I, too, have confirmed that this is a real issue in the Mac OS X version of OOo
2.0.3 on Intel (I'm on Mac OS X 10.4.7/Intel). As the original poster indicated,
the bug happens with mixed data. I used the following data to confirm
(interestingly, this data is presented post-sort):

AD4
AE1
AE3
AE3
AE1
AE2
AE3
AJ5
AJ5

Furthermore, I also confirmed that the bug does not exist on the Linux/Intel
2.0.3 version (running on Ubuntu on the same machine under Parallels). It'd be
nice to see this issue's status changed from unconfirmed as putting the above
data into a spreadsheet and trying to sort it would confirm the problem in a
matter of seconds.

Confirmed, because two different users report the same issue.

Hi Eric,

please have a look at this one too.

Frank

please provide some details of the build version, and URL where it can be down loaded.

I would like to try and re-produce this when the version and origin are known - it would also be useful to 
attach a file containing an example - meanwhile I cannot see this problem in my own build m183

Viewing the Help->About OpenOffice.org window yields the following information:

680m7 (Build:9044)

The version number is listed as 2.0.3 and it's the Intel build for Mac OS X
under X11 that is (was until recently?) available from the
porting.openoffice.org Web site.

The small amount of data in my original port entered into a spreadsheet will
demonstrate the problem as soon as a sort of it is attempted, at least on this
build on this platform.

I can consistently reproduce the problem as reported in 2.0.3 maho 680m7(9044) and again in 2.0.4 
maho build 680m4(9070) but not seen in my own build of m183.

The simplest case I could make is in two cells, enter text values such as A1 with A2 below, select the range 
and go to the sort dialog - in the options tab make sure to un-check the "Range contains column labels". 
Each time the same data is sorted again, the sequence is reversed.

This problem is NOT seen in my own build on Mac OSX Intel OOD680_m4 Build 9070.
(a) config.parms: jim-watsons-computer:~/vanilla/config_office jim$ cat *rms
--with-gnu-cp=/sw/bin/cp --enable-fontconfig --with-jdk-home=/System/Library/Frameworks/
JavaVM.framework/Home/ --with-x

(b) cvs diff attached, env file attached

(c) Xcode 2.4 IDE 759.0 Core 757.0 ToolSupport 733.0

(d) jim-watsons-computer:~/vanilla/config_office jim$ gcc -v      
Using built-in specs.
Target: i686-apple-darwin8
Configured with: /private/var/tmp/gcc/gcc-5363.obj~28/src/configure --disable-checking -enable-
werror --prefix=/usr --mandir=/share/man --enable-languages=c,objc,c++,obj-c++ --program-
transform-name=/^[cg][^.-]*$/s/$/-4.0/ --with-gxx-include-dir=/include/c++/4.0.0 --with-
slibdir=/usr/lib --build=powerpc-apple-darwin8 --with-arch=nocona --with-tune=generic --
program-prefix= --host=i686-apple-darwin8 --target=i686-apple-darwin8
Thread model: posix
gcc version 4.0.1 (Apple Computer, Inc. build 5363)

(e) example file and screenshot attached

Created attachment 39341 [details]
screen short sort tests

Created attachment 39342 [details]
example spreadsheet

Created attachment 39343 [details]
environment file

Created attachment 39344 [details]
cvs diff

What else is different? I did not disable gtk, therefore to build vcl i had to include /sw/lib in the SOLARLIB 
before /usr/lib thus: 
export SOLARLIB=' -L/Users/jim/vanilla/solver/680/unxmacxi.pro/lib -L/sw/lib -L/usr/lib -L/usr/
X11R6/lib'

Created attachment 39345 [details]
list of shared libraries used

okay I'll investigate this on this weekend.
Sorry for delay...

*** Issue 69870 has been marked as a duplicate of this issue. ***

ericb->maho

Eike Rathke proposed to rebuild not using system ICU, and see if the bug is still here.

-> maybe an i18npool issue. 

If I copy libicui18n.dylib.26.0 from my build to maho build, then the bug is removed in maho build. 
If I copy that file from maho build into my own build, then the bug is seen in my build.
I do not use --with-system for any libs. 

I'm using Xcode 2.2.1 :) and I suspect that it is Xcode's bug.
How do you think? anyway I'll update Xcode for next releases.

Maho,

I use XCode 2.2 and I got the same bug. IMHO, one of the system libs is simply buggy.

If ever I can find the time, I'll build today (or this week sure) m185 using internal stuff (nothing system).

Stay tuned.

Created attachment 39390 [details]
diff -u nm_maho.txt nm_jim.txt > nm.diff

Created attachment 39392 [details]
diff -u nm_maho2.txt nm_jim2.txt > nm2.diff

Created attachment 39393 [details]
diff of icu build logs

Created attachment 39394 [details]
same warnings for both builds

I switched from Xcode 2.2 to Xcode 2.4,
the problem seems to disapper. So I think this is a bug
of Xcode...

BTW: I'll provide RC3 builds with Xcode 2.4. Please verify this
issue with newer builds.

thanks,
 maho

Hi Maho,

>Le 29 sept. 06 à 04:26, maho@openoffice.org a écrit :
>
>------- Additional comments from maho@openoffice.org Thu Sep 28 19:26:35 -0700 2006 -------
>I switched from Xcode 2.2 to Xcode 2.4,

I'm still using XCode 2.2

>the problem seems to disapper.

Ufff...  :-/

>So I think this is a bug
>of Xcode...

I know Stephan Bergmann  and Tino Rachui are working on  a strange issue concerning strip : too much 
symbols are stripped with Intel version of strip.
Pavel found a difference between bad and good libicu18n concerning exceptions ...
Jim Watson is doing an important work too for the SDK, and everything seems to design XCode as 
faulty.

...the list is growing :-/

If this is confirmed, I invite all developers to switch for XCode2.4
FYI, it should be possible to build OpenOffice.org using XCode2.4 (inside XCode, yes)

>BTW: I'll provide RC3 builds with Xcode 2.4. Please verify this issue with newer builds.

I'd suggest you ask for a verification on QA list too. this way, all testers will confirm.
Time to download XCode2.4 ...

*** Issue 69981 has been marked as a duplicate of this issue. ***

The sort problem is not seen in my tests using maho 680_m5 Build 9073 packaged as 
OOo_2.0.4rc3_060930_MacOSXIntel_en-US.dmg

With OOo_2.0.4rc3_060930_MacOSXIntel_de.dmg I downloaded yesterday the sort problem is gone. I 
tested the sort with some little spreadsheets and with a greater one where I detected it.

Thanks

This problem on Mac Intel is fixed by building 2.0.4rc3 with Xcode 2.4.

Closed

thanks all!