This is quite a weird one. The old regex was causing issues for some users and even Stack overflow errors while
trying to parse the Regex??
Explanation from Claude:
The character classes here (\u{e0061}-\u{e0066}, \u{e0041}-\u{e0046}, \u{e0030}-\u{e0039}) are all in the Unicode Tags block (U+E0000–U+E007F), which is outside the Basic Multilingual Plane — every one of these code points is represented internally as a surrogate pair. When you build a /u-flag regex with several ranges of astral (non-BMP) code points inside a character class, and then apply a repetition quantifier to that class ({1,6}, or +? as in the actual error), V8's Irregexp compiler has to expand/canonicalize the class and unroll the quantifier. For BMP regexes this is cheap, but the astral-range handling triggers a much deeper recursive expansion in the compiler itself.
This is a known class of V8/Chromium bugs (regex-compiler stack overflow on Unicode-mode patterns with supplementary-plane character-class ranges + quantifiers) — it's not something wrong with your logic, it's the compiling of the pattern that blows the native stack on affected Chromium versions/builds, not the matching. That's why it only hits "some users": it depends on their exact V8 build/stack size, not the input string.
* Move the shelf next to the editor
Fixing a couple of bugs and issues and moving the shelf with pictures beside the image editor for better usage and much better look from the previous scroller implementation
* misc
---------
Co-authored-by: thororen <78185467+thororen1234@users.noreply.github.com>
