Andre Poenitz wrote:
On Thu, Jul 10, 2008 at 06:09:28PM +0200, Abdelrazak Younes wrote:
Index: src/Undo.cpp
===================================================================
--- src/Undo.cpp (revision 25531)
+++ src/Undo.cpp (working copy)
@@ -73,7 +73,7 @@
MathData * ar, BufferParams const & bp,
bool ifb) :
kind(kin), cursor(cur), cell(cel), from(fro), end(en),
- pars(pl), array(ar), bparams(bp), isFullBuffer(ifb)
+ pars(pl), array(ar), bparams(&bp), isFullBuffer(ifb)
{}
/// Which kind of operation are we recording for?
UndoKind kind;
@@ -90,7 +90,7 @@
/// the contents of the saved MathData (for mathed)
MathData * array;
/// Only used in case of full backups
- BufferParams bparams;
+ BufferParams const * bparams;
/// Only used in case of full backups
bool isFullBuffer;
private:
@@ -321,8 +321,8 @@
if (undo.isFullBuffer) {
LASSERT(undo.pars, /**/);
// This is a full document
- otherstack.top().bparams = buffer_.params();
- buffer_.params() = undo.bparams;
+ otherstack.top().bparams = &buffer_.params();
+ buffer_.params() = *undo.bparams;
But if the idea is to store the params _because they change_,
that should be a copy, i.e.
otherstack.top().bparams = new BufferParams(buffer_.params())
with a corresponing delete somewhere, shouldn't it?
Ah... right, but then that would be a memory leak... I guess we need a
shared_pointer here...
Abdel.
