2025/11/17

Newest at the top

2025-11-17 17:18:21 +0100 <lucabtz> i thought there is a way to have that
2025-11-17 17:18:10 +0100 <lucabtz> while mine looks weird
2025-11-17 17:18:04 +0100 <merijn> tomsmeding: :)
2025-11-17 17:18:01 +0100 <lucabtz> idk because some people on the bridge in matrix have a proper name
2025-11-17 17:17:59 +0100 <tomsmeding> I cannot believe my luck
2025-11-17 17:17:53 +0100 <tomsmeding> merijn: it has an install-includes!!!!!!
2025-11-17 17:17:39 +0100 <merijn> lucabtz: You don't, afaik?
2025-11-17 17:17:19 +0100 <lucabtz> how do you connect a matrix account to a IRC one?
2025-11-17 17:17:00 +0100 <tomsmeding> makes sense, thanks!
2025-11-17 17:16:51 +0100 <merijn> tomsmeding: At any rate, the answer is basically IFF they include it in install-includes you can, else you're hosed
2025-11-17 17:16:40 +0100 <tomsmeding> so I believe my only option is writing a fromListNChecked that does almost the same as fromListN but not quite
2025-11-17 17:16:14 +0100 <tomsmeding> possible, but I want a checked fromListN and I can't have that now, because doing a length check before calling V.fromListN forces the entire list and breaks all streaming
2025-11-17 17:15:44 +0100 <merijn> tomsmeding: Which doesn't support any kind of error flow
2025-11-17 17:15:35 +0100 <merijn> tomsmeding: That behaviour is because those functions seem like they're intended to be use with OverloadedList?
2025-11-17 17:15:23 +0100 <pounce> if i do readelf -e /lib64/libtinfo.so.6 | grep VER the library has VERSYM and VERNEED but not VERDEF
2025-11-17 17:15:03 +0100 <pounce> Clint: well, libtinfo might not have those versions. but im not an expert on linkers so im not sure
2025-11-17 17:14:45 +0100 <tomsmeding> (concretely: the fromListN on vectors just ignores additional elements in the list and returns a shorter vector if the list was too short; I think that's boneheaded behaviour and want an error to be thrown instead)
2025-11-17 17:14:43 +0100 <merijn> tomsmeding: https://cabal.readthedocs.io/en/stable/cabal-package-description-file.html#pkg-field-install-inclu…
2025-11-17 17:13:50 +0100Zemy(~Zemy@72.178.108.235) (Ping timeout: 256 seconds)
2025-11-17 17:12:13 +0100 <tomsmeding> to be robust against future changes of that header file, I'd ideally like to be able to use their macro definition
2025-11-17 17:11:51 +0100 <tomsmeding> and they have a little header file (include/vector.h) that defines PHASE_FUSED as [1], and then they write {-# INLINE PHASE_FUSED foo #-}
2025-11-17 17:11:25 +0100 <tomsmeding> merijn: I'm trying to write a new function on `vector` vectors that works with their fusion system
2025-11-17 17:10:54 +0100 <merijn> In short: What are you trying to do?
2025-11-17 17:10:50 +0100Googulator48Googulator
2025-11-17 17:10:45 +0100 <merijn> I guess that also depends whether you mean including it in C or including it in Haskell code, I suppose
2025-11-17 17:10:45 +0100Googulator48(~Googulato@85-238-67-234.pool.digikabel.hu)
2025-11-17 17:10:42 +0100Googulator(~Googulato@2a01-036d-0106-0231-4475-80b4-5cdc-43d6.pool6.digikabel.hu) (Quit: Client closed)
2025-11-17 17:10:03 +0100Zemy_(~Zemy@2600:100c:b0a9:8436:1c98:a0ff:fed8:7a71)
2025-11-17 17:09:58 +0100 <merijn> What that location is, is distro-dependent
2025-11-17 17:09:48 +0100 <merijn> tomsmeding: cabal explicitly supports installing header files in a findable location
2025-11-17 17:09:34 +0100 <merijn> tomsmeding: "It Depends (TM)"
2025-11-17 17:09:08 +0100 <tomsmeding> is it possible to #include a header file using CPP from a _different_ package?
2025-11-17 17:07:26 +0100 <Clint> so haskeline and terminfo require versioned symbols and libtinfo has those symbol versions and it's still complaining?
2025-11-17 17:05:23 +0100 <pounce> ghc isn't from the repo
2025-11-17 17:05:10 +0100 <pounce> Clint: ghc was built against libtinfo on this machine
2025-11-17 17:01:33 +0100 <Clint> presumably you could just rebuild ghc against your libtinfo and it would be happy
2025-11-17 16:47:02 +0100 <pounce> which is where the libtinfo.so.6 mentioned comes from
2025-11-17 16:46:28 +0100 <pounce> i have ncurses-devel and ncurses-libs and ncurses-compat-libs
2025-11-17 16:44:38 +0100 <Square2> missing libtinfo-dev?
2025-11-17 16:44:11 +0100 <pounce> but idk how to fix this :/ it's repo managed
2025-11-17 16:43:14 +0100 <pounce> the only thing i can gather is it might be missing VERDEF
2025-11-17 16:42:22 +0100m1dnight_(~m1dnight@d8D861A17.access.telenet.be) (Ping timeout: 256 seconds)
2025-11-17 16:42:04 +0100 <merijn> Sounds like the libtinfo shipping with your fedora might just be borked?
2025-11-17 16:41:45 +0100 <pounce> merijn: yes, they are
2025-11-17 16:41:39 +0100vardhan(~vardhan@122.172.80.68) (Ping timeout: 265 seconds)
2025-11-17 16:41:38 +0100 <Square2> A quick googling seems to yield a bunch of results that are not tied to ghc/haskell.
2025-11-17 16:41:34 +0100 <merijn> Those look like linker errors to me
2025-11-17 16:38:50 +0100 <Square2> okok
2025-11-17 16:38:42 +0100 <pounce> but also using cabal or stack or anything else has the same effect
2025-11-17 16:38:30 +0100 <pounce> im specifically writing agda, which does call ghc directly