Desktop search in KDE -- Baloo configuration, tweaking, functionality, docs?

[Mark Rousell]

René,

Many thanks for your reply.

On 14/04/2020 10:23, René J.V. Bertin wrote:
> On Tuesday April 14 2020 01:27:46 Mark Rousell wrote:
> 
> Hi,
> 
>> In particular at the moment, my requirements/expectations are:
>> (1) Dynamic indexing of new file content and changed file content. I'd
>> expect this to be almost instant on a freshly installed system with only
>> a few files.
>> (2) Search of arbitrary files types from the GUI.
> 
> I can't help you much other than stating that
> - I have gotten the impression that baloo is a bit of abandonware, or that it's victim of a kind of collaborative babel effect where everyone wants something different. Have you checked if there's a more appropriate ML to ask your questions? This particular list sees very little activity.

Oh, that's very interesting and worrying. I know that the former
maintainer of Baloo, Vishesh Handa, has stepped down. He seemed to be
the driving force behind Baloo and, before that, Nepomuk. However, I
hadn't realised that things might have stalled without him.

Before Baloo, Nepomuk was certainly a do everything package (context
indexing, tagging and semantic data store) but I understand it was too
complex and many people complained about the speed (or lack of). Baloo
was introduced to just do file indexing and searching.

However, I can see that people might well have wanted to add features
back in over time, especially data store indexing that Baloo does not
currently cover.

As for other places to ask, the only other place I can think of is the
KDE-Devel list. Not that this is a dev question but I might find someone
who knows what the situation is on there.

I asked before on the Opensuse list but no one knew the answer there. I
wonder if I would get more responses on support forums but I suspect not.

One would hope that Baloo is not actually abandonware since it is the
state of the art of indexing tech in KDE.

> - I still run a Plasma4 desktop myself (with most components updated to the latest 4.14 versions from git), and baloo works there though I usually suspend it because I have no particular need for it and it definitely burns too much CPU for that. I use an advanced KCM (https://gitlab.com/ericlnu/baloo-kcmadv) to control it more finely - but that one has never been updated to KF5 as far as I can tell ... maybe a nice challenge for you? For searching I use the milou (https://cgit.kde.org/milou.git) thingy, and that one *has* been updated for KF5. It gives me an interface (under Plasma4) that is reminiscent of the Spotlight interface in OS X ... no idea how it has evolved under Plasma5 (or devolved: there's a reason I've been sticking to Plasma4).

Thanks for the link to baloo-kcmadv and milou. I note that milou runs on
top of Baloo.

There is also the independent project, Recoll
https://www.lesbonscomptes.com/recoll/, which is very capable indeed and
fulfils almost all of my requirements, except that it is not integrated
into the UI in the way that I would like to see. I suppose I could learn
to live without transparent UI integration.

> Irrelevant, but:
> Then you must have a very fast machine. I've never seen Win10 be anything but slow as molasses. As usual, the only desktop integrated indexing system I have experience with that does not degrade system performance unjustifiably is

I have a 10 year old Core i7-980X. It was very fast in its day but is
distinctly average now. I really don't want to make this yet another OS
war (not least because I want to move *away* from Windows!) but I have
many Windows 10 support customers and it seems plenty fast enough. In
particular, I have never known dynamic content indexing to noticeably
slow it down on any modern hardware.

> Apple's Spotlight (which also gives you a backdoor to control what's being indexed because it's plugin-based).

Windows Search is also fully plugin based. All content and metadata
extractors are plugins and new ones can be installed, existing ones can
be replaced, they can be removed, and they can be enabled or disabled.
The UI gives you control over what file types and data stores are
indexed. It's actually really capable and flexible, and the UI
integration is superb, and is perhaps the only reason I am personally
still using Windows.

Back in the KDE4 era, the whole semantic desktop concept (with Nepomuk
being a core part of it) was supposed to bring a similar level of
integrated indexing and search (and more, with centralised tagging and a
semantic data store capability) to KDE and Linux but it was perhaps too
ambitious. Baloo is actually very capable now in and of itself (within
its more limited aims) but it does seem that UI integration has suffered.

I'll check out milou.

Thanks once again for your reply.



-- 
Mark Rousell