GNU bug report logs - #28273
[PATCH] gnu: Fix current-guix.

Previous Next

Full log

View this message in rfc822 format

From: help-debbugs <at> gnu.org (GNU bug Tracking System)
To: Christopher Baines <mail <at> cbaines.net>
Cc: tracker <at> debbugs.gnu.org
Subject: bug#28273: closed ([PATCH] gnu: Fix current-guix.)
Date: Thu, 31 Aug 2017 21:05:01 +0000

[Message part 1 (text/plain, inline)]

Your message dated Thu, 31 Aug 2017 22:04:21 +0100
with message-id <20170831220421.5ba1287c <at> cbaines.net>
and subject line Re: [bug#28273] [PATCH] gnu: Fix current-guix.
has caused the debbugs.gnu.org bug report #28273,
regarding [PATCH] gnu: Fix current-guix.
to be marked as done.

(If you believe you have received this mail in error, please contact
help-debbugs <at> gnu.org.)


-- 
28273: http://debbugs.gnu.org/cgi/bugreport.cgi?bug=28273
GNU Bug Tracking System
Contact help-debbugs <at> gnu.org with problems

[Message part 2 (message/rfc822, inline)]

From: Christopher Baines <mail <at> cbaines.net>
To: guix-patches <at> gnu.org
Subject: [PATCH] gnu: Fix current-guix.
Date: Tue, 29 Aug 2017 07:49:30 +0100

Without this change, I get errors like:
  ERROR: In procedure string-drop:
  ERROR: Value out of range 0 to 35: 51

* gnu/packages/package-management.scm (current-guix): Pass exactly the same
  path to git-predicate and local-file, to ensure that the select? function is
  compatible.
---
 gnu/packages/package-management.scm | 11 ++++++-----
 1 file changed, 6 insertions(+), 5 deletions(-)

diff --git a/gnu/packages/package-management.scm b/gnu/packages/package-management.scm
index 67a956dea..a6c97e618 100644
--- a/gnu/packages/package-management.scm
+++ b/gnu/packages/package-management.scm
@@ -312,10 +312,11 @@ generated file."
      #t)))
 
 (define-public current-guix
-  (let ((select? (delay (or (git-predicate
-                             (string-append (current-source-directory)
-                                            "/../.."))
-                            source-file?))))
+  (let* ((repository-root (canonicalize-path
+                           (string-append (current-source-directory)
+                                          "/../..")))
+         (select? (delay (or (git-predicate repository-root)
+                             source-file?))))
     (lambda ()
       "Return a package representing Guix built from the current source tree.
 This works by adding the current source tree to the store (after filtering it
@@ -323,7 +324,7 @@ out) and returning a package that uses that as its 'source'."
       (package
         (inherit guix)
         (version (string-append (package-version guix) "+"))
-        (source (local-file "../.." "guix-current"
+        (source (local-file repository-root "guix-current"
                             #:recursive? #t
                             #:select? (force select?)))))))
 
-- 
2.14.1

[Message part 3 (message/rfc822, inline)]

From: Christopher Baines <mail <at> cbaines.net>
To: ludo <at> gnu.org (Ludovic Courtès)
Cc: 28273-done <at> debbugs.gnu.org
Subject: Re: [bug#28273] [PATCH] gnu: Fix current-guix.
Date: Thu, 31 Aug 2017 22:04:21 +0100

[Message part 4 (text/plain, inline)]

On Thu, 31 Aug 2017 15:18:37 +0200
ludo <at> gnu.org (Ludovic Courtès) wrote:

> Christopher Baines <mail <at> cbaines.net> skribis:
> 
> > Without this change, I get errors like:
> >   ERROR: In procedure string-drop:
> >   ERROR: Value out of range 0 to 35: 51
> >
> > * gnu/packages/package-management.scm (current-guix): Pass exactly
> > the same path to git-predicate and local-file, to ensure that the
> > select? function is compatible.  
> 
> Good catch, LGTM.

Great, I've merged this now.

> I think that fundamentally, we should arrange for ‘git-predicate’ to
> not do any string prefix comparison.  That’s bound to fail.  If we
> could somehow restrict it to comparing inode numbers, that’d be
> perfect.

Hmm, yeah, that might work well... I think the structure would simplify
to:
 - taking the list of files, creating a list of directories
   (rather than a tree)
 - combining these lists
 - finding each inode for every file and directory

The predicate could then check if the provided inode is known, without
switching on the type as it does at the moment.

I'll put it on my list of things to look at :)

[Message part 5 (application/pgp-signature, inline)]

Previous Next

GNU bug tracking system
Copyright (C) 1999 Darren O. Benham, 1997,2003 nCipher Corporation Ltd, 1994-97 Ian Jackson.