When reporting interrupted speech (either in fictional dialogue or transcribed conversations) it is standard to use em-dash as follows:
"I assure you I can."
In this context, by default LibreOffice doesn't convert -- to em-dash, because the default replacement table entry expects a space before the dashes. To get it to work, the entry .*-- or .*--.* has to be added manually to the replacement table.
Steps to Reproduce:
1. Type any line of dialogue, with or without quotation marks, that ends with a trailing em-dash to indicate interruption, e.g. "What--?" or What--? or "So--" or So--
-- did not convert to em-dash
-- should convert to em-dash. The user should not have to manually add the entry with wildcards to the replacement table.
User Profile Reset: Yes
OpenGL enabled: Yes
Version: 220.127.116.11 (x64)
Build ID: 8061b3e9204bef6b321a21033174034a5e2ea88e
CPU threads: 12; OS: Windows 10.0 Build 18363; UI render: Skia/Vulkan; VCL: win
Locale: en-GB (en_GB); UI: en-GB
It is working like this:
You need letter minus minus letter -> and this will work
For example, type this:
You can't--I assure you -> it will work
Yes, it is working for the use-case where the em-dash is between two letters with no spacing (which is the standard for AP style).
However, that's not the only use-case for em-dash. The one I've mentioned in my original bug report (indicating an interruption of dialogue or narrative) is a also a standard use of em-dash, predominantly in fiction or transcribed speech (e.g. Chicago Manual of Style). Other word processors will always automatically convert double hyphens to em-dash, regardless of whether they are preceded/followed by a letter or numeral, an end quotation mark or other punctuation, a space, or a new line.
So what I'm saying is not that it's a bug in the sense of the code performing incorrectly, it's a bug in that the defaults are only set up to account for one way of using em-dash in English, and this is unusual enough compared to e.g. Google Docs or Scrivener to stand out, particularly to anyone who writes fiction or does transcription work.
Ok, UX Team will analize your bug very soon.
No, we have to watch excessive changes to the auto-correct table especially regex globals like this.
Adding a default auto-correction of ".*--.*" to U+2013 (–) would disrupt the far more common "-->" correction to U+2192 (→)
Users are free to make their own adjustments to the auto-correct table.
IMHO => WF
What you are showing is an en-dash, normally em-dash is used for such things. I think a triple dash --- could be automatically converted to em-dash, but as already stated, you can easily set that up yourself. The current behaviour in Writer is to convert -- to em-dash between letters and to en-dash between numbers.
(In reply to lomacar from comment #5)
> What you are showing is an en-dash, normally em-dash is used for such
> things. I think a triple dash --- could be automatically converted to
> em-dash, but as already stated, you can easily set that up yourself. The
> current behaviour in Writer is to convert -- to em-dash between letters and
> to en-dash between numbers.
Not sure that is correct, rather there is no difference in the "--" replacement as 'en-dash' (U+2013) when used between characters or numbers. I just verified that is true with current master in Writer, Calc and Impress.
Looking in source, can't find a use of the U+2014 'em-dash' for auto-correction of two dashes. We do pick up em-dash in a few locales for edit-shell.
And otherwise provide an 'emoji' style "::" auto-correction, i.e. entry o ":---:" becomes '—' (U+2014)
(In reply to V Stuart Foote from comment #4)
> IMHO => WF
Agreed. Have also seen many reports where auto replacement was considered a bug, so we should be careful with .* in general.