxiaoguang]

So if we support $` for inline math, then this issue (and maybe some related legacy issues) could be fully resolved?

[nschloe]

At least you could point anyone running into a $-syntax bug to the working $`...`$ syntax. I myself would only use the latter anyway, and recommend everyone would do the same. (Same for $$ vs ```math.)

Edit: Highlighting why the \[...\] is problematic as well: #27645

[wxiaoguang]

Edit: Highlighting why the \[...\] is problematic as well: #27645

Yes, I also found that \[...\] is problematic, it's not right to introduce it. But I guess we couldn't remove it since it would be a breaking change.

After " Refactor markdown math render #32831 ", the $`x`$ will be supported, then other issues could be closed IMO (encourage users to use the new syntax)

[nschloe]

But I guess we couldn't remove it since it would be a breaking change.

Indeed, unfortunately so.

After " Refactor markdown math render #32831 ", the $x$ will be supported, then other issues could be closed IMO (encourage users to use the new syntax)

Note that we'd also need ```math code blocks to get feature parity with GitHub/GitLab.

[wxiaoguang]

Note that we'd also need ```math code blocks to get feature parity with GitHub/GitLab.

It is already supported, right? https://demo.gitea.com/wxiaoguang/test-repo/issues/1#issuecomment-3436

[nschloe]

Right! Hadn't realized that.