Hacker News Reader

Comment by eviks | Hacker News Reader

Is there a chance for an Everything alternative? That's the awesome tool that would allow you to ignore the depth/breadth completely find any file literally instantly (the power of instant feedback is unapralleled) Do non-NTFS systems (even the modern ones) not offer this?

View on HN

Just adding the Everything is ridiculously fast - nothing comes close. https://www.voidtools.com/en-au/support/everything/

Anything that accesses the MFT directly will be faster than Everything, since Everything goes through the OS. That said, Everything has a superb design and indeed is extremely fast.

Everything is using the MFT as well.

How about FSearch in Linux? [http://cboxdoerfer.github.io/fsearch/]

Yes, FSearch is the one I use, but it's not as great, per FSearch's dev:

> However, FSearch doesn't automatically detect changes made to the file system and update its index then. This is on the roadmap (it's called inotify support) but it'll never work as smooth as Everything on Windows, because the Linux kernel isn't particularly good at reporting filesystem changes

https://github.com/cboxdoerfer/fsearch/issues/26

Everything is comprehensive + instant + always up-to-date, that's so awesome a combo it's a pity it's Windows only

Doesn't everything need to rebuild its database on every reboot? Was a minor annoyance the last time i remember using it.

No, upon start Everything loads its database file into RAM and queries the USN journal for all file system changes since the last time it was running. By applying those changes to its current database it automatically gets a consistent representation of the current state of the file system. This usually is much quicker than a full rebuild.

Only when the USN journal doesn't go back enough in time will Everything rebuild its database by parsing the MFT (which is also much quicker than traversing the file system).

Okay. My system might have been bugged or misconfigured. It was several years ago.

what about deleted files? Do they eventually get flushed from the cache?

Files get deleted immediately, the USN Journal only keeps track of the events (including delete events). It's basically a log of all filesystem events (up until a certain point in time, due to size restrictions).

The journal has data about deletions, too.

https://learn.microsoft.com/en-us/windows-server/administrat...:

“As files, directories, and other NTFS objects are added, deleted, and modified, NTFS enters records into the USN change journal, one for each volume on the computer.”

Mac OS spotlight has more-or-less instant file search. Command-F in finder or mdfind in the terminal.

Except it's neither instant nor comprehensive, and those two features are what make Everything awesome

I think you never actually used the mdfind search, only the bastardized version that is now available as Spotlight. That interface used to be good, now it’s so bad and slow I don’t know why they let this happen. But try mdfind with filenames, it is incredibly fast. Also for file content searches it is very fast also.

I tried a quick search between mdfind, bfs, and find to locate all README.md files in my home directory. bfs and find both found 2930 files and took 8.5s and 47.5s respectively. mdfind took 0.1s but found only 2400 files.

so what's the query to find a file anywhere on my hard drive, including hidden folders/files?

I don’t know what problems people have with Spotlight. It finds everything instantly for me. It’s my launcher, calculator…

For starters, it hides system files by default.

I think that’s a good default.

Most people don’t need to be messing with system files, the ones who do, can go through the extra trouble of checking a few preferences, using command line tools, setting up plist flags, etc.

Oh so you do know what problems people have with Spotlight, you just don't care about those problems to offer anything useful (your generic list isn't that since there are no preferences, cli tools, plist flags that would bring Everything to life on a Mac)

No, I learned from this problem just now and think most users shouldn’t be messing with Library folder, etc, which is why it is hidden, and rightly so, IMO.

I have not used Everything, is it aomething similar to SwiftSearch (bypasses fs and parses ntfs disk directly, so no indexing and instant results) ?

If the parent is talking about the Windows search software Everything (https://www.voidtools.com/), it does index the files (but unsure of how, it's relatively quick compared to other Windows tools that does the same, so probably some magic involved?) at startup, then continuously indexes (as far as I can tell) but then after that the searches are instant.

Edit: Found some answers (https://www.voidtools.com/faq/):

Re indexing:

> How long will it take to index my files?

> "Everything" only indexes file and folder names and generally takes a few seconds to build its database. A fresh install of Windows 10 (about 120,000 files) will take about 1 second to index. 1,000,000 files will take about 1 minute.

Re changes:

> Does "Everything" monitor file system changes?

> Yes, "Everything" does monitor your file systems for all changes. Your search results will update in real-time to reflect any changes. Everything will automatically keep your NTFS indexes up to date with the NTFS USN Journal. Changes will not be missed when Everything is not running as the system maintains the NTFS USN Journal.

Yes, exactly the principle as SwiftSearch - bypassing the slow fs and getting instant results (but ntfs only), and checking fs logs directly to keep track of all the updates

the 'locate' command is faster than everything but arguably not as convenient (being a commandline tool). It seems to support non-ntfs (meaning it doesn't support ntfs) linux filesystems.

edit: catfish is a gui front-end available for locate.

There's plocate too, which is a bit faster still according to the authors.

locate is much slower than Everything and it also has a much less powerful query language. With Everything you usually get results in way less than 100ms, whereas with locate it's several hundred milliseconds for a database with more than a million files.

Reply on news.ycombinator.com