2025/11/03

Newest at the top

2025-11-03 15:22:09 +0100Frostillicus(~Frostilli@pool-71-174-119-69.bstnma.fios.verizon.net) (Ping timeout: 244 seconds)
2025-11-03 15:19:54 +0100 <Leary> My guess: It means that the ghc cabal is using has non-profiling versions of the libraries registered in its global package database, which get in the way of it building its own.
2025-11-03 15:19:49 +0100 <tomsmeding> normally cabal would realise that a particular dependency has not been built for profiling yet, and start doing so
2025-11-03 15:19:23 +0100 <dminuoso> Since it is just plain cabal-install
2025-11-03 15:19:19 +0100 <dminuoso> This is why my question was rather about what cabal is barking about rather.
2025-11-03 15:18:51 +0100 <tomsmeding> probably, and where I sign off because I have no clue
2025-11-03 15:18:41 +0100 <dminuoso> Would this be the time where I admit that I may be using haskell.nix?
2025-11-03 15:18:38 +0100 <tomsmeding> if nuking the store doesn't help, that rules this out
2025-11-03 15:18:15 +0100 <tomsmeding> because this looks like a borked install of parser-combinators and attoparsec in the store
2025-11-03 15:18:02 +0100 <dminuoso> why would I have to nuke stuff from cabal store?
2025-11-03 15:17:36 +0100 <tomsmeding> dminuoso: have you tried rm -r ~/.cabal/store/ghc-9.6.6*? :p
2025-11-03 15:15:12 +0100prdak(~Thunderbi@user/prdak) prdak
2025-11-03 15:14:44 +0100 <dminuoso> Is GHC looking for .p_hi files or soemthing along the lines?
2025-11-03 15:14:13 +0100 <dminuoso> What does "installed the profiling libraries" even mean, here?
2025-11-03 15:12:10 +0100 <dminuoso> I just wanted to debug a performance regression, so I tried `--enable-profiling` and cabal is failing to find modules for an internal dependencies: https://paste.tomsmeding.com/23kpMrLR
2025-11-03 15:08:52 +0100prdak(~Thunderbi@user/prdak) (Read error: Connection reset by peer)
2025-11-03 15:05:01 +0100deptype(~deptype@2406:b400:3a:73c2:422:9c14:cc52:da31)
2025-11-03 15:04:48 +0100deptype(~deptype@2406:b400:3a:73c2:c5bd:c4ff:b106:a7ad) (Remote host closed the connection)
2025-11-03 15:03:03 +0100trickard_(~trickard@cpe-52-98-47-163.wireline.com.au)
2025-11-03 15:02:50 +0100trickard_(~trickard@cpe-52-98-47-163.wireline.com.au) (Read error: Connection reset by peer)
2025-11-03 15:01:58 +0100prdak(~Thunderbi@user/prdak) prdak
2025-11-03 15:00:47 +0100 <sshine> Morj: https://www.icann.org/en/contracted-parties/registry-operators/services/reserved-names/two-charact…
2025-11-03 15:00:37 +0100 <sshine> Mroj: it seems that ICANN changed their policy and generally allow two-letter gTLDs that are not ccTLDs.
2025-11-03 14:59:49 +0100 <dminuoso> First domain shall be category.hask.
2025-11-03 14:59:33 +0100 <dminuoso> .hask is available.
2025-11-03 14:59:30 +0100 <haskellbridge> <Morj> I propose a reunification of serbo-croatia, the tld would be an abbreviation of hrvatsko-serbsko
2025-11-03 14:58:56 +0100 <dminuoso> But yeah.
2025-11-03 14:58:28 +0100 <dminuoso> Morj: Well, one of the imperial bags of cash is just bribing someone at ICANN.
2025-11-03 14:57:40 +0100 <haskellbridge> <Morj> But a gtld can't be two letters, right?
2025-11-03 14:57:40 +0100 <dminuoso> Oh and you need to pay an annual fee of 3 lumps of gold.
2025-11-03 14:57:17 +0100 <dminuoso> Morj: gTLDs are a thing! No need for a full country anymore, you just need about 2 imperial bags of cash.
2025-11-03 14:57:15 +0100 <haskellbridge> <Morj> :3
2025-11-03 14:57:08 +0100Frostillicus(~Frostilli@pool-71-174-119-69.bstnma.fios.verizon.net)
2025-11-03 14:56:37 +0100 <sshine> it already exists :)
2025-11-03 14:56:29 +0100 <sshine> .fun?
2025-11-03 14:56:04 +0100 <haskellbridge> <Morj> Should we create a new country to get an "hs" top-level domain? Or maybe another would work better?
2025-11-03 14:55:30 +0100 <haskellbridge> <Morj> Yes, those
2025-11-03 14:55:24 +0100 <tomsmeding> hyphens?
2025-11-03 14:55:13 +0100Frostillicus(~Frostilli@pool-71-174-119-69.bstnma.fios.verizon.net) (Ping timeout: 264 seconds)
2025-11-03 14:54:58 +0100 <haskellbridge> <Morj> For rust instead I type docs.rs/packagename and it also works, except some packages use underscores instead of tirets in the name
2025-11-03 14:54:12 +0100 <haskellbridge> <Morj> I have search keywords in firefox: h for hoogle and ha for hackage, very useful
2025-11-03 14:53:20 +0100tomsmeding. o O ( https://xkcd.com/1053/ )
2025-11-03 14:53:11 +0100 <[exa]> ok cool. :D
2025-11-03 14:52:57 +0100[exa]discovers hackage search
2025-11-03 14:51:21 +0100prdak(~Thunderbi@user/prdak) (Ping timeout: 250 seconds)
2025-11-03 14:51:10 +0100prdak1(~Thunderbi@user/prdak) (Read error: Connection reset by peer)
2025-11-03 14:49:24 +0100 <Leary> [exa]: Just hackage search, like they said.
2025-11-03 14:49:16 +0100 <dminuoso> And it must facilitate particular access patterns. Maybe you care about inserting a lot
2025-11-03 14:48:33 +0100 <dminuoso> Im not sure a generic version is even that useful. You tend to pick a trie for performance reasons, and then you want it tailored to your algorithm.
2025-11-03 14:47:27 +0100 <dminuoso> I looked for trie packages on hackage, there is not much generic stuff up there.