GitLab

This reverts part of d8730c41.

The commented-out changes would have been wrong because they
were method calls on Entries, not Labels.

This reverts part of d8730c41.

It has mistakenly replaced Entry::set_alignment() with
Widget::set_halign() instead of only replacing Misc::set_alignment().

Now the fields take up their full widths again, so the extra buttons
(New, Find, ..., etc) are not far away.

Conflicts:
	glom/utility_widgets/layoutwidgetbase.cc

Instead of Widget's xalign, which acts on the whole widget,
not just its text content.
See http://www.murrayc.com/permalink/2015/03/02/gtk-aligning-justification-in-text-widgets/

By temporarily unsetting LANGUAGE while getting LC_TIME via
setlocale(LC_TIME, NULL). As Geary does already too.

So we have the correct 4 digit date formats for those too,
even where the default behaviour is incorrect.

Also add a regression test for this.

Bug #742968

* glom/utility_widgets/notebook_noframe.[h|cc]:
  Add a hacky remove_all_pages_without_signalling() method
  to avoid us calling get_parent() while this widget is being
  removed, because that is causing get_parent() to be called on
  a GtkWidget that is no longer valid. This has started happening
  recently so I suspect a change in GTK+ or gtkmm.
  I have not put in the time to identify what is really going
  wrong.
* glom/mode_data/notebook_data.cc: Destructor: Call the new
  workaround method.

Found by clang's scan-build.

get_parent_container_or_self() is not called by anything.
Found by Coverity Scan, which identified dead code because of
an early return, not because the whole method is not used.

Because 1) We don't do this elsewhere and 2) We would have to
handle that excetion somehow, but we don't and we can't.

Found by Coverity Scan as an uncaught exception in main().

Found by scan-build.

Found by Coverity Scan.

Found by Coverity Scan.