2025/09/22

Newest at the top

2025-09-22 19:31:18 +0200Googulator(~Googulato@2a01-036d-0106-0744-4151-2ee3-78cb-4ad2.pool6.digikabel.hu) (Quit: Client closed)
2025-09-22 19:30:55 +0200Googulator9(~Googulato@2a01-036d-0106-0744-4151-2ee3-78cb-4ad2.pool6.digikabel.hu)
2025-09-22 19:28:03 +0200Axma58360(~Axman6@user/axman6) (Ping timeout: 240 seconds)
2025-09-22 19:24:34 +0200Axman6(~Axman6@user/axman6) Axman6
2025-09-22 19:24:21 +0200gorignak(~gorignak@user/gorignak) gorignak
2025-09-22 19:24:04 +0200gorignak(~gorignak@user/gorignak) (Quit: quit)
2025-09-22 19:23:01 +0200kuribas(~user@2a02-1810-2825-6000-fedd-5dda-6057-f5dc.ip6.access.telenet.be) (Ping timeout: 255 seconds)
2025-09-22 19:22:39 +0200ljdarj(~Thunderbi@user/ljdarj) ljdarj
2025-09-22 19:21:38 +0200merijn(~merijn@host-vr.cgnat-g.v4.dfn.nl) merijn
2025-09-22 19:17:12 +0200jmcantrell_jmcantrell
2025-09-22 19:17:12 +0200jmcantrell_(~weechat@user/jmcantrell) jmcantrell
2025-09-22 19:16:16 +0200 <Franciman> i see
2025-09-22 19:14:23 +0200jmcantrell(~weechat@user/jmcantrell) (Quit: WeeChat 4.7.1)
2025-09-22 19:14:19 +0200 <dminuoso> And its not because the problem domain is complicated: its just that all that fixed point nonsense is very much adhoc.
2025-09-22 19:13:52 +0200 <dminuoso> Long story short: One is bad, you should probably use the other, and use a bunch of magic undocumented library functions to make it really extensible.
2025-09-22 19:13:28 +0200 <dminuoso> 8 year old issue about haskellPackages.extend and haskellPackages.override
2025-09-22 19:13:19 +0200 <dminuoso> https://github.com/NixOS/nixpkgs/issues/26561
2025-09-22 19:13:17 +0200 <dminuoso> The best example is this:
2025-09-22 19:13:13 +0200 <dminuoso> But its an absurd abstraction because it leaks implementation at every step of the way
2025-09-22 19:13:00 +0200 <dminuoso> At the end its all done to compose things together
2025-09-22 19:12:51 +0200 <dminuoso> Sometimes its about attrsets
2025-09-22 19:12:51 +0200 <Franciman> i see
2025-09-22 19:12:49 +0200 <dminuoso> Sometimes its about config
2025-09-22 19:12:47 +0200 <dminuoso> Sometimes its about a singular derivation
2025-09-22 19:12:28 +0200 <Franciman> i thought fixpoints was to compute the final derivation
2025-09-22 19:12:20 +0200 <Franciman> ahh i see
2025-09-22 19:12:19 +0200 <dminuoso> In case of NixOS modules config fixed point, its how modules can talk to each other to figure out whats what
2025-09-22 19:12:16 +0200 <inline> on almost every update i saw haskell packages... looks like a fast moving thing
2025-09-22 19:11:59 +0200 <dminuoso> In various degrees.
2025-09-22 19:11:55 +0200 <dminuoso> Franciman: So the fixed points that occur in nixpkgs allow for composability of "things"
2025-09-22 19:11:54 +0200 <inline> and once i installed haskell in there the updates won't ever end....
2025-09-22 19:11:19 +0200 <Franciman> i'm not sure i follow the comment on fixed point logics, wdym dminuoso ?
2025-09-22 19:11:15 +0200 <inline> so 1. level flaky, 2. level flaky due to previous or on it's own, 3. level stable
2025-09-22 19:10:43 +0200 <Franciman> dminuoso: you think we would need full prolog power or just a datalog dialect?
2025-09-22 19:10:21 +0200 <inline> i had 3 levels of encryption, the one via bitlocker (it looses the device from time to time) then i had an encryption level from the VBox, then another of the virtual disk with cryptsetup in the OS
2025-09-22 19:10:21 +0200tzh(~tzh@c-76-115-131-146.hsd1.or.comcast.net)
2025-09-22 19:10:06 +0200 <dminuoso> s/define/avoid/
2025-09-22 19:09:56 +0200 <dminuoso> Software developers have been solving these problems for 5 decades.
2025-09-22 19:09:46 +0200 <dminuoso> Franciman: All that fixed point logic is just trying very hard to define module/packaging/interface boundaries
2025-09-22 19:09:45 +0200 <Franciman> lol
2025-09-22 19:09:00 +0200 <dminuoso> Neither usable nor popular, sounds we could torture a PhD student with it.
2025-09-22 19:08:58 +0200 <Franciman> prolix sounds good
2025-09-22 19:08:55 +0200 <inline> seems like when encryption of devices is the case and windows suddenly looses the device (even though it was initially opened via passphrase and ran 2-3 days)
2025-09-22 19:08:52 +0200 <Franciman> maybe we can inject the fixed point shit about making derivation into the classic fixpoint semantics of logic programs?
2025-09-22 19:08:28 +0200 <dminuoso> Lets make that a thing.
2025-09-22 19:08:23 +0200 <dminuoso> Franciman: Oh prolix!
2025-09-22 19:08:05 +0200 <Franciman> davean: i would like to see a declarative language for nix-style stuff using logic programming paradigm
2025-09-22 19:08:00 +0200 <dminuoso> Its not as if helps readability.
2025-09-22 19:07:54 +0200 <dminuoso> I dont even know why nix people even botter with CR/LF.
2025-09-22 19:07:35 +0200 <inline> i now even updated virtualbox