https://bugs.freedesktop.org/show_bug.cgi?id=79027

--- Comment #3 from Albrecht Müller <[email protected]> ---
The file containing the hypenation problem shows another one that I consider
much worse: It does not reliably update the contents of the cells. When I use a
calculation program the last thing I expect is that I have to double check the
numbers by hand, and that is exactly the consequence of this problem.

I report it here as it seems to be related to the original problem. It occurs
in the same file, and some operations on cells are not performed correctly. One
is hyphenation, the other is updating (recalculating) the cell content. While
the hyphenation problem is easily to reproduce (see previous comment), I did
not find a way to create an example file that shows the other one.  Therefore I
can give only a few observations here. They are based on experiments with a
file I happened to save in a state which allows to reproduce the problem
reliably.

(1) The main symptom is that there are cells that do not display the correct
results of the formula they contain. The numbers displayed seem to come from
some previous calculations despite the fact that automatic recalculation is on
and works for other cells. Pressing F9 to force a recalculation does not change
anything. The formula reference chain seems to be broken.

(2) Maybe the problem has common roots with bug 43003, which I could reproduce
using  LibreOffice calc Version: 4.2.4.2 Build-ID:
63150712c6d317d27ce2db16eb94c2f3d7b699f8. It is not the same problem as I do
not use array functions, only built in functions like sum ("Summe" in German)
and names for cells and cell ranges. Bug 44143 seems to be fixed in this
version of Calc.

(3) The problem seems to require a speadsheed of some minimum complexity. I was
not able to build an example from scratch.

(4) This problem is difficult to observe, as it occurs only rarely and it seems
to disappear in many (but not all) cases when a cell is changed on which the
content of the cell in question depends.  Seems as if the broken reference
chain gets repaired somehow. On the other hand a single table can contain quite
a few cells containing wrong results.

(5) A cell that refers to other cells only by names should show the same value
as the original when it is copied to some other place.  The copy of a cell with
a wrong value shows the correct value at the target location. The net effect is
that you can have two cells containing exactly the same formulas referring to
exactly the same cells but showing different values. This behavior differs from
the example in bug 43003. When I copied the cells F1 (after having changed A1
to $A$1) to some other places all the copies showed the same wrong value.

(6) Undo does not work in the sense that it does not restore exactly the
previous state: Adding a blank character at the end of formula in some cell
which shows a wrong result usually does not change the semantics of the formula
but forces the cell to show the correct result. The undo operation does not
restore the wrong result.  I tried this trick with the example given in bug
43003. It did work there on cell F1 (containing "=INC(A1)"), but not on matrix
B1:C1 (containing '=INCDINC(A1)"). The formula reference chain seems to remain
broken in this case.

-- 
You are receiving this mail because:
You are the assignee for the bug.
_______________________________________________
Libreoffice-bugs mailing list
[email protected]
http://lists.freedesktop.org/mailman/listinfo/libreoffice-bugs

Reply via email to