On Mon, Feb 24, 2020 at 04:37:00PM -0500, Scott Kostyshak wrote:
> On Mon, Feb 24, 2020 at 09:56:41PM +0100, Enrico Forestieri wrote:
> > On Mon, Feb 24, 2020 at 02:02:32PM -0500, Scott Kostyshak wrote:
> > > On Mon, Feb 24, 2020 at 06:39:25PM +0100, Enrico Forestieri wrote:
> > > > On Mon, Feb 24, 2020 at 11:05:51AM -0500, Scott Kostyshak wrote:
> > > > >
> > > > > Apparently another approach would be to add the following:
> > > > >
> > > > > memset(&padded_event, 0, sizeof(padded_event));
> > > > >
> > > > > Valgrind does not complain when this line is added to the union patch.
> > > >
> > > > I am baffled. The last suggestion I have is trying
> > > >
> > > > alignas(32) xcb_selection_notify_event_t nev = {0};
> > >
> > > Valgrind still gives the error. I attach the full Valgrind log.
> >
> > Sigh. The final attempt would be doing the same as it is done in Qt:
> >
> > struct alignas(32) padded_event
> > : xcb_selection_notify_event_t {};
> > padded_event nev = {};
> >
> > If also this fails, I give up.
>
> It works! Please commit since I have no idea what is going on in this
> patch. In case it saves you time, attached is the diff and here's a
> commit message you can correct:
>
> Cleaner fix to ensure 32-bit XCB events
>
> This fix still satisfies Valgrind and is cleaner than the approach
> at 19c41bd0: instead of using calloc we now use the C++11 specifier
> alignas. For more info, see the following ML thread:
>
>
> https://www.mail-archive.com/search?l=mid&q=20200219024908.5n4x4osni55gylo3%40tallinn
>
> Thanks for the patch. Although this was frustrating for you, the
> non-valgrind-waiting part was fun for me since I got to learn a little
> bit more about memory and initialization.
Done, thank you.
--
Enrico
--
lyx-devel mailing list
lyx-devel@lists.lyx.org
http://lists.lyx.org/mailman/listinfo/lyx-devel
