bug#26238: core-updates: Fix time zone bug in `date` 8.26

bug#26238: core-updates: Fix time zone bug in `date` 8.26

[Ludovic Courtès]

Leo Famulari <leo@famulari.name> skribis:

> From 466a0ebad18ee5e38f280acf19ab19aadffa2206 Mon Sep 17 00:00:00 2001
> From: Leo Famulari <leo@famulari.name>
> Date: Fri, 24 Mar 2017 12:18:21 -0400
> Subject: [PATCH] gnu: coreutils: Fix a time-zone bug in date 8.26.
>
> * gnu/packages/base.scm (coreutils)[replacement]: New field.
> (coreutils/fixed): New variable.
> (coreutils-minimal)[source]: Use the source of coreutils/fixed.
> * gnu/packages/commencement.scm (coreutils-final)[replacement]: New field.
> * gnu/packages/patches/coreutils-fix-time-zone-conversion.patch: New file.
> * gnu/local.mk (dist_patch_DATA): Add it.

LGTM, thanks!

Ludo’.

bug#26238: core-updates: Fix time zone bug in `date` 8.26

[Ludovic Courtès]

Marius Bakke <mbakke@fastmail.com> skribis:

> Leo Famulari <leo@famulari.name> writes:
>
>> On Sun, Mar 26, 2017 at 10:29:38PM +0200, Marius Bakke wrote:
>>> I don't object to the graft, as it does sound like a serious regression.
>>> My reaction was mostly a "flight" response after seeing the gnulib patch
>>> that this depends upon (the parse_datetime2 API change).
>>> 
>>> https://git.savannah.gnu.org/cgit/gnulib.git/commit/?id=4e6e16b3f43ce96302b1e52e48730c1f15e18c14
>>> 
>>> Are you working on backporting it?
>>
>> Ah, I failed to understand that it would be required to backport the
>> gnulib commit as well. That does look hairy.
>
> Maybe grafting 8.27 is better. It would be good to build it out on a new
> core-updates branch quickly to catch any problems it might introduce.

Let’s merge this one first.  :-)

Ludo’.

bug#26238: core-updates: Fix time zone bug in `date` 8.26

[Leo Famulari]

[-- Attachment #1: Type: text/plain, Size: 1111 bytes --]

On Mon, Mar 27, 2017 at 01:36:32PM +0200, Ludovic Courtès wrote:
> Marius Bakke <mbakke@fastmail.com> skribis:
> 
> > Leo Famulari <leo@famulari.name> writes:
> >
> >> On Sun, Mar 26, 2017 at 10:29:38PM +0200, Marius Bakke wrote:
> >>> I don't object to the graft, as it does sound like a serious regression.
> >>> My reaction was mostly a "flight" response after seeing the gnulib patch
> >>> that this depends upon (the parse_datetime2 API change).
> >>> 
> >>> https://git.savannah.gnu.org/cgit/gnulib.git/commit/?id=4e6e16b3f43ce96302b1e52e48730c1f15e18c14
> >>> 
> >>> Are you working on backporting it?
> >>
> >> Ah, I failed to understand that it would be required to backport the
> >> gnulib commit as well. That does look hairy.
> >
> > Maybe grafting 8.27 is better. It would be good to build it out on a new
> > core-updates branch quickly to catch any problems it might introduce.
> 
> Let’s merge this one first.  :-)

Yeah, and I think we'll have to skip this bug fix for now,
unfortunately. Unless somebody else feels up to the task of backporting
the gnulib changes.

[-- Attachment #2: signature.asc --]
[-- Type: application/pgp-signature, Size: 833 bytes --]

bug#26238: core-updates: Fix time zone bug in `date` 8.26

[Ludovic Courtès]

Hi Leo!

ludo@gnu.org (Ludovic Courtès) skribis:

> Leo Famulari <leo@famulari.name> skribis:
>
>> From 466a0ebad18ee5e38f280acf19ab19aadffa2206 Mon Sep 17 00:00:00 2001
>> From: Leo Famulari <leo@famulari.name>
>> Date: Fri, 24 Mar 2017 12:18:21 -0400
>> Subject: [PATCH] gnu: coreutils: Fix a time-zone bug in date 8.26.
>>
>> * gnu/packages/base.scm (coreutils)[replacement]: New field.
>> (coreutils/fixed): New variable.
>> (coreutils-minimal)[source]: Use the source of coreutils/fixed.
>> * gnu/packages/commencement.scm (coreutils-final)[replacement]: New field.
>> * gnu/packages/patches/coreutils-fix-time-zone-conversion.patch: New file.
>> * gnu/local.mk (dist_patch_DATA): Add it.
>
> LGTM, thanks!

Was there anything wrong with this patch?  Otherwise I still think it’s
worth applying to ‘master’.

Thanks,
Ludo’.

bug#26238: core-updates: Fix time zone bug in `date` 8.26

[Marius Bakke]

[-- Attachment #1: Type: text/plain, Size: 1116 bytes --]

Ludovic Courtès <ludo@gnu.org> writes:

> Hi Leo!
>
> ludo@gnu.org (Ludovic Courtès) skribis:
>
>> Leo Famulari <leo@famulari.name> skribis:
>>
>>> From 466a0ebad18ee5e38f280acf19ab19aadffa2206 Mon Sep 17 00:00:00 2001
>>> From: Leo Famulari <leo@famulari.name>
>>> Date: Fri, 24 Mar 2017 12:18:21 -0400
>>> Subject: [PATCH] gnu: coreutils: Fix a time-zone bug in date 8.26.
>>>
>>> * gnu/packages/base.scm (coreutils)[replacement]: New field.
>>> (coreutils/fixed): New variable.
>>> (coreutils-minimal)[source]: Use the source of coreutils/fixed.
>>> * gnu/packages/commencement.scm (coreutils-final)[replacement]: New field.
>>> * gnu/packages/patches/coreutils-fix-time-zone-conversion.patch: New file.
>>> * gnu/local.mk (dist_patch_DATA): Add it.
>>
>> LGTM, thanks!
>
> Was there anything wrong with this patch?  Otherwise I still think it’s
> worth applying to ‘master’.

It depends on this gnulib patch, which does not apply cleanly and needs
a rather difficult backport:

https://git.savannah.gnu.org/cgit/gnulib.git/commit/?id=4e6e16b3f43ce96302b1e52e48730c1f15e18c14

[-- Attachment #2: signature.asc --]
[-- Type: application/pgp-signature, Size: 487 bytes --]

bug#26238: core-updates: Fix time zone bug in `date` 8.26

[Ludovic Courtès]

Morning!

Marius Bakke <mbakke@fastmail.com> skribis:

> Ludovic Courtès <ludo@gnu.org> writes:
>
>> Hi Leo!
>>
>> ludo@gnu.org (Ludovic Courtès) skribis:
>>
>>> Leo Famulari <leo@famulari.name> skribis:
>>>
>>>> From 466a0ebad18ee5e38f280acf19ab19aadffa2206 Mon Sep 17 00:00:00 2001
>>>> From: Leo Famulari <leo@famulari.name>
>>>> Date: Fri, 24 Mar 2017 12:18:21 -0400
>>>> Subject: [PATCH] gnu: coreutils: Fix a time-zone bug in date 8.26.
>>>>
>>>> * gnu/packages/base.scm (coreutils)[replacement]: New field.
>>>> (coreutils/fixed): New variable.
>>>> (coreutils-minimal)[source]: Use the source of coreutils/fixed.
>>>> * gnu/packages/commencement.scm (coreutils-final)[replacement]: New field.
>>>> * gnu/packages/patches/coreutils-fix-time-zone-conversion.patch: New file.
>>>> * gnu/local.mk (dist_patch_DATA): Add it.
>>>
>>> LGTM, thanks!
>>
>> Was there anything wrong with this patch?  Otherwise I still think it’s
>> worth applying to ‘master’.
>
> It depends on this gnulib patch, which does not apply cleanly and needs
> a rather difficult backport:
>
> https://git.savannah.gnu.org/cgit/gnulib.git/commit/?id=4e6e16b3f43ce96302b1e52e48730c1f15e18c14

Oh right, sorry for being sloppy.

A while back you suggested using directly 8.27 as the replacement:

  https://debbugs.gnu.org/cgi/bugreport.cgi?bug=26238#32

I think it would be a good idea.

Thoughts?

Ludo’.

bug#26238: core-updates: Fix time zone bug in `date` 8.26

[Marius Bakke]

[-- Attachment #1.1: Type: text/plain, Size: 1802 bytes --]

Ludovic Courtès <ludo@gnu.org> writes:

> Morning!
>
> Marius Bakke <mbakke@fastmail.com> skribis:
>
>> Ludovic Courtès <ludo@gnu.org> writes:
>>
>>> Hi Leo!
>>>
>>> ludo@gnu.org (Ludovic Courtès) skribis:
>>>
>>>> Leo Famulari <leo@famulari.name> skribis:
>>>>
>>>>> From 466a0ebad18ee5e38f280acf19ab19aadffa2206 Mon Sep 17 00:00:00 2001
>>>>> From: Leo Famulari <leo@famulari.name>
>>>>> Date: Fri, 24 Mar 2017 12:18:21 -0400
>>>>> Subject: [PATCH] gnu: coreutils: Fix a time-zone bug in date 8.26.
>>>>>
>>>>> * gnu/packages/base.scm (coreutils)[replacement]: New field.
>>>>> (coreutils/fixed): New variable.
>>>>> (coreutils-minimal)[source]: Use the source of coreutils/fixed.
>>>>> * gnu/packages/commencement.scm (coreutils-final)[replacement]: New field.
>>>>> * gnu/packages/patches/coreutils-fix-time-zone-conversion.patch: New file.
>>>>> * gnu/local.mk (dist_patch_DATA): Add it.
>>>>
>>>> LGTM, thanks!
>>>
>>> Was there anything wrong with this patch?  Otherwise I still think it’s
>>> worth applying to ‘master’.
>>
>> It depends on this gnulib patch, which does not apply cleanly and needs
>> a rather difficult backport:
>>
>> https://git.savannah.gnu.org/cgit/gnulib.git/commit/?id=4e6e16b3f43ce96302b1e52e48730c1f15e18c14
>
> Oh right, sorry for being sloppy.
>
> A while back you suggested using directly 8.27 as the replacement:
>
>   https://debbugs.gnu.org/cgi/bugreport.cgi?bug=26238#32
>
> I think it would be a good idea.
>
> Thoughts?

Here is a WIP patch that adds a graft for coreutils 8.27. I stole Leos
code for the (gnu packages commencement) part. It builds with
`./pre-inst-env guix build coreutils`, but my after reconfiguring my
system on it, `date --version` still shows 8.26.

Any idea what's wrong?


[-- Warning: decoded text below may be mangled, UTF-8 assumed --]
[-- Attachment #1.2: 0001-gnu-coreutils-Replace-with-8.27.patch --]
[-- Type: text/x-patch, Size: 3768 bytes --]

From 2636e565fa4847dc0d1c786345e98053cf783cc1 Mon Sep 17 00:00:00 2001
From: Marius Bakke <mbakke@fastmail.com>
Date: Tue, 18 Apr 2017 20:13:45 +0200
Subject: [PATCH] gnu: coreutils: Replace with 8.27.

* gnu/packages/base.scm (coreutils-8.27): New variable.
(coreutils)[replacement]: New field.
(coreutils-minimal)[source]: Inherit from COREUTILS-8.27.
[replacement]: New field.
* gnu/packages/commencement.scm (coreutils-final): Set replacement #f on
coreutils derivation.

Co-authored-by: Leo Famulari <leo@famulari.name>
---
 gnu/packages/base.scm         | 15 +++++++++++++++
 gnu/packages/commencement.scm | 20 ++++++++++++--------
 2 files changed, 27 insertions(+), 8 deletions(-)

diff --git a/gnu/packages/base.scm b/gnu/packages/base.scm
index 5c36b612f..36eaabb36 100644
--- a/gnu/packages/base.scm
+++ b/gnu/packages/base.scm
@@ -285,6 +285,7 @@ used to apply commands with arbitrarily long arguments.")
   (package
    (name "coreutils")
    (version "8.26")
+   (replacement coreutils-8.27)
    (source (origin
             (method url-fetch)
             (uri (string-append "mirror://gnu/coreutils/coreutils-"
@@ -361,11 +362,25 @@ functionality beyond that which is outlined in the POSIX standard.")
    (license gpl3+)
    (home-page "https://www.gnu.org/software/coreutils/")))
 
+(define coreutils-8.27
+  (package
+    (inherit coreutils)
+    (version "8.27")
+    (source (origin
+              (method url-fetch)
+              (uri (string-append "mirror://gnu/coreutils/coreutils-"
+                                  version ".tar.xz"))
+              (sha256
+               (base32
+                "0sv547572iq8ayy8klir4hnngnx92a9nsazmf1wgzfc7xr4x74c8"))))))
+
 (define-public coreutils-minimal
   ;; Coreutils without its optional dependencies.
   (package
     (inherit coreutils)
     (name "coreutils-minimal")
+    (replacement #f)
+    (source (package-source coreutils-8.27))
     (outputs '("out"))
     (inputs '())))
 
diff --git a/gnu/packages/commencement.scm b/gnu/packages/commencement.scm
index 92f6e6c2e..c9d851c4d 100644
--- a/gnu/packages/commencement.scm
+++ b/gnu/packages/commencement.scm
@@ -868,15 +868,19 @@ exec ~a/bin/~a-~a -B~a/lib -Wl,-dynamic-linker -Wl,~a/~a \"$@\"~%"
   ;; The final Coreutils.  Treat them specially because some packages, such as
   ;; Findutils, keep a reference to the Coreutils they were built with.
   (package-with-bootstrap-guile
-   (package-with-explicit-inputs coreutils
-                                 %boot5-inputs
-                                 (current-source-location)
 
-                                 ;; Use the final Guile, linked against the
-                                 ;; final libc with working iconv, so that
-                                 ;; 'substitute*' works well when touching
-                                 ;; test files in Gettext.
-                                 #:guile guile-final)))
+
+   (let ((p (package-with-explicit-inputs coreutils
+                                          %boot5-inputs
+                                          (current-source-location)
+
+                                          ;; Use the final Guile, linked against the
+                                          ;; final libc with working iconv, so that
+                                          ;; 'substitute*' works well when touching
+                                          ;; test files in Gettext.
+                                          #:guile guile-final)))
+           (package (inherit p)
+                    (replacement #f)))))
 
 (define grep-final
   ;; The final grep.  Gzip holds a reference to it (via zgrep), so it must be
-- 
2.12.2


[-- Attachment #2: signature.asc --]
[-- Type: application/pgp-signature, Size: 487 bytes --]

bug#26238: core-updates: Fix time zone bug in `date` 8.26

[Ludovic Courtès]

Marius Bakke <mbakke@fastmail.com> skribis:

> Here is a WIP patch that adds a graft for coreutils 8.27. I stole Leos
> code for the (gnu packages commencement) part. It builds with
> `./pre-inst-env guix build coreutils`, but my after reconfiguring my
> system on it, `date --version` still shows 8.26.
>
> Any idea what's wrong?

Bah!  The system uses ‘coreutils-final’ due to this stanza in (gnu
system):

     ;; The packages below are also in %FINAL-INPUTS, so take them from
     ;; there to avoid duplication.
     (map canonical-package
          (list guile-2.0 bash coreutils findutils grep sed
                diffutils patch gawk tar gzip bzip2 xz lzip))

Not sure how to handle it.  We could explicitly write ‘coreutils-8.27’
above, which would solve the problem.  It would also make the graft much
less useful, except for people explicitly doing:

  guix package -i coreutils

So we might just as well remove the graft.

Thoughts?

Ludo’.

bug#26238: core-updates: Fix time zone bug in `date` 8.26

[Marius Bakke]

[-- Attachment #1.1: Type: text/plain, Size: 1268 bytes --]

Ludovic Courtès <ludo@gnu.org> writes:

> Marius Bakke <mbakke@fastmail.com> skribis:
>
>> Here is a WIP patch that adds a graft for coreutils 8.27. I stole Leos
>> code for the (gnu packages commencement) part. It builds with
>> `./pre-inst-env guix build coreutils`, but my after reconfiguring my
>> system on it, `date --version` still shows 8.26.
>>
>> Any idea what's wrong?
>
> Bah!  The system uses ‘coreutils-final’ due to this stanza in (gnu
> system):
>
>      ;; The packages below are also in %FINAL-INPUTS, so take them from
>      ;; there to avoid duplication.
>      (map canonical-package
>           (list guile-2.0 bash coreutils findutils grep sed
>                 diffutils patch gawk tar gzip bzip2 xz lzip))
>
> Not sure how to handle it.  We could explicitly write ‘coreutils-8.27’
> above, which would solve the problem.  It would also make the graft much
> less useful, except for people explicitly doing:
>
>   guix package -i coreutils
>
> So we might just as well remove the graft.

Okay, that makes sense. I dropped the graft in the patch below and can
confirm it works on my system. Note that coreutils-8.27 is now public so
that `guix package -i coreutils` will pull in 8.27 regardless. PTAL.


[-- Warning: decoded text below may be mangled, UTF-8 assumed --]
[-- Attachment #1.2: 0001-gnu-Add-workaround-for-date-regression-in-coreutils-.patch --]
[-- Type: text/x-patch, Size: 2207 bytes --]

From 5be25413a76ca84c18e4f4174629af5fe6e87db1 Mon Sep 17 00:00:00 2001
From: Marius Bakke <mbakke@fastmail.com>
Date: Tue, 18 Apr 2017 20:13:45 +0200
Subject: [PATCH] gnu: Add workaround for `date` regression in coreutils@8.26.

See <https://bugs.gnu.org/23035> and <https://bugs.gnu.org/26238>.

* gnu/packages/base.scm (coreutils-8.27): New variable.
* gnu/system.scm (%base-packages): Use that instead of COREUTILS.
---
 gnu/packages/base.scm | 16 ++++++++++++++++
 gnu/system.scm        |  2 +-
 2 files changed, 17 insertions(+), 1 deletion(-)

diff --git a/gnu/packages/base.scm b/gnu/packages/base.scm
index 5c36b612f..a95ca2888 100644
--- a/gnu/packages/base.scm
+++ b/gnu/packages/base.scm
@@ -361,6 +361,22 @@ functionality beyond that which is outlined in the POSIX standard.")
    (license gpl3+)
    (home-page "https://www.gnu.org/software/coreutils/")))
 
+;; We add version 8.27 here for use in (gnu system) due to a time
+;; zone bug in `date' versions 8.25 - 8.26.
+;; https://debbugs.gnu.org/cgi/bugreport.cgi?bug=23035
+;; https://debbugs.gnu.org/cgi/bugreport.cgi?bug=26238
+(define-public coreutils-8.27
+  (package
+    (inherit coreutils)
+    (version "8.27")
+    (source (origin
+              (method url-fetch)
+              (uri (string-append "mirror://gnu/coreutils/coreutils-"
+                                  version ".tar.xz"))
+              (sha256
+               (base32
+                "0sv547572iq8ayy8klir4hnngnx92a9nsazmf1wgzfc7xr4x74c8"))))))
+
 (define-public coreutils-minimal
   ;; Coreutils without its optional dependencies.
   (package
diff --git a/gnu/system.scm b/gnu/system.scm
index 89c4150f9..f6ab7ded8 100644
--- a/gnu/system.scm
+++ b/gnu/system.scm
@@ -385,7 +385,7 @@ explicitly appear in OS."
          ;; The packages below are also in %FINAL-INPUTS, so take them from
          ;; there to avoid duplication.
          (map canonical-package
-              (list guile-2.0 bash coreutils findutils grep sed
+              (list guile-2.0 bash coreutils-8.27 findutils grep sed
                     diffutils patch gawk tar gzip bzip2 xz lzip))))
 
 (define %default-issue
-- 
2.12.2


[-- Attachment #2: signature.asc --]
[-- Type: application/pgp-signature, Size: 487 bytes --]

bug#26238: core-updates: Fix time zone bug in `date` 8.26

[Ludovic Courtès]

Heya,

Marius Bakke <mbakke@fastmail.com> skribis:

>> Not sure how to handle it.  We could explicitly write ‘coreutils-8.27’
>> above, which would solve the problem.  It would also make the graft much
>> less useful, except for people explicitly doing:
>>
>>   guix package -i coreutils
>>
>> So we might just as well remove the graft.
>
> Okay, that makes sense. I dropped the graft in the patch below and can
> confirm it works on my system. Note that coreutils-8.27 is now public so
> that `guix package -i coreutils` will pull in 8.27 regardless. PTAL.

Indeed.

> From 5be25413a76ca84c18e4f4174629af5fe6e87db1 Mon Sep 17 00:00:00 2001
> From: Marius Bakke <mbakke@fastmail.com>
> Date: Tue, 18 Apr 2017 20:13:45 +0200
> Subject: [PATCH] gnu: Add workaround for `date` regression in coreutils@8.26.
>
> See <https://bugs.gnu.org/23035> and <https://bugs.gnu.org/26238>.
>
> * gnu/packages/base.scm (coreutils-8.27): New variable.
> * gnu/system.scm (%base-packages): Use that instead of COREUTILS.

Go for it!

Thanks,
Ludo’.

bug#26238: core-updates: Fix time zone bug in `date` 8.26

[Marius Bakke]

[-- Attachment #1: Type: text/plain, Size: 524 bytes --]

Ludovic Courtès <ludo@gnu.org> writes:

>> From 5be25413a76ca84c18e4f4174629af5fe6e87db1 Mon Sep 17 00:00:00 2001
>> From: Marius Bakke <mbakke@fastmail.com>
>> Date: Tue, 18 Apr 2017 20:13:45 +0200
>> Subject: [PATCH] gnu: Add workaround for `date` regression in coreutils@8.26.
>>
>> See <https://bugs.gnu.org/23035> and <https://bugs.gnu.org/26238>.
>>
>> * gnu/packages/base.scm (coreutils-8.27): New variable.
>> * gnu/system.scm (%base-packages): Use that instead of COREUTILS.
>
> Go for it!

Pushed!

[-- Attachment #2: signature.asc --]
[-- Type: application/pgp-signature, Size: 487 bytes --]