I’m attempting to use Obsidian on my Dropbox root folder (half a million files, only 400-3000 or so markdown files, but e.g. 130,000 PNGs. Performance-wise, this actually works surprisingly well, except that [[ completion is useless, since it includes all these “attachments” in the search.
Would it be possible to add a setting to include only markdown files in this search? Maybe, as a compromise, “attachments” could also be listed if they’re “near” the focal markdown file in the file tree (say, in the same folder, or a deeper folder (perhaps just one folder deeper))?
Also thinking about this usage scenario with Obsidian being “file explorer” replacement, but currently different obstacles, including the one you described, prevent me from diving into such laborious “project”.
As a philosophical motivating push for this: One of the great advantages of a markdown-based system like Obsidian (or Zettlr or Foam or …) over a database-based system (like Roam or innumerable others) is lack of silo lock-in. But having to keep all my markdown files and their attachments–and nothing more!–in a single folder on my hard drive is a mini-silo (albeit a much less restrictive one).
It would be great if Obsidian (and Zettlr and Foam!) thought of themselves more as crawlers (like the Google crawler), and less the absolute owners of a neatly-defined data fiefdom.
Is there any hope of adding this option? This is really the thing that prevents me from using Obsidian seriously. I’m imagining a menu option like “Include these extensions in search:” with a default entry of “md,pdf,png”, or maybe just “” meaning all files are included in the search. An advanced version could accept a regular expression that must match the full file basename (or even full file path), but it would be enough split the list by comma, and include only files (case insensitive) end in those characters.
This might be useful for speeding up Obsidian for other tasks on large file trees, but this should at least be an option for paring down the autocomplete list.
(An example of another case where excluding files by extension might make large file trees workable is for following links–it would be useful if ctrl+click on wikilinks only considered a subset of my files when deciding whether the link destination exists, or a new markdown file should be created.)
+1
When creating MOCs and searching internal notes by doing [[searchterm , a lot of attachments and picture files get in the way and drown out the .md notes that I’m looking for.