Obsidian version: v1.4.14
Installer version: v1.4.12
Operating system: Windows 10 Home 10.0.22621
Login status: logged in
Catalyst license: none
Insider build toggle: off
Live preview: on
Legacy editor: off
Base theme: dark
Community theme: none
Snippets enabled: 0
Restricted mode: on
Thanks @Alicila for linking my forum post and thanks @CawlinTeffid for pointing to the Feature Request (I would like to double the request for such a switch!!).
I’d like to ask for some clarification on the algorithm’s spec… and maybe we are not talking about the same thing here: So both, @Alicila and me are talking about the Wiki-link autocomplete (which is not the same thing as the Quick Switcher) and both of us also have significantly less than 10’000 notes (Alicila writes they have ca. 7700+ files (including attachments) and I have even less: ca. 3600+ files plus ca. 700+ attachments).
@Alicila : how is QuickSwitcher behaving in your case? In my case it still works (and importantly: without any performance issues!!). Just the Wiki-Link Modal is behaving differently, which I really don’t understand why.
From the information given, I’m deducing that there are not one but two “performance optimizations” in place, one for QuickSwitcher and one for Wiki-Links Modals, however, with different thresholds… May somebody of the mods/devs check this hypothesis?
I have a large vault. I have many notes with filenames with spaces in them. I use the Various Complements plugin. I can set trigger for 2 or 3 words in Vault Complement and not remember not getting a find. On the plus side, no need to start with [[, either. Give it a shot, if you like the sound of all this.
I use it for custom dictionaries as well.
Thanks @WhiteNoise. May you elaborate a bit further on “searchable items”: How about wiki-links to non-existing notes? How about aliases? Do they all count, too?
Is this documented somewhere? Moreover, for a standard user without coding-skills, I believe there is no way to get a rough estimate on how many “searchable items” there are in their vault… and hence no information to deduce how far they are from the 10k limit.
I don’t understand this (and here I’m just doubling what @Alicila has already brought into the discussion): On my computer quick switcher is very fast and accurate in performing fuzzy searches. However, the wiki-link autocomplete modals are not. How should this be just due to computer power rather than due to a difference in the search algorithms?
Are the algorithms open source? If not, it would still be important to know: Are the search algorithms qualitatively dependent on computer speed or not? Or in other words: Would they deliver the same results on every machine (if you want, say, after waiting for 5 seconds)?
To underline the urgency of our request: It is not about the time it takes for the algorithm to dump the results. Much more it is about reliability. We need to know, whether the note we are expecting to retrieve via wiki-link-autocomplete does or does not exist. Otherwise, this will quickly end up in lots of notes with similar names. This is especially true, since the “optimised” algorithm is sensitive to spaces and dashes.