Bug#892845: stretch-pu: package openni2/2.2.0.33+dfsg-7+deb9u1

[Julien Cristau <jcristau@debian.org>]

Control: tag -1 + confirmed

On Sat, Jul 07, 2018 at 02:48:23PM +0300, Adrian Bunk wrote:
> Control: tags -1 - moreinfo
> 
> On Mon, Apr 02, 2018 at 03:47:49PM +0200, Julien Cristau wrote:
> > Control: tag -1 moreinfo
> > 
> > On Tue, Mar 13, 2018 at 19:38:48 +0200, Adrian Bunk wrote:
> > 
> > > Package: release.debian.org
> > > Severity: normal
> > > Tags: stretch
> > > User: release.debian.org@packages.debian.org
> > > Usertags: pu
> > > 
> > >   * Fix armhf baseline violation and armel FTBFS caused by NEON usage.
> > >     (Closes: #874220)
> > > 
> > > This aims at a small change without affecting other architectures.
> > > 
> > > Taking the fix from unstable verbatim was also not possible due to
> > > 2.2.0.33+dfsg-10 not building on stretch - the following compile
> > > error is present with gcc 6 but for some reason not with gcc 7:
> > > 
> > > Linux/XnLinuxUSBDevice.cpp: In function 'XnBool handleGetStringDescriptor(XnUSBDevice*, __u16, __u16, __u8)':
> > > Linux/XnLinuxUSBDevice.cpp:401:18: error: dereferencing type-punned pointer will break strict-aliasing rules [-Werror=strict-aliasing]
> > >    *(__u16*)&buf[2] = __cpu_to_le16(USB_LANGUAGE_ENGLISH_US);
> > >                   ^
> > > 
> > > Fixing the typo in ThirdParty/PSCommon/BuildSystem/Platform.generic
> > > (see #874220) might be advisable for unstable, but for stretch the
> > > different change avoids this change (that would also affect other
> > > architectures).
> > 
> > Was the newly-building package tested on armel?
> 
> No, and I don't expect it to get users on armel.
> 
> The point of this pu is to fix the baseline violation on armhf,
> if building the package on armel is considered a problem I can
> error out on armel in debian/rules.
> 
Thanks; I think that'd be best.  Acked with that caveat.

Cheers,
Julien