So I can confirm two different behaviors but both wrong (between 3.6.5.2 & 4.1 master build pulled a few days ago).

Because of this I'm going to leave at version 4.0.1.2 as it's worse than it was in 3.6.5.2 (IMO) but both are wrong.


I have been able to confirm the issue on:
Version: 4.1.0.0.alpha0+Build ID: 6e3e6ef7257e93743a72719581ef6fe0016e58e
Date:   Thu Apr 11 15:24:38 2013 +0200 
Platform: Bodhi Linux 2.2 x64

+ + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + +
As I've been able to confirm this problem I am marking as:

New (confirmed)
Major - data loss, xml filter is definitely broken.
High - default, seems appropriate.

Keywords - regression - I am putting regression only because it has gotten worse, it seems like some work was done in the area of the filter and it broke it more. Plus a bibisect might get us closer to useful code pointers

Whiteboard Status - bibisectrequest

+ + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + +
LibreOffice is powered by a team of volunteers, every bug is confirmed (triaged) by human beings who mostly give their time for free. We invite you to join our triaging by checking out this link:
https://wiki.documentfoundation.org/QA/BugTriage and join us on freenode at #libreoffice-qa

There are also other ways to get involved including with marketing, UX, documentation, and of course developing -  http://www.libreoffice.org/get-help/mailing-lists/. 

Lastly, good bug reports help tremendously in making the process go smoother, please always provide reproducible steps (even if it seems easy) and attach any and all relevant material

So after attempting to bibisect - this is not a regression, it's just broken. Updating version to 3.6alpha (verified with bibisect).

Removing bibisectrequest as it's useless in this case

dangit - so another attempt and it seems like it was broken, then fixed, then broken again - apologies for spamming comments


Bibisect below

+++++++++++++++++++++++++++++++++++++++++++++++++++++++++++++++++++++++++++++++

f6275e54c182c5a8ed2368418445d84beb165d58 is the first bad commit
commit f6275e54c182c5a8ed2368418445d84beb165d58
Author: Bjoern Michaelsen <bjoern.michaelsen@canonical.com>
Date:   Mon Dec 10 11:35:50 2012 +0000

    source-hash-d9412fb4755377b8358a46a249cfe29a22ea9451
    
    commit d9412fb4755377b8358a46a249cfe29a22ea9451
    Author:     David Tardon <dtardon@redhat.com>
    AuthorDate: Tue Sep 11 07:31:43 2012 +0200
    Commit:     David Tardon <dtardon@redhat.com>
    CommitDate: Tue Sep 11 07:36:32 2012 +0200
    
        fdo#53520 rename portuguese dicts
    
        Change-Id: I70cb4856f1db4722e886407d1c2fdf6a73b9a7f3

:100644 100644 0953f1686760ca6fac561e87d200e640875764f2 44293ef104571a1eaf6bfecd43d4a01702bdf52c M	ccache.log
:100644 100644 d86efaac0c94c20bff57e3036b4417574d73ce4a 86a37681bb8cfd44d04534e14b93dfd45cf0fe63 M	commitmsg
:100644 100644 5728ca3baa483958445168e9b9fccb6f075c5897 ba1824fe7a9f7615a41309f3f318d62e2c4a7589 M	dev-install.log
:100644 100644 8acfb864b3fb560156573153b434b749e43dcf60 5ee43aee8b36e6868002119bee285d15f5000fe0 M	make.log
:040000 040000 580424baebbc6a88398ef8d8478f1b4c5d6c6219 a363beeec352e9ce16ad9cc261f62b98edc79d12 M	opt


# bad: [5b4b36d87517a6ea96ff8c84c46b12f462fc9a1a] source-hash-8450a99c744e9005f19173e4df35d65640bcf5c4
# good: [65fd30f5cb4cdd37995a33420ed8273c0a29bf00] source-hash-d6cde02dbce8c28c6af836e2dc1120f8a6ef9932
git bisect start 'latest' 'oldest'
# skip: [16b0b88cbd4ef0f51816e97277e40c5cf78f7bf9] source-hash-099198a4224778fe6e43f5dc13b5b9b1b4dc828c
git bisect skip 16b0b88cbd4ef0f51816e97277e40c5cf78f7bf9
# skip: [1b33e76db9e293bea69b7f814ee90a52a10f7a57] source-hash-cac1f33e839469d884730350e46a21d92fb442f2
git bisect skip 1b33e76db9e293bea69b7f814ee90a52a10f7a57
# good: [bc003d6498b1bb1188ed2387d87f6723d5a329c7] source-hash-c8218367a0f52206591a5048848fa5292405b6c3
git bisect good bc003d6498b1bb1188ed2387d87f6723d5a329c7
# bad: [62b2ae5ee4cc77ad0b399c5a716ef526023d13ab] source-hash-e9960f36675a025c0536dec30ae56c50f4adecb1
git bisect bad 62b2ae5ee4cc77ad0b399c5a716ef526023d13ab
# bad: [aefa1263d1642e8d9677d385f492530ae0725fc6] source-hash-e63bba0013e5ce34cd04559632206bb7c891eebe
git bisect bad aefa1263d1642e8d9677d385f492530ae0725fc6
# good: [da69f6b7ef762284f67b1562f7aec6e37206e65b] source-hash-8638f1e72a3fe830c0e8dcc1bd847d4fb9e599ee
git bisect good da69f6b7ef762284f67b1562f7aec6e37206e65b
# good: [358ad9e115542633c030659959937b107661624f] source-hash-b679a2a02180c017bd8b596fb2e4f283bad93b75
git bisect good 358ad9e115542633c030659959937b107661624f
# bad: [f6275e54c182c5a8ed2368418445d84beb165d58] source-hash-d9412fb4755377b8358a46a249cfe29a22ea9451
git bisect bad f6275e54c182c5a8ed2368418445d84beb165d58
# good: [99fb4d5c4493406d720bb58eedd6c79b2c304c77] source-hash-982b7cb498c3ea1106c5d2184f84989d99b1d942
git bisect good 99fb4d5c4493406d720bb58eedd6c79b2c304c77
# good: [fe309e4a83889bed615770908d76f717e3797e05] source-hash-337ef5808dd8e55c06d00b222e69c5ba287acab5
git bisect good fe309e4a83889bed615770908d76f717e3797e05

From commit 3420be984986bcff03d6d127b913fc07372fe89f there is a sequence of commits where attachment 76297 [details] either loads blank or fails to load; from commit a2a10b59876951b6493419713e9054ceabd3d6cc onwards the file loads again but there are rows/cells missing.

There is a separate problem with the file which has always existed (inherited from OOo) that the cell references in the merged cells are wrong. For instance, B1:C1 loads as "=B1+1", but should be "=A1+1" according to what Excel shows. Another bug should really be raised for this, but as it can't be reproduced in any version released since 2012, a proper report may have to wait for this bug to be fixed.

Adding Cc: to pjotr@guineapics.de


commit 3420be984986bcff03d6d127b913fc07372fe89f
Author: Peter Jentsch <pjotr@guineapics.de>
Date:   Tue May 1 00:24:01 2012 +0200

    fixed problem with template recursion in spreadsheetml import.
    
    Changed recursion to iteration in various places (creating column styles and both row styles and rows)
    Recursion when creating cells/columns remains.
    
    Change-Id: Ie5d27c576069b79225e7901671227c0ff8dce75d

...

commit a2a10b59876951b6493419713e9054ceabd3d6cc
Author: Peter Jentsch <pjotr@guineapics.de>
Date:   Tue May 1 00:09:39 2012 +0200

    optimize font-decl template for libxslt (using exslt functions)
    
    Change-Id: I5b8469db0729fdb722d22ae9a0a8688658db39f0

I try to fix this issue in the upcoming weeks. I have a trial version Microsoft Office to verify the fix. Test documents with a corresponding XLS version (to verify the import result) are very welcome.