I repurposed one of his old laptops (that overheats, sadly) as frontend, making room for an Ubuntu partition.
I installed the trusty #recoll to index text documents and emails.
For images, I ended up using #KPhotoAlbum, that has very flexible tagging possibilities, making it useful on more than photo albums. Chronological view and image stacking are also great.
I used #Geeqie, then more recently #Czkawka for finding image #duplicates, that I then stacked manually in KPhotoAlbum.

Damn, #Recoll can be a real life-saver if you're looking for a document in your pile of (hidden) folders/emails/attachments/etc. It indexes pretty much everything and helps dig up stuff. Neat!

https://www.lesbonscomptes.com/recoll/pages/index-recoll.html

#Recoll l’eina de cerca de text complet per a Linux https://edutictac.noho.st/blog/sasogu/2023/03/01/recoll-leina-de-cerca-de-text-complet-per-a-linux/

Just like I am a fan of desktop search engines like #Recoll (bringing the power of something like Google to my own personal data), I am excited about the potential of the application of LLMs like #ChatGPT *[footnote1] to my own library of articles, books and notes.

"Show me who has written about the rise of ecofascism in 1930s in Germany, and summarize my notes on it."

So, I‘ve been using the combination of #Recoll and #Synapse #Launcher in production for some weeks now and it works satisfying well! Finding my docs as easy as with my #Mac. #gtk #vala

Upstream merge request submitted of course, so I hope this fork is will become part of #Debian once. #Synapse #Recoll

There you go: https://codeberg.org/Letterus/synapse-project/releases @Codeberg

#Debian and #Ubuntu packages available @opensuse. I'm quite happy I now could install this package on my office machine and I'm finding my documents immediately, getting them presented in a pretty way! #Recoll #Synapse #Zeitgeist #FullText #Search

Okay, #Debian #Packaging now. Bwah, I hate it. I probably gonna postpone it to next year. 😁 #Synapse #plugin #Recoll

I think I improved the code by setting a timeout so that #Recoll will only be called when the user stops typing. As you can see via the logged lines in the terminal window this reduces the number of calls to Recall massively and thus the load on the machine. #Synapse

Code is here @Codeberg: https://codeberg.org/Letterus/synapse-project/src/branch/recoll_integration/src/plugins/recoll-plugin.vala

Since this is my first time doing concurrent/async programming in #Vala I'm grateful for comments helping to improve those lines. 😉

To add some more information here: This (German) wiki entry provides information which #Debian #packages to install to make #Recoll perform best: https://wiki.ubuntuusers.de/Recoll/

And this article of #LinuxMagazine provides an in-depth explanation of the features of Recoll and why it should be preferred over using #Xapian directly for handling user provided search strings: https://www.linux-magazine.com/Issues/2018/212/Tutorials-Recoll

So, this already looks very promising. #synapse #recoll #integration #endOfTheYearHacking
Time to create a new @Codeberg repo and make a #Debian #package this evening I think.

So, my #FLOSS todo list for the next days:
- get that #Recoll plugin for #Synapse working again
- write that script to make @qownnotes use @pandoc and @libreoffice to generate a PDF booklet
- make that #iBook G4 running again and set up an usable PIM suite using #vdirsyncer, #khal, #khard and #mutt.

If I'm somewhat recreated by that time I'm gonna restart coding on #ObjGTK. But after Christmas I'm really too exhausted to get into deep work too soon. 🤔 🤷‍♂️

Can anyone #recommend a full-text, indexing search for #debian #linux? It seems like there's no singular solution, like Spotlight on OSX.

I have #Akonadi and #Tracker3 running. I've considered #Recoll and #SOLR.

In practice, I wind up using #grep and locate. Grep is a a worst-case personal search engine.

Bonus points if it doesn't require lots of config and if it is lightweight when idle.

This should be easy in 2022. Right?

Just found a true 'gem' of a Desktop #Search solution for #Linux called #Recoll: https://www.lesbonscomptes.com/recoll/pages/index-recoll.html

@nobodyinperson I keep them separated. I organize literature via #Zotero. Sometimes I use #Recoll when I look for a specific information in my library. When I write papers I make a snapshot of the papers a I want so cite. That is when I use #kbibtex alot in combination with #RKWard.

@brecht @floppy That sounds like a good line of thought, however storing the contents is also needed for full text search, and that's precisely what search engines have been doing for this whole millennium without much legal fuss.

The difficulty only came recently with snippets, which are all about showing the contents to some party other than the one which did the scraping.

I think my setup for search and saving using #Recoll-we and #singlefileZ is legally safe.

@kinen Дельный совет был, спасибо. Очень впечатлил #recoll на #xapian движке. Странно, что инфы мало, как будто непопулярный, а ведь индексирует быстро, ищет моментально. На NVME дикие объемы индексировал - все стабильно и предсказуемо, и никаких капризов как у elastic.

#bigdata #search

Recommended #opensource desktop #search: #Recoll

Why? Scan docs, images (Exif), ebooks, calendars... · Inside archives too · Restrict directories · Simple search or complex query · Text summaries · Stemming

More recommendations: https://tuxwise.net/recommended-software/

https://www.lesbonscomptes.com/recoll/

@ashshuota SingleFileZ, which saves to a place indexed by #Recoll.

Dommage, mais je renonce à #Recoll comme moteur de recherche local. Sympathique mais trop gourmand en ressources, trop instable.
=> https://recoll.org

En revanche j'ai trouvé un substitut prometteur : #drill-search => https://drill.software/

Ça n'indexe pas (encore) le contenu des fichiers mais ça dépote, quasiment sans rien pomper dans les ressources du système.

À suivre.