Created attachment 194992 [details] sample document Open new writer document Make number list as below 1.あああ 2.いいい 3.ううう Set symbols of list to Multibyte number "１,２,３..." １.あああ ２.いいい ３.ううう When inserting singlebyte alphanumeric characters at the beginning of the list content, the digits of the symbols become single-byte characters.(See a sample document) 1.1あああ ２.いいい 3.aううう And,It is unclear if it is the same issue, but the symbols in the numbered list settings dialog also appear to be half-width digits.(See a screen shot)
Created attachment 194993 [details] Screen shot of symbol preview in dialogue
What I see in the attachment: 1. Place cursor before number 4: the font is 游明朝 (I don't have it, so is substituted by something else), which is the Asian font of the "Numbering Symbol" character style used in Format > Bullets and Numbering > Customize. 2. Place the cursor at the beginning of item's text (i.e. before "えええ") 3. Type some English character e.g. "a" Result: The numbering font for this item (4) switches to Alef, which is the Western font for the "Numbering Symbol" character style. 4. Type Japanese again at the beginning of the line Result: numbering font reverts back to character style's Asian font. Can you please confirm that this is the issue you describe?
I am not the bug reporter but here are my two cents as a CJK user: (In reply to Stéphane Guillou (stragu) from comment #2) > Can you please confirm that this is the issue you describe? The fundamental issue is not changes of fonts between western ones and Asian ones, but the characters used for the list numbering. 1, 2, 3, ... are the ordinary digits, Unicode 0x0030 to 0x0039. １, ２, ３, ... are so-called "full-width" [1] digits used by CJK encodings, Unicode 0xFF10 to 0xFF19. Most CJK fonts have both sets of characters, and the second set are usually much wider than the first set, to match the width of kanji/hanzi/Han characters. I can more or less reproduce the report, as the list numbering switches between the half-width and full-width (the reporter call them single-byte and multi-byte in the original report) sets when I explicitly choose the full-width numbering, and type half-width digits or latin letters immediately after the numbering. You can set both the western font and the Asian font of the document as a single CJK font (the open source Noto Sans CJK / Source Han Sans will do) and see the difference more clearly. 1. https://help.libreoffice.org/latest/en-US/text/shared/00/00000005.html?DbPAR=SHARED#bm_id911622865848718
Thanks for the explanation. I'll let Jonathan help with this one as I'm out of my depth.
I was able to reproduce the issue. I've also manually verified that Unicode DIGITs are used instead of the intended FULLWIDTH DIGITs when the line starts with a Latin character. Besides aesthetic concerns, I think we shouldn't overrule a user who has explicitly chosen the full-width number format. We already have the "Native Numbering" format for users who do want numbering to be reactive to the following text, and among the rest this seems like an outlier. We don't perform this conversion for any of the ideographic, RTL, or CTL formats, for example; it's only this one. I'm not certain, but from briefly looking at the code I also suspect this behavior isn't intentional. Based on the above, I'm confirming this bug.