Bug#893006: stretch-pu: package boost1.62/1.62.0+dfsg-4+deb9u1

[Julien Cristau]

Control: tag -1 moreinfo

On Wed, Apr 04, 2018 at 10:25:30PM +0200, Philipp Huebner wrote:
> Hi,
> 
> Am 02.04.2018 um 12:57 schrieb Julien Cristau:
> > On Thu, Mar 15, 2018 at 14:51:10 +0100, Philipp Huebner wrote:
> >> I would like to fix #883987 in boost1.62 for Stretch.
> >> The changes are basically the same as what is currently in testing and I 
> >> got the
> >> maintainer's go-ahead in
> >> http://lists.alioth.debian.org/pipermail/pkg-boost-devel/2018-March/004184.html
> >>
> > What made these partial specializations not be necessary anymore?  That
> > seems like critical missing information if we are to make a decision
> > here, to know if/what we might be breaking instead.
> 
> my guess is that only upstream can really answer that.
> 
> My work colleague confirmed that we're basically using the same code as
> in the example at the upstream bug tracker and it fails when using gcc-6
> on Stretch:
> https://svn.boost.org/trac10/ticket/12534
> 
> 
> Best wishes,
> -- 
>  .''`.   Philipp Huebner 
> : :'  :  pgp fp: 6719 25C5 B8CD E74A 5225  3DF9 E5CA 8C49 25E4 205F
> `. `'`
>   `-
> 

Bug#893006: stretch-pu: package boost1.62/1.62.0+dfsg-4+deb9u1

[Philipp Huebner]

Hi,

Am 02.04.2018 um 12:57 schrieb Julien Cristau:
> On Thu, Mar 15, 2018 at 14:51:10 +0100, Philipp Huebner wrote:
>> I would like to fix #883987 in boost1.62 for Stretch.
>> The changes are basically the same as what is currently in testing and I got 
>> the
>> maintainer's go-ahead in
>> http://lists.alioth.debian.org/pipermail/pkg-boost-devel/2018-March/004184.html
>>
> What made these partial specializations not be necessary anymore?  That
> seems like critical missing information if we are to make a decision
> here, to know if/what we might be breaking instead.

my guess is that only upstream can really answer that.

My work colleague confirmed that we're basically using the same code as
in the example at the upstream bug tracker and it fails when using gcc-6
on Stretch:
https://svn.boost.org/trac10/ticket/12534


Best wishes,
-- 
 .''`.   Philipp Huebner 
: :'  :  pgp fp: 6719 25C5 B8CD E74A 5225  3DF9 E5CA 8C49 25E4 205F
`. `'`
  `-



signature.asc
Description: OpenPGP digital signature

Bug#893006: stretch-pu: package boost1.62/1.62.0+dfsg-4+deb9u1

[Julien Cristau]

On Thu, Mar 15, 2018 at 14:51:10 +0100, Philipp Huebner wrote:

> Package: release.debian.org
> Severity: normal
> Tags: stretch
> User: release.debian@packages.debian.org
> Usertags: pu
> 
> Hi,
> 
> I would like to fix #883987 in boost1.62 for Stretch.
> The changes are basically the same as what is currently in testing and I got 
> the
> maintainer's go-ahead in
> http://lists.alioth.debian.org/pipermail/pkg-boost-devel/2018-March/004184.html
> 
What made these partial specializations not be necessary anymore?  That
seems like critical missing information if we are to make a decision
here, to know if/what we might be breaking instead.

Cheers,
Julien

Bug#893006: stretch-pu: package boost1.62/1.62.0+dfsg-4+deb9u1

[Andreas Beckmann]

Control: tag -1 stretch

On 2018-04-01 14:51, Adam D. Barratt wrote:
> Andreas - you tagged #883987 as "buster sid" despite the version
> tracking implying that stretch was affected. Do you remember the
> details as to why?

Most likely:

On Tue, 12 Dec 2017 14:13:18 +0100 (CET) Pierre Saramito
 wrote:> Hi Andreas,
>
> This problem do neither comes from Rheolef-6.7 nor from CGAL-4.11:
> it comes from Boost-1.62 combined with g++ 7.2 in Debian sid and testing.

So gcc-7 triggered the tags.

Adding stretch since the problem is apparently reproducible there ...
(and "stretch buster sid" (rather than "") carries for me the meaning
that this was set intentionally in this "unclear" case)


Andreas

Bug#893006: stretch-pu: package boost1.62/1.62.0+dfsg-4+deb9u1

[Adam D. Barratt]

On Sun, 2018-04-01 at 14:42 +0200, Philipp Huebner wrote:
> Hi,
> 
> Am 31.03.2018 um 23:34 schrieb Adam D. Barratt:
> > It's unclear to me that this actually affects stretch in any
> > meaningful
> >  way.
> > 
> > While the version tracking information does include the stretch
> > version
> > in the "found" list, the bug is tagged "buster sid". From reading
> > the
> > bug log, it looks like this is because the issue only occurs when
> > using
> >  GCC 7, which is not present in stretch.
> 
> that is definitely incorrect. Work colleagues of mine ran into this
> bug on plain Stretch systems, and it disappeared when using the
> patched boost packages I provided.
> 
> I cannot share the code because that's privileged information, but
> you can take my word for it. I am already providing the patched
> packages to dozens of machines via a local repository.
> 

Andreas - you tagged #883987 as "buster sid" despite the version
tracking implying that stretch was affected. Do you remember the
details as to why?

Regards,

Adam

Bug#893006: stretch-pu: package boost1.62/1.62.0+dfsg-4+deb9u1

[Philipp Huebner]

Hi,

Am 31.03.2018 um 23:34 schrieb Adam D. Barratt:
> It's unclear to me that this actually affects stretch in any meaningful
>  way.
> 
> While the version tracking information does include the stretch version
> in the "found" list, the bug is tagged "buster sid". From reading the
> bug log, it looks like this is because the issue only occurs when using
>  GCC 7, which is not present in stretch.

that is definitely incorrect. Work colleagues of mine ran into this bug
on plain Stretch systems, and it disappeared when using the patched
boost packages I provided.

I cannot share the code because that's privileged information, but you
can take my word for it. I am already providing the patched packages to
dozens of machines via a local repository.

Obviously I would like them very much to go back to Debian so all users
can benefit from it.

Kind regards,
-- 
 .''`.   Philipp Huebner 
: :'  :  pgp fp: 6719 25C5 B8CD E74A 5225  3DF9 E5CA 8C49 25E4 205F
`. `'`
  `-





signature.asc
Description: OpenPGP digital signature

Bug#893006: stretch-pu: package boost1.62/1.62.0+dfsg-4+deb9u1

[Adam D. Barratt]

Control: tags -1 + moreinfo

On Thu, 2018-03-15 at 14:51 +0100, Philipp Huebner wrote:
> I would like to fix #883987 in boost1.62 for Stretch.
> The changes are basically the same as what is currently in testing
> and I got the
> maintainer's go-ahead in
> http://lists.alioth.debian.org/pipermail/pkg-boost-devel/2018-March/0
> 04184.html
> 

It's unclear to me that this actually affects stretch in any meaningful
 way.

While the version tracking information does include the stretch version
in the "found" list, the bug is tagged "buster sid". From reading the
bug log, it looks like this is because the issue only occurs when using
 GCC 7, which is not present in stretch.

Regards,

Adam

Bug#893006: stretch-pu: package boost1.62/1.62.0+dfsg-4+deb9u1

[Philipp Huebner]

Package: release.debian.org
Severity: normal
Tags: stretch
User: release.debian@packages.debian.org
Usertags: pu

Hi,

I would like to fix #883987 in boost1.62 for Stretch.
The changes are basically the same as what is currently in testing and I got the
maintainer's go-ahead in
http://lists.alioth.debian.org/pipermail/pkg-boost-devel/2018-March/004184.html

The source debdiff is attached.
The patched binary packages have been tested and the bug confirmed to be
solved.

Best wishes,
Philipp
debdiff --diffstat boost1.62_1.62.0+dfsg-4.dsc 
boost1.62_1.62.0+dfsg-4+deb9u1.dsc


diffstat for boost1.62-1.62.0+dfsg boost1.62-1.62.0+dfsg

 changelog |9 +++
 patches/flat-map-remove-partial-specializations.patch |   52 ++
 patches/series|1 
 3 files changed, 62 insertions(+)

diff -Nru boost1.62-1.62.0+dfsg/debian/changelog 
boost1.62-1.62.0+dfsg/debian/changelog
--- boost1.62-1.62.0+dfsg/debian/changelog  2016-11-12 19:46:50.0 
+0100
+++ boost1.62-1.62.0+dfsg/debian/changelog  2018-03-14 09:54:41.0 
+0100
@@ -1,3 +1,12 @@
+boost1.62 (1.62.0+dfsg-4+deb9u1) stretch; urgency=medium
+
+  [ Steve M. Robbins ]
+  * patches/flat-map-remove-partial-specializations.patch: New.  Upstream
+patch to remove now-unnecessary partial specializations.  Closes:
+#883987.
+
+ -- Philipp Huebner   Wed, 14 Mar 2018 09:54:41 +0100
+
 boost1.62 (1.62.0+dfsg-4) unstable; urgency=medium
 
   * New patch upstream-add-degree-reverse_graph.patch.
diff -Nru 
boost1.62-1.62.0+dfsg/debian/patches/flat-map-remove-partial-specializations.patch
 
boost1.62-1.62.0+dfsg/debian/patches/flat-map-remove-partial-specializations.patch
--- 
boost1.62-1.62.0+dfsg/debian/patches/flat-map-remove-partial-specializations.patch
  1970-01-01 01:00:00.0 +0100
+++ 
boost1.62-1.62.0+dfsg/debian/patches/flat-map-remove-partial-specializations.patch
  2018-01-06 21:44:32.0 +0100
@@ -0,0 +1,52 @@
+From 5e4a107e82ab3281688311d22d2bfc2fddcf84a3 Mon Sep 17 00:00:00 2001
+From: =?UTF-8?q?Ion=20Gazta=C3=B1aga?= 
+Date: Fri, 25 Nov 2016 23:56:33 +0100
+Subject: [PATCH] Fixes Ticket #12534: flat_map fails to compile if included
+ after type_traits is instantiated under gcc
+
+---
+ doc/container.qbk   |  1 +
+ include/boost/container/detail/pair.hpp | 30 --
+ 2 files changed, 1 insertion(+), 30 deletions(-)
+
+diff --git a/include/boost/container/detail/pair.hpp 
b/include/boost/container/detail/pair.hpp
+index 63d1dead..4abff4b4 100644
+--- a/boost/container/detail/pair.hpp
 b/boost/container/detail/pair.hpp
+@@ -428,36 +428,6 @@ inline void swap(pair& x, pair& y)
+ }  //namespace container_detail {
+ }  //namespace container {
+ 
+-
+-//Without this specialization recursive flat_(multi)map instantiation fails
+-//because is_enum needs to instantiate the recursive pair, leading to a 
compilation error).
+-//This breaks the cycle clearly stating that pair is not an enum avoiding any 
instantiation.
+-template
+-struct is_enum;
+-
+-template
+-struct is_enum< ::boost::container::container_detail::pair >
+-{
+-   static const bool value = false;
+-};
+-
+-template
+-struct is_enum< ::std::pair >
+-{
+-   static const bool value = false;
+-};
+-
+-template 
+-struct is_class;
+-
+-//This specialization is needed to avoid instantiation of pair in
+-//is_class, and allow recursive maps.
+-template 
+-struct is_class< ::boost::container::container_detail::pair >
+-{
+-   static const bool value = true;
+-};
+-
+ #ifdef BOOST_NO_CXX11_RVALUE_REFERENCES
+ 
+ template
diff -Nru boost1.62-1.62.0+dfsg/debian/patches/series 
boost1.62-1.62.0+dfsg/debian/patches/series
--- boost1.62-1.62.0+dfsg/debian/patches/series 2016-11-12 19:46:50.0 
+0100
+++ boost1.62-1.62.0+dfsg/debian/patches/series 2018-03-14 09:54:08.0 
+0100
@@ -8,3 +8,4 @@
 # fixed alternatively? boost-context-use-sysv-not-aapcs.patch
 no-gcc-march-options.patch
 upstream-add-degree-reverse_graph.patch
+flat-map-remove-partial-specializations.patch