[Buildroot] [git commit branch/2019.02.x] qt5base: Add patch to fix compile issue with gcc9

[Buildroot] [git commit branch/2019.02.x] qt5base: Add patch to fix compile issue with gcc9

[Julien Béraud]

Backport:
https://github.com/qt/qtbase/commit/a52d7861edfb5956de38ba80015c4dd0b596259b#diff-0b4819f9df9c43297ba840d1ac401599
Signed-off-by: Julien Beraud <julien.beraud@orolia.com>
---
 .../0006-Fix-compile-issue-with-gcc-9.patch   | 32 +++++++++++++++++++
 1 file changed, 32 insertions(+)
 create mode 100644 package/qt5/qt5base/5.11.3/0006-Fix-compile-issue-with-gcc-9.patch

diff --git a/package/qt5/qt5base/5.11.3/0006-Fix-compile-issue-with-gcc-9.patch b/package/qt5/qt5base/5.11.3/0006-Fix-compile-issue-with-gcc-9.patch
new file mode 100644
index 0000000000..a10ab06e35
--- /dev/null
+++ b/package/qt5/qt5base/5.11.3/0006-Fix-compile-issue-with-gcc-9.patch
@@ -0,0 +1,32 @@
+From e094806951ff7337b5b0c534db479e3808f153a7 Mon Sep 17 00:00:00 2001
+From: Allan Sandfeld Jensen <allan.jensen@qt.io>
+Date: Tue, 13 Nov 2018 16:00:23 +0100
+Subject: [PATCH 1/1] Fix compile issue with gcc 9
+
+It appears messenne_twisters in the latest libstdc++ has one more
+requirement before it is willing to construct with our
+SystemGenerator struct as an sseq provider.
+
+Change-Id: If38151d1fa6f40a80274acc26d9ed6b4ac6049fe
+Reviewed-by: Giuseppe D'Angelo <giuseppe.dangelo@kdab.com>
+Reviewed-by: Thiago Macieira <thiago.macieira@intel.com>
+Signed-off-by: Julien Beraud <julien.beraud@orolia.com>
+---
+ src/corelib/global/qrandom.cpp | 1 +
+ 1 file changed, 1 insertion(+)
+
+diff --git a/src/corelib/global/qrandom.cpp b/src/corelib/global/qrandom.cpp
+index ebf9864b15..577736a867 100644
+--- a/src/corelib/global/qrandom.cpp
++++ b/src/corelib/global/qrandom.cpp
+@@ -218,6 +218,7 @@ struct QRandomGenerator::SystemGenerator
+ #endif // Q_OS_WINRT
+ 
+     static SystemGenerator &self();
++    typedef quint32 result_type;
+     void generate(quint32 *begin, quint32 *end) Q_DECL_NOEXCEPT_EXPR(FillBufferNoexcept);
+ 
+     // For std::mersenne_twister_engine implementations that use something
+-- 
+2.23.0
+
-- 
2.23.0

[Buildroot] [git commit branch/2019.02.x] qt5base: Add patch to fix compile issue with gcc9

[Peter Korsgaard]

>>>>> "Julien" == Julien B?raud <julien.beraud@orolia.com> writes:

 > Backport:
 > https://github.com/qt/qtbase/commit/a52d7861edfb5956de38ba80015c4dd0b596259b#diff-0b4819f9df9c43297ba840d1ac401599
 > Signed-off-by: Julien Beraud <julien.beraud@orolia.com>

Ehh, I got a bit confused first as this looked like a git commit
notification.

Why would this be applicable for 2019.02.x? We don't offer gcc9 support
in 2019.02.x.

-- 
Bye, Peter Korsgaard

[Buildroot] [git commit branch/2019.02.x] qt5base: Add patch to fix compile issue with gcc9

[Julien Béraud]

On 19/09/2019 17:26, Peter Korsgaard wrote:
>>>>>> "Julien" == Julien B?raud <julien.beraud@orolia.com> writes:
> 
>   > Backport:
>   > https://github.com/qt/qtbase/commit/a52d7861edfb5956de38ba80015c4dd0b596259b#diff-0b4819f9df9c43297ba840d1ac401599
>   > Signed-off-by: Julien Beraud <julien.beraud@orolia.com>
> 
> Ehh, I got a bit confused first as this looked like a git commit
> notification.

Sorry I am just referring to the commit in the upstream project

> 
> Why would this be applicable for 2019.02.x? We don't offer gcc9 support
> in 2019.02.x.

This issue happens when trying to build qt5 on buildroot on a machine 
that has gcc9 (native) installed. It is not a cross-compilation issue. 
Do you mean that you don't want to support that ?

Julien

[Buildroot] [git commit branch/2019.02.x] qt5base: Add patch to fix compile issue with gcc9

[Peter Korsgaard]

>>>>> "Julien" == Julien B?raud <julien.beraud@orolia.com> writes:

 > On 19/09/2019 17:26, Peter Korsgaard wrote:
 >>>>>>> "Julien" == Julien B?raud <julien.beraud@orolia.com> writes:
 >> 
 >> > Backport:
 >> > https://github.com/qt/qtbase/commit/a52d7861edfb5956de38ba80015c4dd0b596259b#diff-0b4819f9df9c43297ba840d1ac401599
 >> > Signed-off-by: Julien Beraud <julien.beraud@orolia.com>
 >> 
 >> Ehh, I got a bit confused first as this looked like a git commit
 >> notification.

 > Sorry I am just referring to the commit in the upstream project

 >> 
 >> Why would this be applicable for 2019.02.x? We don't offer gcc9 support
 >> in 2019.02.x.

 > This issue happens when trying to build qt5 on buildroot on a machine 
 > that has gcc9 (native) installed. It is not a cross-compilation issue. 
 > Do you mean that you don't want to support that ?

No, that we do want to support - But this wasn't clear from your commit
message.

What about qt 5.6? Do we need a similar patch there?

-- 
Bye, Peter Korsgaard

[Buildroot] [git commit branch/2019.02.x] qt5base: Add patch to fix compile issue with gcc9

[Thomas Petazzoni]

On Thu, 19 Sep 2019 16:07:33 +0000
Julien B?raud <julien.beraud@orolia.com> wrote:

> On 19/09/2019 17:26, Peter Korsgaard wrote:
> >>>>>> "Julien" == Julien B?raud <julien.beraud@orolia.com> writes:  
> >   
> >   > Backport:
> >   > https://github.com/qt/qtbase/commit/a52d7861edfb5956de38ba80015c4dd0b596259b#diff-0b4819f9df9c43297ba840d1ac401599
> >   > Signed-off-by: Julien Beraud <julien.beraud@orolia.com>  
> > 
> > Ehh, I got a bit confused first as this looked like a git commit
> > notification.  
> 
> Sorry I am just referring to the commit in the upstream project

That's not what Peter is talking about. He is talking about the prefix
of your patch. Normally, patches are like this:

 [PATCH] package/qt5base: something

or:

 [PATCH 2019.02.x] package/qt5base: something

But you sent it as:

 [git commit branch/2019.02.x] ...

If you look at the mailing list traffic, e-mails prefixed with [PATCH]
are actual patches submitted by contributors, while e-mails prefixed
with [git commit] are auto-generated e-mails that get sent when the
Buildroot maintainers merge some patches and push them as commits to
the official Git repo.

Hence, your patch was quite confusing, because its title prefix looked
like a Git commit notification.

Thomas
-- 
Thomas Petazzoni, CTO, Bootlin
Embedded Linux and Kernel engineering
https://bootlin.com

[Buildroot] [git commit branch/2019.02.x] qt5base: Add patch to fix compile issue with gcc9

[Thomas Petazzoni]

On Thu, 19 Sep 2019 07:36:05 +0000
Julien B?raud <julien.beraud@orolia.com> wrote:

> Backport:
> https://github.com/qt/qtbase/commit/a52d7861edfb5956de38ba80015c4dd0b596259b#diff-0b4819f9df9c43297ba840d1ac401599
> Signed-off-by: Julien Beraud <julien.beraud@orolia.com>

With an extended commit log explaining that it fixes an issue when
building Qt5 on a machine with gcc9 as its system compiler, and a blank
line before the SoB:

Acked-by: Thomas Petazzoni <thomas.petazzoni@bootlin.com>

(I actually had the exact same patch sitting around in one of my
branch, as I encountered the same issue working on 2019.02 Qt5 project)

Thanks,

Thomas
-- 
Thomas Petazzoni, CTO, Bootlin
Embedded Linux and Kernel engineering
https://bootlin.com

[Buildroot] [git commit branch/2019.02.x] qt5base: Add patch to fix compile issue with gcc9

[Julien Béraud]

On 19/09/2019 21:48, Thomas Petazzoni wrote:
> That's not what Peter is talking about. He is talking about the prefix
> of your patch. Normally, patches are like this:
> 
>   [PATCH] package/qt5base: something
> 
> or:
> 
>   [PATCH 2019.02.x] package/qt5base: something
> 
> But you sent it as:
> 
>   [git commit branch/2019.02.x] ...
> 
> If you look at the mailing list traffic, e-mails prefixed with [PATCH]
> are actual patches submitted by contributors, while e-mails prefixed
> with [git commit] are auto-generated e-mails that get sent when the
> Buildroot maintainers merge some patches and push them as commits to
> the official Git repo.
> 
> Hence, your patch was quite confusing, because its title prefix looked
> like a Git commit notification.

Thanks for the explanation. So the correct way to send a patch for a 
specific branch is :
[PATCH branch-name] ... ?

Regards,
Julien

[Buildroot] [git commit branch/2019.02.x] qt5base: Add patch to fix compile issue with gcc9

[Thomas Petazzoni]

On Fri, 20 Sep 2019 07:50:34 +0000
Julien B?raud <julien.beraud@orolia.com> wrote:

> Thanks for the explanation. So the correct way to send a patch for a 
> specific branch is :
> [PATCH branch-name] ... ?

Yes, with the exception that you can/should omit branch-name when the
target branch is master. So essentially, it's only for special cases (a
patch for the next branch, when the next branch is open, or for a
LTS/maintenance branch) that you need to specify the branch name. When
there's no branch name, we assume your patch targets master.

Thomas
-- 
Thomas Petazzoni, CTO, Bootlin
Embedded Linux and Kernel engineering
https://bootlin.com

[Buildroot] [git commit branch/2019.02.x] qt5base: Add patch to fix compile issue with gcc9

[Julien Béraud]

On 20/09/2019 09:53, Thomas Petazzoni wrote:

>> Thanks for the explanation. So the correct way to send a patch for a
>> specific branch is :
>> [PATCH branch-name] ... ?
> 
> Yes, with the exception that you can/should omit branch-name when the
> target branch is master. So essentially, it's only for special cases (a
> patch for the next branch, when the next branch is open, or for a
> LTS/maintenance branch) that you need to specify the branch name. When
> there's no branch name, we assume your patch targets master.

Thanks very much.

[Buildroot] [git commit branch/2019.02.x] qt5base: Add patch to fix compile issue with gcc9

[Julien Béraud]

On 19/09/2019 18:46, Peter Korsgaard wrote:

> 
> What about qt 5.6? Do we need a similar patch there?
> 

I can't build 5.6 due to other compilation errors so I can't tell sorry.

Regards,
Julien

[Buildroot] [git commit branch/2019.02.x] qt5base: Add patch to fix compile issue with gcc9

[Peter Korsgaard]

commit: https://git.buildroot.net/buildroot/commit/?id=90d8317a940996fea9d4e6246f4c7df474e874f1
branch: https://git.buildroot.net/buildroot/commit/?id=refs/heads/2019.02.x

Fixes an issue when building Qt5 on a machine that has gcc9 as the
system compiler.

Original commit in qt5base:
https://github.com/qt/qtbase/commit/a52d7861edfb5956de38ba80015c4dd0b596259b

Signed-off-by: Julien Beraud <julien.beraud@orolia.com>
Acked-by: Thomas Petazzoni <thomas.petazzoni@bootlin.com>
[Peter: drop patch number]
Signed-off-by: Peter Korsgaard <peter@korsgaard.com>
---
 .../5.11.3/0006-Fix-compile-issue-with-gcc-9.patch | 32 ++++++++++++++++++++++
 1 file changed, 32 insertions(+)

diff --git a/package/qt5/qt5base/5.11.3/0006-Fix-compile-issue-with-gcc-9.patch b/package/qt5/qt5base/5.11.3/0006-Fix-compile-issue-with-gcc-9.patch
new file mode 100644
index 0000000000..0a5507456d
--- /dev/null
+++ b/package/qt5/qt5base/5.11.3/0006-Fix-compile-issue-with-gcc-9.patch
@@ -0,0 +1,32 @@
+From e094806951ff7337b5b0c534db479e3808f153a7 Mon Sep 17 00:00:00 2001
+From: Allan Sandfeld Jensen <allan.jensen@qt.io>
+Date: Tue, 13 Nov 2018 16:00:23 +0100
+Subject: [PATCH] Fix compile issue with gcc 9
+
+It appears messenne_twisters in the latest libstdc++ has one more
+requirement before it is willing to construct with our
+SystemGenerator struct as an sseq provider.
+
+Change-Id: If38151d1fa6f40a80274acc26d9ed6b4ac6049fe
+Reviewed-by: Giuseppe D'Angelo <giuseppe.dangelo@kdab.com>
+Reviewed-by: Thiago Macieira <thiago.macieira@intel.com>
+Signed-off-by: Julien Beraud <julien.beraud@orolia.com>
+---
+ src/corelib/global/qrandom.cpp | 1 +
+ 1 file changed, 1 insertion(+)
+
+diff --git a/src/corelib/global/qrandom.cpp b/src/corelib/global/qrandom.cpp
+index ebf9864b15..577736a867 100644
+--- a/src/corelib/global/qrandom.cpp
++++ b/src/corelib/global/qrandom.cpp
+@@ -218,6 +218,7 @@ struct QRandomGenerator::SystemGenerator
+ #endif // Q_OS_WINRT
+ 
+     static SystemGenerator &self();
++    typedef quint32 result_type;
+     void generate(quint32 *begin, quint32 *end) Q_DECL_NOEXCEPT_EXPR(FillBufferNoexcept);
+ 
+     // For std::mersenne_twister_engine implementations that use something
+-- 
+2.23.0
+