16:43 <nekobot> [haikuports] Begasus pushed 1 commit to branch master: https://github.com/haikuports/haikuports/compare/385ab20efba4...37f109102239

16:43 <nekobot> • Begasus (37f10910): LLVM, bump revision for version 18, 19 and 20 (#12209)…

16:44 <Begasus[m]> stress testing tomorrow :)

17:00 <nekobot> • OscarL (5a05c991): intel_microcode: update to release 20250211 (#12228)

17:00 <nekobot> [haikuports] korli pushed 1 commit to branch master: https://github.com/haikuports/haikuports/compare/37f109102239...5a05c991263c

17:52 <Begasus[m]> closing down here

17:52 <Begasus[m]> cu peeps!

17:52 <erysdren> cya Begasus

18:36 <Shaka444> Are we still using ffmpeg as the MediaKit backend?

20:28 <OscarL> https://oftc.irclog.whitequark.org/ still broken :-(

20:30 <OscarL> Last week I was thinking about downloading a copy of the logs. Should have acted sooner :-)

21:11 <^rooker> I've only booted Haiku once from a USB stick, and I fell in love with the "good old" UI-response-snappiness of 80s code, I had almost forgotten.

21:11 <^rooker> Hi everyone!

21:11 <Habbie> hi!

21:12 <erysdren> hiya

21:12 <^rooker> I failed trying to get the network (wifi) device to work though :(... So I gave up, due to lack of time.

21:12 <^rooker> But I'm still intrigued to learn more about the very designs of Haiku/BeOS - especially "xattrs"...

21:13 <^rooker> Anyone eager to share how life feels like with an OS that takes filesystem key/value information seriously and into its design? :D

21:14 <^rooker> Sorry, wrong emoji mapping. I wanted a smiley (colon-D) not lauging-in-tears :D - I'm serious about my question and interest.

21:16 <^rooker> Does anyone know where I can find François Revol? He wrote an interesting article on xattr-namespaces (file:///home/pb/Downloads/dcmi-952135759-2.pdf) in 2011. I've tried contacting him, but got no reply.

21:16 <Habbie> you pasted a local path

21:16 <OscarL> ^rooker: regarding attributes... you can read about TrackerBase (from BeOS days) https://blog.birdhouse.org/2019/01/27/trackerbase/ and https://github.com/shacker/TrackerBase, and more recently, about its use in "SEN": https://discuss.haiku-os.org/t/announcement-sen-personal-knowledge-graphs-book-featuring-haiku/13582

21:16 <^rooker> Sorry: wrong like - https://dcpapers.dublincore.org/files/articles/952135759/dcmi-952135759.pdf

21:16 <Habbie> :)

21:17 <^rooker> @oscarL: thanks for the links!

21:18 <OscarL> ^rooker: François Revol is in here as mmu_man, and albeit not usually that active on this channels last few years, still with us :-)

21:19 <^rooker> Oh! Thanks for the info.

21:19 <^rooker> I'm working in the field of digital-preservation. So I'm doing archival-coding and setups for large-scale mixed-data collection handling in institutions world-wide.

21:20 <HaikuUser2> hello

21:21 <^rooker> OscarL: And on my quest for "better annotation ways" - I've come across Object-Storage (claiming to do FS-metadata) - then a friend pointed me to "old school BeOS - now Haiku" as reference. And the article by François is exactly what I'm currently working on.

21:21 <OscarL> :-)

21:22 <^rooker> I'm seriously convinced that it's about time (and the right thing to do) - to finally make use of key/value features like xattrs and offload many "PC-legacy" meta+data issues to the filesystem - as database.

21:22 <OscarL> user grexe (Gregor Rosenauer), the one working on that "SEN" system, is probably the one using the most out of "file system as a database" features nowadays.

21:23 <^rooker> Do you know how/where I could contact him? (preferrably email?)

21:23 <^rooker> You mean: https://discuss.haiku-os.org/u/grexe/summary?

21:23 <OscarL> that user, yes. same handle over github, IIRC.

21:26 <^rooker> okay, found his github - but there's no button to contact him.

21:27 <^rooker> sorry! found an email address there.

21:28 <OscarL> hello HaikuUser2, welcome.

21:32 <OscarL> guess I scared them away... happens regularly enough as to make me think it must be /me issue :-)

21:41 <^rooker> @OscarL: I wrote grexe a "first contact" email. Hope he finds it well.

21:42 <^rooker> My experiments so far, de-embedding embedded metadata from mixed-data-formats folder-trees show how plaintext life could be!

21:43 <^rooker> So what else, aside from "0-byyte files as contact SQLite-replacement" awesome use-cases does Haiku offer with xattrs?

21:48 <OscarL> Not much on the base OS, AFAIK. The system and features are there, but it is up to users to come up with use cases (thus why I mentioned TrackerBase and SEN)

21:48 <phschafft> morning.

21:49 <OscarL> Some audio related apps make use of them instead of, say, id3 tags in music files.

21:51 <OscarL> IIRC, and at least back in the BeOS days, one could insert an audio CD with CD-Text, and be able to mount it as a regular volume, with the tracks appearing as regular .wav files, and having the CD-text data parsed into the file attributes.

21:52 <phschafft> ^rooker: being a bit from a metadata background as well. so maybe there is some time for us to chat. not sure if it is now.

21:52 <OscarL> (so if you "copy/pasted" a .wav file into BFS... you'd not only get the audio track ripped, but with the attributes title/album/etc already filed in)

21:55 <OscarL> bbl

21:55 <phschafft> (It's late for me and I'm a bit out of focus)

21:58 <^rooker> From Gregor's article to PKGs (Personal Knowledge Graphs) to this: https://datacentricmanifesto.org/

21:59 <^rooker> "We have uncovered a root cause of the messy state of Information Architecture in large institutions and on the web today. It is the prevailing application-centric mindset that gives applications priority over data. The remedy is to flip this on its head. Data is the center of the universe; applications are ephemeral."

22:01 <^rooker> @phschafft: If it's late for you, we're in the same timezone :) AT

22:01 <phschafft> I also feel that most people think too syntatic, and too little semantic.

22:02 <^rooker> @paschafft: feel free to write me at haiku.irc2025 AT arkthis.com.

22:03 <^rooker> sorry typo. late here, too.

22:03 <phschafft> e.g. a user doesn't want to use that program. what they really want to do is doing this task. and they may prefer one program over another to do that.

22:03 <phschafft> the difference is that there is this fliping, same as given in your quote.

22:03 <^rooker> @phschafft: Absolutely agree! And I think I've found ways of tackling / transforming exactly that.

22:03 <^rooker> fliping?

22:04 <phschafft> the perspective between application and data (and data transformation)

22:04 <phschafft> e.g. editing a text file is just transforming it from one state to another. no part of that transformation is about what software we use for that.

22:04 <^rooker> I know. I've spent the last 3 years bending my head from "files-in-folders" to "URIs for everything" as ObjectIDs - with metadata finally staying /with/ the object, independent of its data-(file)-format.

22:05 <^rooker> And I'm doing meta/data import/export for almost 20y now.

22:05 <phschafft> software is simply not part of the definition of 0) data nor 1) transformation.

22:06 <^rooker> ...that's why I think offloading database and search/query and relationships-by-links to the filesystem is a good way to go.

22:06 <^rooker> Most use cases for sqlite, csv and excel in general would probably resolve into "right-click-edit-metadata" and local index/search interfaces.

22:06 <phschafft> I thought so as well for many years. currently I'm on (what I think) is the next step and removing the filesystem as a concept ;)

22:07 <^rooker> I'm amazed to see how well Wikidata's data "feels" like and is to work with.

22:08 <^rooker> Imagining "just the same" level of RDF and SPARQL on local filesystem data - even supported by cloud storage by default.

22:08 <phschafft> I have a few problems with it. but our (talking about the company I work with) systems can just use wikidata data natively.

22:09 <^rooker> I'm sure there are issues, yet this is MediaWiki's work - and Wikidata started in 2012 - and is a not-for-profit (AFAIK?) foundation or so.

22:09 <^rooker> And they're serving "the world" quite well on their budget, I guess?

22:09 <phschafft> totally agree.

22:10 <^rooker> Anyways, is there any production-stable way to use haiku's FS (befs?) on Linux?

22:10 <Habbie> my debian trixie (testing, soon to be 13) just mounted befs for me, is what i know

22:10 <^rooker> I'm looking for a sexy filesystem to support theoretically unlimited-size xattrs.

22:10 <phschafft> and they follow the RDF model nearly perfectly. I just feel that the RDF model is a subset of the truth^TM. but it is part of it.

22:10 <^rooker> Debian 13, Trixie - Roger that. Thx.

22:11 <erysdren> Arch Linux can also mount BeFS, but i've never tried writing to it. only reading.

22:11 <^rooker> Well, I'd say "the metadata truths" are always domain-specific and declared by their URIs, all treated equally?

22:12 <^rooker> It might even be "legal" to use "file:///" as URI prefix? So I don't see namespace or ^TM issues there. Or am I not getting you right?

22:12 <phschafft> ^rooker: I think I should have worded that better.

22:12 <phschafft> again, it's late for the two of us.

22:12 <^rooker> @pschafft: Sorry if I said anything wrong.

22:13 <^rooker> Alright, let's call it a day.

22:13 <^rooker> Thanks for engaging, and I'll continue my adventurous quest anyways! :)

22:13 <phschafft> all fine. just that we should maybe move the details to tomorrow ;)

22:13 <^rooker> Goodnight.

22:13 <phschafft> Nachti!

22:13 <^rooker> Pfiat Di.