Georg Baum wrote:
If you do
if (cur.pit() > cur.lastpit())
cur.pit() = cur.lastpit();
then it is even suitable for 1.4.3 IMHO. I am not sure whether the test is
necessary, but cur.pit() > cur.lastpit() is clearly wrong.
There are also other places where this is used, always without such a
test. So I'm not sure about it.
Attached an updated patch that fixes the crash and bug.
Joost
Index: insets/insettabular.C
===================================================================
--- insets/insettabular.C (revision 15086)
+++ insets/insettabular.C (working copy)
@@ -684,6 +684,7 @@
if (insertAsciiString(cur.bv(), clip, false)) {
// content has been replaced,
// so cursor might be invalid
+ cur.pit() = cur.lastpit();
cur.pos() = cur.lastpos();
bvcur.setCursor(cur);
break;
@@ -733,8 +734,10 @@
// cursor follows cell:
cur.idx() = tabular.getCellNumber(i, j);
// select this cell only:
+ cur.pit() = 0;
cur.pos() = 0;
cur.resetAnchor();
+ cur.pit() = cur.lastpit();
cur.pos() = cur.top().lastpos();
cur.setCursor(cur);
cur.setSelection();
@@ -1757,6 +1760,7 @@
}
// cursor position might be invalid now
+ cur.pit() = cur.lastpit();
cur.pos() = cur.lastpos();
cur.clearSelection();
}
