Package: guix;
Reported by: Christopher Baines <mail <at> cbaines.net>
Date: Sun, 9 Feb 2020 19:54:02 UTC
Severity: normal
To reply to this bug, email your comments to 39531 AT debbugs.gnu.org.
Toggle the display of automated, internal messages from the tracker.
View this report as an mbox folder, status mbox, maintainer mbox
bug-guix <at> gnu.org:bug#39531; Package guix.
(Sun, 09 Feb 2020 19:54:02 GMT) Full text and rfc822 format available.Christopher Baines <mail <at> cbaines.net>:bug-guix <at> gnu.org.
(Sun, 09 Feb 2020 19:54:02 GMT) Full text and rfc822 format available.Message #5 received at submit <at> debbugs.gnu.org (full text, mbox):
From: Christopher Baines <mail <at> cbaines.net> To: bug-guix <at> gnu.org Subject: guix pull on aarch64-linux glibc derivation has incorrect output Date: Sun, 09 Feb 2020 19:53:12 +0000
[Message part 1 (text/plain, inline)]
Hey, When attempting to guix pull using the aarch64-linux system, I'm seeing some issues with derivation outputs. I tried with a newer and older commit, and the result is the same. → guix pull --commit=27b09f3ab11a30821a5ce0b071aac1bc6156497d --system=aarch64-linux --profile=/tmp/testprofile2 Updating channel 'guix' from Git repository at 'https://git.savannah.gnu.org/git/guix.git'... Building from this channel: guix https://git.savannah.gnu.org/git/guix.git 27b09f3 Computing Guix derivation for 'aarch64-linux'... - guix pull: error: derivation `/gnu/store/800ky8qa4az7yx36gsg9ak6bih3530qm-glibc-2.29.drv' has incorrect output `/gnu/store/8v34v81q86klja9rihaixkypcml5ad5j-glibc-2.29-debug', should be `/gnu/store/w3iq60ias1qlrjigbj75ssda09hwg21i-glibc-2.29-debug' → guix pull --commit=4cb7c3d6a0e6ee50f8f9f54243402560b8908378 --system=aarch64-linux --no-grafts --profile=/tmp/testprofile Updating channel 'guix' from Git repository at 'https://git.savannah.gnu.org/git/guix.git'... Building from this channel: guix https://git.savannah.gnu.org/git/guix.git 4cb7c3d Computing Guix derivation for 'aarch64-linux'... / guix pull: error: derivation `/gnu/store/800ky8qa4az7yx36gsg9ak6bih3530qm-glibc-2.29.drv' has incorrect output `/gnu/store/8v34v81q86klja9rihaixkypcml5ad5j-glibc-2.29-debug', should be `/gnu/store/w3iq60ias1qlrjigbj75ssda09hwg21i-glibc-2.29-debug' Thanks, Chris
[signature.asc (application/pgp-signature, inline)]
bug-guix <at> gnu.org:bug#39531; Package guix.
(Sun, 19 Apr 2020 22:15:01 GMT) Full text and rfc822 format available.Message #8 received at 39531 <at> debbugs.gnu.org (full text, mbox):
From: Ludovic Courtès <ludo <at> gnu.org> To: Christopher Baines <mail <at> cbaines.net> Cc: 39531 <at> debbugs.gnu.org Subject: Re: bug#39531: guix pull on aarch64-linux glibc derivation has incorrect output Date: Mon, 20 Apr 2020 00:14:50 +0200
[Message part 1 (text/plain, inline)]
Hi, Christopher Baines <mail <at> cbaines.net> skribis: > When attempting to guix pull using the aarch64-linux system, I'm seeing > some issues with derivation outputs. I tried with a newer and older > commit, and the result is the same. > > > → guix pull --commit=27b09f3ab11a30821a5ce0b071aac1bc6156497d --system=aarch64-linux --profile=/tmp/testprofile2 > Updating channel 'guix' from Git repository at 'https://git.savannah.gnu.org/git/guix.git'... > Building from this channel: > guix https://git.savannah.gnu.org/git/guix.git 27b09f3 > Computing Guix derivation for 'aarch64-linux'... - > guix pull: error: derivation `/gnu/store/800ky8qa4az7yx36gsg9ak6bih3530qm-glibc-2.29.drv' has incorrect output `/gnu/store/8v34v81q86klja9rihaixkypcml5ad5j-glibc-2.29-debug', should be `/gnu/store/w3iq60ias1qlrjigbj75ssda09hwg21i-glibc-2.29-debug' The problem here is that we’re building the trampoline, “compute-guix-derivation”, for AArch64. It builds if substitutes are available (likely) and fails to build otherwise. And then we try to execute it locally, and since your machine is not AArch64, it fails. The first patch attached does what I thought was all it would take to fix it. But then I realized that the second patch is needed so that ‘make-config.scm’ uses a Guile for the right system, same for “module-import.drv” and so on. Together, these two patches solve the problem (not retroactively though), but we need to check the implications of changing the default value of #:guile-for-build. Thanks, Ludo’.
[0001-build-self-Always-build-the-trampoline-for-the-curre.patch (text/x-patch, inline)]
From 366729d477960127424076bdac51e7c937857081 Mon Sep 17 00:00:00 2001
From: =?UTF-8?q?Ludovic=20Court=C3=A8s?= <ludo <at> gnu.org>
Date: Sun, 19 Apr 2020 23:14:56 +0200
Subject: [PATCH 1/2] build-self: Always build the trampoline for the current
system.
Previously we'd build 'compute-guix-derivation' for the target system.
Running it would then fail (unless qemu-binfmt support was set up or
something similar).
Fixes <https://bugs.gnu.org/39531>.
Reported by Christopher Baines <mail <at> cbaines.net>.
* build-aux/build-self.scm (load-path-expression): Add #:system and
honor it.
(gexp->script): Likewise.
(build-program): Pass #:system to 'gexp->script'.
---
build-aux/build-self.scm | 24 +++++++++++++++++-------
1 file changed, 17 insertions(+), 7 deletions(-)
diff --git a/build-aux/build-self.scm b/build-aux/build-self.scm
index f86c79f0d0..58ecf20cf4 100644
--- a/build-aux/build-self.scm
+++ b/build-aux/build-self.scm
@@ -148,14 +148,17 @@
;;; 1ae16033f34cebe802023922436883867010850f (March 2018.)
;;;
-(define (load-path-expression modules path)
+(define* (load-path-expression modules path
+ #:key (system (%current-system)))
"Return as a monadic value a gexp that sets '%load-path' and
'%load-compiled-path' to point to MODULES, a list of module names. MODULES
are searched for in PATH."
(mlet %store-monad ((modules (imported-modules modules
- #:module-path path))
+ #:module-path path
+ #:system system))
(compiled (compiled-modules modules
- #:module-path path)))
+ #:module-path path
+ #:system system)))
(return (gexp (eval-when (expand load eval)
(set! %load-path
(cons (ungexp modules) %load-path))
@@ -165,12 +168,14 @@ are searched for in PATH."
(define* (gexp->script name exp
#:key (guile (default-guile))
- (module-path %load-path))
+ (module-path %load-path)
+ (system (%current-system)))
"Return an executable script NAME that runs EXP using GUILE, with EXP's
imported modules in its search path."
(mlet %store-monad ((set-load-path
(load-path-expression (gexp-modules exp)
- module-path)))
+ module-path
+ #:system system)))
(gexp->derivation name
(gexp
(call-with-output-file (ungexp output)
@@ -186,7 +191,8 @@ imported modules in its search path."
(write '(ungexp set-load-path) port)
(write '(ungexp exp) port)
(chmod port #o555))))
- #:module-path module-path)))
+ #:module-path module-path
+ #:system system)))
(define (date-version-string)
@@ -364,7 +370,11 @@ interface (FFI) of Guile.")
#$pull-version)
#:system system)
derivation-file-name))))))
- #:module-path (list source))))
+ #:module-path (list source)
+
+ ;; Always build the trampoline for the local system. The
+ ;; system being targeted will be passed as an argument.
+ #:system %system)))
(define (call-with-clean-environment thunk)
(let ((env (environ)))
--
2.26.0
[0002-DRAFT-gexp-guile-for-build-defaults-to-a-system-depe.patch (text/x-patch, inline)]
From d35b68374543f9eed7dc1f49c73d49a31332e2b3 Mon Sep 17 00:00:00 2001
From: =?UTF-8?q?Ludovic=20Court=C3=A8s?= <ludo <at> gnu.org>
Date: Sun, 19 Apr 2020 23:58:10 +0200
Subject: [PATCH 2/2] DRAFT gexp: #:guile-for-build defaults to a
system-dependent derivation.
Until now, the default #:guile-for-build would be Guile lowered to
whichever system was current when %GUILE-FOR-BUILD was set. This would
break when passing #:system to a value other than %CURRENT-SYSTEM.
* guix/gexp.scm (lower-gexp): #:guile-for-build defaults to #f.
(gexp->derivation): Likewise.
(imported-files/derivation): Likewise.
(imported-files): Likewise.
(imported-modules): Likewise.
(compiled-modules): Likewise.
---
guix/gexp.scm | 12 ++++++------
1 file changed, 6 insertions(+), 6 deletions(-)
diff --git a/guix/gexp.scm b/guix/gexp.scm
index 4ac0411da1..753da8cb4d 100644
--- a/guix/gexp.scm
+++ b/guix/gexp.scm
@@ -784,7 +784,7 @@ second element is the derivation to compile them."
(system (%current-system))
(target 'current)
(graft? (%graft?))
- (guile-for-build (%guile-for-build))
+ (guile-for-build #f)
(effective-version "2.2")
deprecation-warnings)
@@ -885,7 +885,7 @@ derivations--e.g., code evaluated for its side effects."
(env-vars '())
(modules '())
(module-path %load-path)
- (guile-for-build (%guile-for-build))
+ (guile-for-build #f)
(effective-version "2.2")
(graft? (%graft?))
references-graphs
@@ -1357,7 +1357,7 @@ to a tree suitable for 'interned-file-tree'."
#:key (name "file-import")
(symlink? #f)
(system (%current-system))
- (guile (%guile-for-build)))
+ (guile #f))
"Return a derivation that imports FILES into STORE. FILES must be a list
of (FINAL-PATH . FILE) pairs. Each FILE is mapped to FINAL-PATH in the
resulting store path. FILE can be either a file name, or a file-like object,
@@ -1406,7 +1406,7 @@ to the source files instead of copying them."
;; The following parameters make sense when creating
;; an actual derivation.
(system (%current-system))
- (guile (%guile-for-build)))
+ (guile #f))
"Import FILES into the store and return the resulting derivation or store
file name (a derivation is created if and only if some elements of FILES are
file-like objects and not local file names.) FILES must be a list
@@ -1426,7 +1426,7 @@ as returned by 'local-file' for example."
(define* (imported-modules modules
#:key (name "module-import")
(system (%current-system))
- (guile (%guile-for-build))
+ (guile #f)
(module-path %load-path))
"Return a derivation that contains the source files of MODULES, a list of
module names such as `(ice-9 q)'. All of MODULES must be either names of
@@ -1455,7 +1455,7 @@ last one is created from the given <scheme-file> object."
#:key (name "module-import-compiled")
(system (%current-system))
target
- (guile (%guile-for-build))
+ (guile #f)
(module-path %load-path)
(extensions '())
(deprecation-warnings #f))
--
2.26.0
bug-guix <at> gnu.org:bug#39531; Package guix.
(Mon, 20 Apr 2020 07:18:01 GMT) Full text and rfc822 format available.Message #11 received at 39531 <at> debbugs.gnu.org (full text, mbox):
From: Christopher Baines <mail <at> cbaines.net> To: Ludovic Courtès <ludo <at> gnu.org> Cc: 39531 <at> debbugs.gnu.org Subject: Re: bug#39531: guix pull on aarch64-linux glibc derivation has incorrect output Date: Mon, 20 Apr 2020 08:17:11 +0100
[Message part 1 (text/plain, inline)]
Ludovic Courtès <ludo <at> gnu.org> writes: > Christopher Baines <mail <at> cbaines.net> skribis: > >> When attempting to guix pull using the aarch64-linux system, I'm seeing >> some issues with derivation outputs. I tried with a newer and older >> commit, and the result is the same. >> >> >> → guix pull --commit=27b09f3ab11a30821a5ce0b071aac1bc6156497d --system=aarch64-linux --profile=/tmp/testprofile2 >> Updating channel 'guix' from Git repository at 'https://git.savannah.gnu.org/git/guix.git'... >> Building from this channel: >> guix https://git.savannah.gnu.org/git/guix.git 27b09f3 >> Computing Guix derivation for 'aarch64-linux'... - >> guix pull: error: derivation `/gnu/store/800ky8qa4az7yx36gsg9ak6bih3530qm-glibc-2.29.drv' has incorrect output `/gnu/store/8v34v81q86klja9rihaixkypcml5ad5j-glibc-2.29-debug', should be `/gnu/store/w3iq60ias1qlrjigbj75ssda09hwg21i-glibc-2.29-debug' > > The problem here is that we’re building the trampoline, > “compute-guix-derivation”, for AArch64. It builds if substitutes are > available (likely) and fails to build otherwise. And then we try to > execute it locally, and since your machine is not AArch64, it fails. > > The first patch attached does what I thought was all it would take to > fix it. But then I realized that the second patch is needed so that > ‘make-config.scm’ uses a Guile for the right system, same for > “module-import.drv” and so on. > > Together, these two patches solve the problem (not retroactively > though), but we need to check the implications of changing the default > value of #:guile-for-build. Hey, Thanks for investigating :) I completely forget why I encountered this, but it was probably around getting the channel instance derivations in to the Guix Data Service, but I think I got that working, so I guess I somehow avoided this issue. Thanks, Chris
[signature.asc (application/pgp-signature, inline)]
GNU bug tracking system
Copyright (C) 1999 Darren O. Benham,
1997,2003 nCipher Corporation Ltd,
1994-97 Ian Jackson.