Apache OpenOffice (AOO) Bugzilla – Issue 20491
Q-PCD MSInteroperability-31: stable sorting
Last modified: 2013-08-07 15:14:39 UTC
Task tracking system for Childs PLEASE CREATE CHILD-TASKS IN ISSUEZILLA! Source Customer Category Calc Product Requirement Sort grouped results as in other applications Customer Need/Problem Need ability to sort previously sorted data and maintain the initial sorting rather than just the original data. See customer example sent to Dieter and Andreas. Comment - Eng Effort MED Eng Owner Niklas Nebel Product Concept When a cell range is sorted by a set of sort keys in some other applications, those rows that have equal values for all sort keys retain their position relative to each other. In SO/OOo they end up in a random order. If the data had some meaningful order before the sort operation, the other applications' behavior is preferable. We will change the implementation of sorting so that the original order is preserved where sorting criteria are equal. Functional Specification -
added keyword Q-PCD
*** Issue 27925 has been marked as a duplicate of this issue. ***
Could someone with the necessary rights please add "sorting algorithm unstable" to the summary and/or keywords to avoid further duplicates?
I can confirm this problem, and even worse, the sorting order changes each time the document is reloaded into OpenOffice and an identical sorting operation performed. It should at least be consistent, our user confidence level has gone through the floor since discovering this issue.
according to the announcement on releases (http://www.openoffice.org/servlets/ReadMsg?list=releases&msgNo=7503) this issue will be re-targeted to OOo Later.
*** Issue 52506 has been marked as a duplicate of this issue. ***
One can overcome this issue by replacing the "sort ascending" or "sort descending" toolbar buttons with a macro published here: http://codesnippets.services.openoffice.org/Calc/Calc.ImproveSortingCapabilities.snip However, we are still longing for a fix for this issue.
Maybe adding the words sorting to the issue name, could make it clearer...
Its duplicate with 27925 and 22884
I confirm there is still this "sort"-problem in version 2.3 !!
Set target.
Accepted
Started
Plaese have a look at issue #7277 and issue #22758 Mandatory improvements to the sorting functions are: (1) Calc must find out by itself, wether the first row is a header or is containing data. Yes that´s possible, Excel does so. Excel recognizes different data types between first row (text in every column) and the following rows (not text at least in one column). If all cells of the list, header and data rows, contain nothing but text, Excel recognizes different formatting in the first row. (Issue #7277) (2) Sorting must base on the column, where the active cell is. (Issue #22758) (3) Subsequent sorting of a list setting the active cell in different columns should result in a hierarchically sorted list corresponding to the sequence of columns, where the active cell was selected. (Issue #20491)
Created attachment 56313 [details] The patch file is about i20491#-v1.
Created attachment 56378 [details] The patch file is about i20491#-v2.
@maoyg: A quick glance revealed that the sort on original position is done already after any sort field. Please change to do that after all sort fields have been processed and nRes==0
Created attachment 56404 [details] The patch file is about i20491#-v3.
adjusting summary
*** Issue 22884 has been marked as a duplicate of this issue. ***
taking the issue
I added the last patch to CWS "calc47".
Reassigning to QA for verification
Created attachment 58692 [details] TestCaseSpecification
Created attachment 58693 [details] Testdocument for Test Case Specification
verified in internal build cws_calc47
Proposal for modified TestCaseSpecification in order to really proof "stable sorting": http://wiki.services.openoffice.org/wiki/User:Stefanw/Werkstatt/Testcases_Sorting#Stable_sorting
Created attachment 59042 [details] For the records: testcase for this issue
Created attachment 59043 [details] Testfile for testcase
verified in DEV300_m38 I confirm stable sorting. Good job so far. Thank you! However, test procedure fails due to another new regression filed as issue #97619.
verified, fixed --> closed.