Crash on AddRecord

[Francois Van Lerberghe]

Mmmh, I've talk too fast.

With V4RB v4.8.1, no more crash
replacing the LOCATE() function with some IF().

With V4RB v4.9.1 or v4.9, the crash is always there,
on the very same record. This is the crash log :

0   libvkernel_fat_release.dylib      0x1917d7ee
fbl::vu_strcoll_uu_icu(unsigned short const*, int, unsigned short const*,
int, fbl::I_Collator*) + 130
1   libvkernel_fat_release.dylib      0x190b6d54
fbl::ENode_Pred_Equal_Str::long_val(unsigned int) + 228
2   libvkernel_fat_release.dylib      0x190b6104
fbl::ENode_Pred_CmpOp::llong_val(unsigned int) + 24
3   libvkernel_fat_release.dylib      0x190a4d07
fbl::ENode_Func_If::llong_val(unsigned int) + 31
4   libvkernel_fat_release.dylib      0x1916f105
fbl::AssignToValue(fbl::smart_ptr<fbl::I_Value>,
fbl::smart_ptr<fbl::I_ENode>, unsigned int) + 191
5   libvkernel_fat_release.dylib      0x19056703
fbl::Method_SQL::Evaluate(unsigned int, fbl::smart_ptr<fbl::I_Value>) + 221
6   libvkernel_fat_release.dylib      0x19117a71
fbl::policyMethodYes::DoMethod(fbl::FldStorage*, unsigned int,
fbl::I_Value*) + 95
7   libvkernel_fat_release.dylib      0x194007fc
fbl::FldStorageMaster_T<fbl::policyFileNo, fbl::policyNullNo,
fbl::policyMethodYes>::ReadValue(unsigned int, fbl::smart_ptr<fbl::I_Value>)
+ 62
8   libvkernel_fat_release.dylib      0x190f4372
fbl::Field_Imp::ReadValue(unsigned int) + 60
9   libvkernel_fat_release.dylib      0x1913fdb7
fbl::Table::ReEvaluateAllMethods(unsigned int) + 101
10  libvkernel_fat_release.dylib      0x1913fe8c
fbl::Table::AddRecord_WithOut_Triggers_ex() + 156
11  libvkernel_fat_release.dylib      0x1913f101
fbl::Table::AddRecord_WithOut_OnEachStatement_Triggers_ex() + 187
12  libvkernel_fat_release.dylib      0x1913f304 fbl::Table::AddRecord() +
478
13  libvkernel_fat_release.dylib      0x19148b0b
fbl::Table_Indirect::AddRecord() + 229
14  libvkernel_fat_release.dylib      0x19153542
fbl::vsql::Cursor::AddRecord() + 258
15  ..._macho_ub_4_9_1.rbx_0.dylib    0x18dc068a
Cursor_AddRecord(REALobjectStruct*) + 114

With V4RB v5beta, the crash is always there, on the very same record,
but with another crash log :

0   ???                               0000000000 0 + 0
1   libvkernel_fat_release.dylib      0x191e01db
fbl::ENode_Pred_Equal_Str::long_val(unsigned int) + 243
2   libvkernel_fat_release.dylib      0x191de4a2
fbl::ENode_Pred_CmpOp::llong_val(unsigned int) + 24
3   libvkernel_fat_release.dylib      0x191cb096
fbl::ENode_Func_If::llong_val(unsigned int) + 32
4   libvkernel_fat_release.dylib      0x192a9359
fbl::AssignToValue(fbl::smart_ptr<fbl::I_Value>,
fbl::smart_ptr<fbl::I_ENode>, unsigned int) + 278
5   libvkernel_fat_release.dylib      0x19174593
fbl::Method_SQL::Evaluate(unsigned int, fbl::smart_ptr<fbl::I_Value>) + 229
6   libvkernel_fat_release.dylib      0x19248e67
fbl::policyMethodYes::DoMethod(fbl::FldStorage*, unsigned int,
fbl::I_Value*) + 93
7   libvkernel_fat_release.dylib      0x195553bc
fbl::FldStorageMaster_T<fbl::policyFileNo, fbl::policyNullNo,
fbl::policyMethodYes>::ReadValue(unsigned int, fbl::smart_ptr<fbl::I_Value>)
+ 48
8   libvkernel_fat_release.dylib      0x1921c05b
fbl::Field_Imp::ReadValue(unsigned int) + 57
9   libvkernel_fat_release.dylib      0x1927b960
fbl::Table::ReEvaluateAllMethods(unsigned int) + 94
10  libvkernel_fat_release.dylib      0x1927bf28
fbl::Table::AddRecord_WithOut_Triggers_ex() + 102
11  libvkernel_fat_release.dylib      0x1927ba45
fbl::Table::AddRecord_WithOut_OnEachStatement_Triggers_ex() + 171
12  libvkernel_fat_release.dylib      0x19277e80 fbl::Table::AddRecord() +
428
13  libvkernel_fat_release.dylib      0x1927f5e1
fbl::Table_Indirect::AddRecord() + 217
14  libvkernel_fat_release.dylib      0x19289c29
fbl::vsql::Cursor::AddRecord() + 223
15  V4RB_macho_ub_5b.rbx_0.dylib      0x18ed69d2
Cursor_AddRecord(REALobjectStruct*) + 117

I don¹t understand what¹s happening since I use IF() function in many other
tables without exhibit any problem ... for the moment :-(.

François Van Lerberghe
Rue Thier Monty, 15 A
4570 Marchin
Belgique

le 21/04/12 10:15, Francois Van Lerberghe <fvanlerberghe at freegates.be> a
écrit :

> The function LOCATE() seems to be the culprit here. Replacing it with a
> bunch of IF() solve my problem : no more crash.
> I hope that workaround will not decrease too much the speed...
> 
> Ruslan, is there another workaround more efficient ?

-------------- next part --------------
An HTML attachment was scrubbed...
URL: <http://lists.macserve.net/pipermail/valentina/attachments/20120421/9cd04248/attachment-0001.html>