Re: mdds broken

2013-05-22 Thread Stephan Bergmann

On 05/22/2013 03:08 AM, Kohei Yoshida wrote:

On 05/21/2013 05:17 AM, Stephan Bergmann wrote:

Building recent master, at least on Fedora 18 with --enable-dbgutil,
fails in various tests due to problems in mdds headers, likely since
http://cgit.freedesktop.org/libreoffice/core/commit/?id=9785579ed0ec8e2e3e826c7e4cbbcaf51f8cb0d4
Update mdds to 0.8.1.

[...]

Anyway, this commit

http://cgit.freedesktop.org/libreoffice/core/commit/?id=4169a1264f8b491e80740c82bada9f6c34e4d26e


*should* fix it.  There were other uno api test breakages even after the
fix, though, but none of them looked related to the mdds problem.


But the mdds-related attempt to compare iterators from different 
sequences failure I described in the original mail is still there.


Stephan
___
LibreOffice mailing list
LibreOffice@lists.freedesktop.org
http://lists.freedesktop.org/mailman/listinfo/libreoffice


Re: mdds broken

2013-05-22 Thread Stephan Bergmann

On 05/22/2013 09:08 AM, Stephan Bergmann wrote:

But the mdds-related attempt to compare iterators from different
sequences failure I described in the original mail is still there.


FYI, with the attached mdds-broken.patch to disable the tests affected 
by that mdds-related attempt to compare iterators from different 
sequences failure, and a local revert of 
http://cgit.freedesktop.org/libreoffice/core/commit/?id=d256dbede60533369d1aac64cca34721183f6a8a 
fdo#46808, Convert chart2::Title service to new style to work around a 
broken JunitTest_chart2_unoapi (sch.ChXChartDocument, sch.ChartTitle; of 
which Noel G. is informed), a current master 
(http://cgit.freedesktop.org/libreoffice/core/commit/?id=ea2f06d7884f2b765394bff8a1e935f67f6c961d 
Ensure SolarMutex is locked at DBG_TESTSOLARMUTEX checkpoints) make 
check works for me again (Fedora 18 x86_64, --enable-dbgutil).


Stephan
diff --git a/sc/Module_sc.mk b/sc/Module_sc.mk
index 0b3970f..ae8a53d 100644
--- a/sc/Module_sc.mk
+++ b/sc/Module_sc.mk
@@ -66,7 +66,6 @@ $(eval $(call gb_Module_add_subsequentcheck_targets,sc,\
 CppunitTest_sc_namedrangeobj \
 CppunitTest_sc_namedrangesobj \
 CppunitTest_sc_tablesheetobj \
-CppunitTest_sc_tablesheetsobj \
 CppunitTest_sc_editfieldobj_cell \
 CppunitTest_sc_editfieldobj_header \
 CppunitTest_sc_modelobj \
diff --git a/sc/qa/unoapi/sc.sce b/sc/qa/unoapi/sc.sce
index 4ececd9..ac3b540 100644
--- a/sc/qa/unoapi/sc.sce
+++ b/sc/qa/unoapi/sc.sce
@@ -88,7 +88,6 @@
 -o sc.ScIndexEnumeration_FunctionDescriptionEnumeration
 -o sc.ScIndexEnumeration_LabelRangesEnumeration
 -o sc.ScIndexEnumeration_NamedRangesEnumeration
--o sc.ScIndexEnumeration_ScenariosEnumeration
 -o sc.ScIndexEnumeration_SheetCellRangesEnumeration
 -o sc.ScIndexEnumeration_SheetLinksEnumeration
 -o sc.ScIndexEnumeration_SpreadsheetViewPanesEnumeration
___
LibreOffice mailing list
LibreOffice@lists.freedesktop.org
http://lists.freedesktop.org/mailman/listinfo/libreoffice


Re: mdds broken

2013-05-22 Thread Kohei Yoshida

Fixed.

http://cgit.freedesktop.org/libreoffice/core/commit/?id=6958b6a3ce9f28bed2a529939e7740452929a5b3

This particular one has nothing to do with mdds, but the error was on 
the Calc side.


Kohei
___
LibreOffice mailing list
LibreOffice@lists.freedesktop.org
http://lists.freedesktop.org/mailman/listinfo/libreoffice


mdds broken

2013-05-21 Thread Stephan Bergmann
Building recent master, at least on Fedora 18 with --enable-dbgutil, 
fails in various tests due to problems in mdds headers, likely since 
http://cgit.freedesktop.org/libreoffice/core/commit/?id=9785579ed0ec8e2e3e826c7e4cbbcaf51f8cb0d4 
Update mdds to 0.8.1.


I came up with the attached quick-hack workaround for the first problem 
encountered, but only to run into the next problem below in 
JunitTest_sc_unoapi soon after, so gave up.


Kohei, please fix these problems.

Stephan



/usr/lib/gcc/x86_64-redhat-linux/4.7.2/../../../../include/c++/4.7.2/debug/safe_iterator.h:493:
error: attempt to compare iterators from different sequences.

Objects involved in the operation:
iterator lhs @ 0x0x2aaad88af368 {
type = 
N11__gnu_debug14_Safe_iteratorIN9__gnu_cxx17__normal_iteratorIPPN4mdds17multi_type_vectorINS3_3mtv18custom_block_func1ILi51ENS5_21default_element_blockILi51EN2sc12CellTextAttrEE5blockENSt9__cxx19986vectorISE_SaISE_ENSt7__debug6vectorISE_SI_
 (mutable iterator);
  state = past-the-end;
  references sequence with type 
`NSt7__debug6vectorIPN4mdds17multi_type_vectorINS1_3mtv18custom_block_func1ILi51ENS3_21default_element_blockILi51EN2sc12CellTextAttrEE5blockESaISC_EEE'
 @ 0x0x2aaad88af368
}
iterator rhs @ 0x0x2aaacca5b3b0 {
type = 
N11__gnu_debug14_Safe_iteratorIN9__gnu_cxx17__normal_iteratorIPKPN4mdds17multi_type_vectorINS3_3mtv18custom_block_func1ILi51ENS5_21default_element_blockILi51EN2sc12CellTextAttrEE5blockENSt9__cxx19986vectorISE_SaISE_ENSt7__debug6vectorISE_SJ_
 (constant iterator);
  state = past-the-end;
  references sequence with type 
`NSt7__debug6vectorIPN4mdds17multi_type_vectorINS1_3mtv18custom_block_func1ILi51ENS3_21default_element_blockILi51EN2sc12CellTextAttrEE5blockESaISC_EEE'
 @ 0x0x2aaacca5b3b0
}



#13 signal handler called
#14 0x003c80635ba5 in __GI_raise (sig=sig@entry=6) at 
../nptl/sysdeps/unix/sysv/linux/raise.c:63
#15 0x003c80637358 in __GI_abort () at abort.c:90
#16 0x003c82eb193d in __gnu_debug::_Error_formatter::_M_error 
(this=0x2aaacca5b120) at ../../../../../libstdc++-v3/src/c++11/debug.cc:777
#17 0x2aaacd327a87 in __gnu_debug::operator==__gnu_cxx::__normal_iteratormdds::multi_type_vectormdds::mtv::custom_block_func151, mdds::mtv::default_element_block51, sc::CellTextAttr  ::block**, 
std::__cxx1998::vectormdds::multi_type_vectormdds::mtv::custom_block_func151, mdds::mtv::default_element_block51, sc::CellTextAttr  ::block*, 
std::allocatormdds::multi_type_vectormdds::mtv::custom_block_func151, mdds::mtv::default_element_block51, sc::CellTextAttr  ::block*  , 
__gnu_cxx::__normal_iteratormdds::multi_type_vectormdds::mtv::custom_block_func151, mdds::mtv::default_element_block51, sc::CellTextAttr  ::block* const*, 
std::__cxx1998::vectormdds::multi_type_vectormdds::mtv::custom_block_func151, mdds::mtv::default_element_block51, sc::CellTextAttr  ::block*, 
std::allocatormdds::multi_type_vectormdds::mtv::custom_block_func151, mdds::mtv::default_element_block51, sc::CellTextAttr  ::block*  , std::__debug::vectormdds::mu!

lti_type_v
ectormdds::mtv::custom_block_func151, mdds::mtv::default_element_block51, sc::CellTextAttr  ::block*, 
std::allocatormdds::multi_type_vectormdds::mtv::custom_block_func151, mdds::mtv::default_element_block51, 
sc::CellTextAttr  ::block*   (__lhs=, __rhs=) at 
/usr/lib/gcc/x86_64-redhat-linux/4.7.2/../../../../include/c++/4.7.2/debug/safe_iterator.h:490

#18 0x2aaacd322697 in mdds::multi_type_vectormdds::mtv::custom_block_func151, 
mdds::mtv::default_element_block51, sc::CellTextAttr  ::get_block_position 
(this=0x2aaad87abbb0, pos_hint=..., pos=0, start_row=@0x2aaacca5b448: 0, 
block_index=@0x2aaacca5b440: 0) at 
workdir/unxlngx6/UnpackedTarball/mdds/include/mdds/multi_type_vector_def.inl:507
#19 0x2aaacd31da97 in mdds::multi_type_vectormdds::mtv::custom_block_func151, 
mdds::mtv::default_element_block51, sc::CellTextAttr  ::setsc::CellTextAttr 
(this=0x2aaad87abbb0, pos_hint=..., pos=0, value=...) at 
workdir/unxlngx6/UnpackedTarball/mdds/include/mdds/multi_type_vector_def.inl:209
#20 0x2aaacd317da7 in ScColumn::SetCell (this=0x2aaad87abbb0, 
rBlockPos=..., nRow=0, pNewCell=0x2aaafa5c91d0) at 
sc/source/core/data/column2.cxx:1508
#21 0x2aaacd333aa0 in ScColumn::Insert (this=0x2aaad87abbb0, rBlockPos=..., 
nRow=0, pNewCell=0x2aaafa5c91d0) at sc/source/core/data/column3.cxx:83
#22 0x2aaacd29bc02 in ScColumn::CopyToColumn (this=0x2aab64078cf0, 
rCxt=..., nRow1=0, nRow2=3, nFlags=255, bMarked=false, rColumn=..., 
pMarkData=0x2aaacca5c010, bAsLink=false) at sc/source/core/data/column.cxx:1466
#23 0x2aaacd29b6db in ScColumn::CopyToColumn (this=0x2aab64078cf0, 
rCxt=..., nRow1=0, nRow2=1048575, nFlags=255, bMarked=true, rColumn=..., 
pMarkData=0x2aaacca5c010, bAsLink=false) at sc/source/core/data/column.cxx:1386
#24 0x2aaacd5a8ec8 in ScTable::CopyToTable (this=0x2aab64078a60, rCxt=..., 
nCol1=0, nRow1=0, nCol2=1023, nRow2=1048575, nFlags=255, bMarked=true, 

Re: mdds broken

2013-05-21 Thread Kohei Yoshida

On 05/21/2013 05:17 AM, Stephan Bergmann wrote:
Building recent master, at least on Fedora 18 with --enable-dbgutil, 
fails in various tests due to problems in mdds headers, likely since 
http://cgit.freedesktop.org/libreoffice/core/commit/?id=9785579ed0ec8e2e3e826c7e4cbbcaf51f8cb0d4 
Update mdds to 0.8.1.
So, the root cause of this had been there for some time, but the code 
was never used in the libreoffice code until now.


I'll skip the details of the crash in the dbgutil build since I'm a bit 
unsettled to touch that topic. Perhaps we can fancy talking that over 
with a beer or two at the next conference.  I'll just say that it's one 
of those C++ mysteries that I didn't know existed until today.


Anyway, this commit

http://cgit.freedesktop.org/libreoffice/core/commit/?id=4169a1264f8b491e80740c82bada9f6c34e4d26e

*should* fix it.  There were other uno api test breakages even after the 
fix, though, but none of them looked related to the mdds problem.


Kohei

--
Kohei Yoshida, LibreOffice Calc hacker, SUSE.

___
LibreOffice mailing list
LibreOffice@lists.freedesktop.org
http://lists.freedesktop.org/mailman/listinfo/libreoffice