Am Dienstag, 1. Januar 2019 17:03:55 CET schrieb Scott Kostyshak
<skost...@lyx.org>:
> On Tue, Jan 01, 2019 at 07:43:02AM +0100, Kornel Benko wrote:
> > Am Montag, 31. Dezember 2018 15:05:27 CET schrieb Scott Kostyshak
> > <skost...@lyx.org>:
> > > The following fails for me on 2.3.x but passes on master:
> > >
> > > ctest -R 'check_Length'
> > >
> > > This is on my updated system (Ubuntu 18.04 with an updated TL 2018).
> > > What does this test actually test and how can I reproduce it manually?
> > > Does it fail for anyone else on 2.3.x? On my old system, it passes on
> > > 2.3.x.
> > >
> > > Scott
> > >
> >
> > If I understand correctly, it should check if the length conversions work
> > as the should.
> > Compare the output of the command
> > "check_Length"
> > with expected data in "src/tests/regfiles/Length"
>
> Thanks. When I execute the file directly, I get a bunch of length output
> (which I guess is expected) and then I get
>
> "Segmentation fault (core dumped)"
>
> at the end.
>
> I do not get a segmentation fault with master's version. The .cpp files
> are identical. There is a difference in an included file:
>
> --- /home/scott/lyxbuilds/2.3.x/repo/src/Length.h 2018-08-08
> 18:22:27.482962193 -0400
> +++ src/Length.h 2018-07-30 17:28:33.680910801 -0400
> @@ -20,8 +20,6 @@
>
> namespace lyx {
>
> -class MetricsBase;
> -
> // Solaris/x86 version 9 and earlier define these
> #undef PC
> #undef SP
> @@ -98,12 +96,6 @@
> */
> int inPixels(int text_width, int em_width = 0) const;
>
> - /** return the on-screen size of this length
> - *
> - * This version of the function uses the current inset width as
> - * width and the EM value of the current font.
> - */
> - int inPixels(MetricsBase const &) const;
> /// return the value in Big Postscript points.
> /// Caution: Inaccurate for em, ex, mu and percent units.
> int inBP() const;
> $
>
> I'm not sure if this is worth spending time on, since it works in
> master, and I don't know if it causes any real issue to the user in
> 2.3.x. Let me know if you disagree, and I will spend time to try to fix
> the issue in 2.3.x.
>
> Scott
I would try to get a backtrace under gdb.
Here, I don't get a core dump.
Kornel
signature.asc
Description: This is a digitally signed message part.
