| Message ID | 86pmoucl3g.fsf@163.com |
|---|---|
| State | Accepted |
| Headers | show |
| Series | [bug#51747] : gnu: nix: Update to 2.4. | expand |
| Context | Check | Description |
|---|---|---|
| cbaines/applying patch | fail | View Laminar job |
| cbaines/issue | success | View issue |
Zhu Zihao schreef op za 15-01-2022 om 00:51 [+0800]: > Patches updated. > > The test is disabled by default, when I tried enable it test it fails, > It looks like that the RPATH of test binary is not complete. Just leave > it disabled. I didn't see anyting RUNPATH (*) related. It failed to run some python script though with ‘file does not exist’ (which could maybe indicate RUNPATH problems, but not in this case), it turned out that the shebang wasn't patched. Adding python to the native-inputs appears to fix things > I have no idea to gi.ve user a hint of differnce between cpuid and > libcpuid. Maybe add a comment or add it to description. But I don't > think user can find these hints in first sight. See attachement for a variant of the 'libcpuid' package that builds for me (--system=x86-64-linux, with tests) and cross-builds to --target=i586-pc-gnu (tests automatically disabled because of cross- compilation). 'kmod' only exists on Linux, so I modified the inputs and 'absolutize' to only use it when building for Linux. I don't think many users look at the description either, but it seems a logical place to look in case of trouble. I added a sentence. I modified the package definition outside emacs and outside a git checkout, so the indentation probably needs some fixing. Greetings, Maxime. (*) IIRC Guix uses RUNPATH and not RPATH.
Maxime Devos <maximedevos@telenet.be> writes: > [[PGP Signed Part:Undecided]] > Zhu Zihao schreef op za 15-01-2022 om 00:51 [+0800]: >> Patches updated. >> >> The test is disabled by default, when I tried enable it test it fails, >> It looks like that the RPATH of test binary is not complete. Just leave >> it disabled. > > I didn't see anyting RUNPATH (*) related. It failed to run some python > script though with ‘file does not exist’ (which could maybe indicate > RUNPATH problems, but not in this case), it turned out that the shebang > wasn't patched. Adding python to the native-inputs appears to fix > things I use -K to keep failed build and ldd the test binary found that RUNPATH is broken, but I miss that we don't add python into inputs.. Penny wise and pound foolish LOL. I know Guix use RUNPATH, but RPATH is short to type and we need not to distinguish between them in our context.
From d81e689fa1acd5eb4a2fee325609216a24a9b887 Mon Sep 17 00:00:00 2001 From: Zhu Zihao <all_but_last@163.com> Date: Fri, 14 Jan 2022 17:30:22 +0800 Subject: [PATCH 6/6] gnu: nix: Update to 2.5.1. * gnu/packages/package-management.scm(nix): Update to 2.5.1. --- gnu/packages/package-management.scm | 4 ++-- 1 file changed, 2 insertions(+), 2 deletions(-) diff --git a/gnu/packages/package-management.scm b/gnu/packages/package-management.scm index 594d6e9d4b..9882f0a411 100644 --- a/gnu/packages/package-management.scm +++ b/gnu/packages/package-management.scm @@ -662,7 +662,7 @@ (define-public guix-icons (define-public nix (package (name "nix") - (version "2.4") + (version "2.5.1") (source (origin (method git-fetch) @@ -671,7 +671,7 @@ (define-public nix (commit version))) (file-name (git-file-name name version)) (sha256 - (base32 "1abgfw5ndqklm0x533li32l4azifz3f6lhaxm6s74b704043r7m2")) + (base32 "1m8rmv8i6lg83pmalvjlq1fn8mcghn3ngjv3kw1kqsa45ymj5sqq")) (patches (search-patches "nix-dont-build-html-doc.diff")))) (build-system gnu-build-system) -- 2.34.0