The autocompletion in CM6 does a fantastic job of delivering lightning fast results even when its given an array of thousands of lengthy strings.
Sometimes the results are a bit too fuzzy for certain contexts though, since the results can include the individual tokenized characters of the query spread out over many words.
Is there any way to limit the amount of fuzziness for specific contexts?
There’s not currently any such configuration. What would be an example of a case where it is too fuzzy? And does it properly sort ‘bad’ matches below better matches?
Query: hair
Hairstyle
chair
How to hide a window until it is ready
A very small chair
The sorting has worked well most of the time. But in cases like the one above, a result like ‘chair’ can actually fall below the longer, fuzzier result if the word chair is not at the start of the candidate.
There was an issue where “by-word” matches (where the input matches the first letter of a word) were scored very high even though a bunch of words weren’t matched. This patch fixes that, and should move the long completion to the bottom in your example. Does that help?
Also, you bolded the “i” in “until” in your example, but I’m seeing it match the in “it”. Was that a transcription error or is it actually looking like that in your tooltip?
The patch helps a lot. Thank you. I think the new sorting order eliminates the need to configure fuzziness for my use case. I’ll let you know if anything changes once I’ve had a chance to play with it some more.
And yes that was a transcription error.