Age | Commit message (Collapse) | Author | Files | Lines |
|
This is 'libreoffice-3-3-3' - the stable branch for the 3.3.3 release.
Only very safe changes, reviewed by three people are allowed.
If you want to commit more complicated fix for the next 3.3.x release,
please use the 'libreoffice-3-3' branch.
If you want to build something cool, unstable, and risky, use master.
|
|
|
|
Signed-off-by: Kohei Yoshida <kyoshida@novell.com>
|
|
Sometimes a simple sheet removal causes segfault in certain types of
documents. Not always reproducible, but this fix is certainly safer.
(n#680272)
Signed-off-by: Noel Power <nopower@novell.com>
|
|
A hyphen character caused auto-complete to kick in, changing the user's choice of conditional operator to "=". This was not changed back if auto-complete option is overridden.
Signed-off-by: Kohei Yoshida <kyoshida@novell.com>
|
|
|
|
Failure to invalidate sheet streams during reference update caused
formula cross referenceing between sheets to totallly get borked.
The bug was originally reported in i#116833.
Signed off by Petr Mladek <pmladek@suse.cz>
|
|
This fixes fdo#34350.
Signed-off-by: Michael Meeks <michael.meeks@novell.com>
|
|
Commit hash: e301739e4439be4214de0ff9b1bd6c2142d6806e
Signed-off-by: Thorsten Behrens <thb@documentfoundation.org>
|
|
Block selection with keyboard was incorrect when the selected range
included hidden cells. This also fixes one crasher on block
selection during formula reference mode i.e. type '=SUM(' and select
range via keyboard afterward.
|
|
Turns out that the token array's recalc mode is set to "recalc always"
when it contains a volatile token during compilation. We can re-use
that to mark the token array volatile before the interpretation starts.
The end result is the same, with better performance since we can avoid
re-scanning of the token array.
Signed-off-by: Michael Meeks <michael.meeks@novell.com>
|
|
When a volatile token is inside a conditional function (such as IF),
*and* the evaluation of that conditional skips the volatile token
it would incorrectly mark the cell non-volatile. The solution is
to scan through all tokens in the token array in the beginning of
the calculation and mark the cell volatile if one is found.
|
|
FT_PREC overwrote BTN_GENERAL_PREC so that French translation was truncated
|
|
When making range selection by mouse, Calc wouldn't highlight the
new range in the column and row headers. It was because we weren't
calling ScTabView::SelectionChanged() when the selection range
is updated. This SelectionChanged() call then puts in queue a
re-paint request for column and row headers.
|
|
Signed-off-by: Michael Meeks <michael.meeks@novell.com>
|
|
When the window is large enough to show several 'Page X' strings,
the page number was not properly incremented. This commit fixes
it.
|
|
When a cell with non-default formatting attribute starts with non-first
row in a column, the filter would incorrectly apply the same format to
all the cells above it if they didn't have any formats. This commit
fixes it.
|
|
I created a patch between before and after the merge commit, and
reverse-applied it. This is because if you revert a merge commit
in git you are denying all future merges from that branch. I wasn't
sure if we really wanted that.
Signed-off-by: Petr Mladek <pmladek@suse.cz>
|
|
This reverts commit 0fe55349b05fae496a86a3fa958d184f0453916c.
Signed-off-by: Petr Mladek <pmladek@suse.cz>
|
|
This is 'libreoffice-3-3' - the stable branch for the 3.3.x releases.
Only safe changes, reviewed by anoter person are allowed.
If you want to build something cool, unstable, and risky, use master.
|
|
This is 'libreoffice-3-3-0' - the stable branch for the 3.3.0 release.
Only very safe changes, reviewed by three people are allowed.
If you want to commit more complicated fix for the next 3.3.x release,
please use the 'libreoffice-3-3' branch.
If you want to build something cool, unstable, and risky, use master.
|
|
|
|
|
|
Conflicts:
sc/source/core/data/table2.cxx
sc/source/core/data/table3.cxx
|
|
|
|
|
|
Signed-off-by: Michael Meeks <michael.meeks@novell.com>
|
|
|
|
|
|
We apparently put numerical sheet name into quotes somewhere else,
so checking for a number followed by a '.' and flagging it as bad
is not only necessary, it's not appropriate. E.g. we do support
other formula syntax where the sheet-to-reference separator is not
'.' (Excel A1 and Excel R1C1), thereby making this test even more
inappropriate.
Signed-off-by: Jan Holesovsky <kendy@suse.cz>
|
|
The same problem happens even when the object is not a cell note,
but, say, a regular text box. Let's not check whether or not the
object is a cell note.
Signed-off-by: Caolán McNamara <caolanm@redhat.com>
|
|
Original patch is from Caolan. I re-worked it to handle more use
cases.
Signed-off-by: Caolán McNamara <caolanm@redhat.com>
|
|
|
|
|
|
|
|
|
|
|
|
Also, increment the char pointer and use it directly, which is
faster than accessing char via [] in each iteration.
|
|
* Apparently ScTable::GetFirstDataPos() never worked as intended and selected
the first non-empty row of the first non-empty column instead of also taking
following columns into account where non-empty rows may be on top of that.
* Caller of ScDocument::ShrinkToDataArea() must check the return value, if
false there is no data in the area passed and the values obtained may be out
of bounds, not in order or unmodified.
* In ScExternalRefMgr a call of ScDocument::ShrinkToUsedDataArea() was
intended instead of ScDocument::ShrinkToDataArea(). For this changed the
return of ShrinkToUsedDataArea() to flag if there is any data and added an
out parameter to flag if the area was shrunk.
* Sanitized arguments and slightly optimized ScTable::ShrinkToUsedDataArea()
to not have to check for >0 and <MAX on each iteration.
(transplanted from 31b53a2ec7e32f9560464f8a3b9f6c3e07b29083)
|
|
This string was hard-coded in the code to be always in English. This
change makes it localizable.
Signed-off-by: Caolán McNamara <caolanm@redhat.com>
|
|
I had accidentally removed the part that constructed correct string
to display. Recovered it, and modified it to use OUStringBuffer
instead of sprintf to construct the string.
Signed-off-by: Norbert Thiebaud <nthiebaud@gmail.com>
|
|
|
|
|
|
This fixes fdo#31805.
|
|
This commit happily fixes fdo#31796.
|
|
|
|
|
|
|
|
Visible range of the current sheet is used to decide whether or not
to render cell cursor. Failing to do that was causing the cursor
to disappear when switching sheets while the cursor on the current
sheet was invisible. (fdo#31753)
|
|
I was supposed to use SCSIZE to store column / row position but was
incorrectly using sal_uInt8. This caused import of outlines from
Excel document to entirely get screwed when the outline positions
were > 255, and it was understandably so. :-P (fdo#31833)
|