On 2016-10-03 11:29, Richard Brown wrote:
Hi Carlos,
So you say that applications may use that index. What for? Perhaps, instead of doing a directory search for display on "open file" they use the index?
I'm just curious, wanting to learn things, not saying that it is wrong. I actually use and like the search tool in gnome, called from xfce. I don't notice slowdowns. I forgot that I also have tracker. I don't know how to call it.
I believe the logic used by the GNOME applications in question is that 'users do not care where their files of type BLAH are, they are loading an application to play Music/view photos/watch videos and they need to be shown their Music, Photos, and Videos, regardless of where they are.
I believe this was also described as part of the KDE concept of the 'Semantic Desktop'
Such an approach requires an index
In fact, if you think of the large amounts of Music/Photos/Videos that people have these days, there is really no alternative BESIDES having an index of some type - there is no way in a million years you want your music app to parse through dozens and dozens of music folders for thousands and thousands of music files just to add the two new tracks that someone added later.
I see. Perhaps we need a filesystem type that includes this information in a fast searchable manner, instead of having to create databases ;-)
It's WAY easier and more efficient to have an intelligent indexer like tracker keeping an eye on all directories of potential interest and incrementally updating a single, multi-use index as files change, than requiring half a dozen apps to maintain their own indexes.
Yes, true. I understand that tracker listens on (something?) to detect new files automatically, without searching for them.
It would be nice to have a single search engine used by both gnome and kde. IMHO :-)
I agree, and I think the strongest candidate for such a role would be tracker given it's relative lightweightness compared to the KDE indexers. But the whole KDE stack is so tightly coupled to their chosen indexer, I think we might have a hard time encouraging that kind of change.
I don't care which one is used, as long as it is only one and it works. Or create a new indexer built by both teams, so that it caters to the needs of both. An independent project. Let's hope :-) Anyway, I now understand why applications needs and indexer. Thanks. -- Cheers / Saludos, Carlos E. R. (from 13.1 x86_64 "Bottle" at Telcontar)