Problem description: Calc crash when auto filter enabled and empty column header and try to sort table. Steps to reproduce: 1. add column with empty header but with a non-empty cell 2. enable auto filter 3. try to sort table Current behavior: calc crash (maybe segmentation fault) Expected behavior: calc not crash but it could warning the user about header problem. Operating System: Windows 7 Version: 4.2.4.2 release
For the test, could you give a try to dev version? (see http://dev-builds.libreoffice.org/daily/libreoffice-4-2/Win-x86@42/current/) It could be a dup of fdo#78079 (put in See Also)
I tried the dev build, but the crash remain. I give you the step to reproduce bug: - enter "header" in A1 - enter "Data" in A2 and B2 - enter a comment in A2 - leave B1 empty - go to A2 - click on button sort ascending - we can see that header are sorted with others data -> so wrong - i undo the the sort - i retry to sort acscending --> CRASH i have tested this with : Crash with : - Windows XP 4.16, 4.25 (libo-42~2014-05-18_18.05.37_LibreOfficeDev_4.2.5.0.0_Win_x86.msi) - Windows Seven 4.16 - Linux 4.14, 4.20, 4.22 No crash with Debian wheezy LO 3.5.4 I try to define you the bug : When a table include an empty header, the sort not work properly (datas and header are sorted, i think header must stay in place : 1st line ) LO crash only when a comment is inserted on a cell, try to sort, undo the sort, retry the sort.
On pc Debian x86-64 with master sources updated yesterday, I could reproduce the crash once. I meant, I tested by launching LO from console and reproduced the crash. Then I gave a try with gdb to retrieve a backtrace, no crash. Then I gave a new try from console only, no crash! :-( Put it back to UNCONFIRMED.
Confirmed: Ubuntu 14.04 x64 LibreOffice 4.2.4.2 New (confirmed) Major (crasher but only after a specific set of steps that might not be reproduced often) High Going to try to bibisect now
So ridiculous - confirmed it the first time, now I can't for the life of me. @gtrelat - any way that you can get a backtrace for us? Is it consistently crashing for you? https://wiki.documentfoundation.org/Development/How_to_debug#Start_LibreOffice_from_the_debugger_.28gdb.29
Created attachment 100792 [details] gdb log file
Created attachment 101470 [details] bt console On pc Debian x86-64 with master sources updated yesterday, I could reproduce the crash, even in gdb. I attached console logs and bt.
Kohei/Markus/Eike: one for you?
So this is completely unrelated to auto-filter, but somehow to Undo of Sort with cell comment and then re-sort instead (which also the bt suggests). Updating summary for this. Could reproduce in 4-2-6 following steps in comment 2, but no crash in 4-3-0 or master.
FYI the sort undo redo code is totally different now on master, 4.3 and the latest 4.2 branch (although the one in the 4.2 branch was not my intent). This needs to be re-tested with the new code.
Created attachment 105557 [details] To reproduce bugs
Comment on attachment 105557 [details] To reproduce bugs Hi, we do some check with the latest code : 4.2.6-secfix -> crash (seg fault) 4.3.1.2 -> no crash Thanks for the fix in 4.3 branch. But, it remain our principal problem : one or more empty cell in the row header. When sorting or filtering, the row header is not recognize as expected. This cause strange behaviors for the sort and the auto-filter functionality. I attach a calc file to reproduce bugs. example for the sort : (steps to reproduce) 1) select a cell 2) do a sort (Ascendant or Descending) --> header row not recognized, so the row is sorted with others data example for the filter : (steps to reproduce) 1) select a cell 2) do a auto-filter --> a popup dialog appears which ask if we want to use the first row as header. --> answer "Yes" or "No" have the same effects
Can't reproduce crash anymore in current 4-2 (to be 4.2.7)