GNOME’s tracker is in many ways a negative program, especially due to its resource consumption and the fact that idexes things. I never figured out how to disable it (I have been able to uninstall it for some years, until it became a mandatory dependency).
I believe that GNOME should seriously think of ways to make tracker an optional component, especially given the idiosyncrasy of a lot of people in the free software community (me included) against anything that indexes data (even when this data stays sealed in the user’s hard disk), or is wasteful of resources. On the other hand, I believe that when a user chooses to have it, tracker should be available and do its job.
That view is not really consistent with the word “software” and I think you should reconsider. If you have ever used a search engine, or a search feature in some application, or a database, then you probably do not actually hold that view. The modern computer, from its developmental roots in tabulating punch card machines, is a tool mainly for computing, storing, collating and indexing data. It’s why we have things like data structures to begin with. The field is actually called “information technology”…
A search engine is a dedicated software + hardware whose job is addressing a huge amount of search requests and maximazing its performance in that. That greatly moves the balance of wasting energy for indexing vs. wasting energy for searching everything from scratch in favor of wasting energy for indexing.
My laptop on the other hand is not a dedicated search engine. Its resources are used most of the time for doing whatever else, and only seldom for searching. A program like tracker instead decides that a certain amount of my laptop’s resources should be used for searching (indexing) in any case.
Thank you for the link. The problem in my opinion is this:
Tracker is a core dependency of GNOME, and some things will not work as expected if you disable it completely.
Designing a GNOME desktop in which tracker can be disabled means that the only inconvenience when tracker is disabled is a longer waiting time for searching, but not that “things will not work as expected” when tracker is disabled.
Yes, that page could be more specific about what would actually break. The problem of putting info there is then it might go out of date, and IMO out of date info is worse than no info at all.
Here’s what I think would happen in practice if you set tracker-miner-fs-3 to index no locations at all:
Nautilus will only search filenames, not content, and will not use any index
GTK file open dialog will do the same
GNOME Music will show no music
GNOME Photos will show no photos
GNOME Videos will show no videos
Possibly a few other things i’m forgetting.
As discussed many times (this post is now 20 years old!) every “option” doubles the cost of integration testing. If we support “filesystem indexing enabled” and a “filesystem indexing disabled” variants, we now have to test both.
Indeed, but the trade-off is the search is faster when you do actually need to do it. The idea of a file index is not exactly a new idea either, locate and updatedb have been around in Unix for 40 years, and there is a version in GNU… Personally I see Tracker as a more advanced version of those commands.
If that’s all it sounds more than acceptable to me.
I understand that, but forcing the user’s CPU to work is not a light decision like it might be the choice of the look of windows in GNOME. In the second case it is perfectly legit for GNOME to say “this is how we want our desktop to look like, and that’s all”, but in the first case I wouldn’t go as bold. And so “double testing” sounds a bit like a duty for cases like these.
It is not an “evil” thing. It is just a thing that some people will never like to have around.
I believe I search in the content of files more or less once a month, in average. I still use the now archived gnome-search-tool for that. As much as I might need once a month to search in the content of files I do not want that the content of my files gets indicized, ever.