kdiff3-user — User questions about KDiff3.

Re: [Kdiff3-user] kdiff3 Ver 0.9.95 stack buffer overrun

[Yang T. <yan...@gm...>]

Hi,

2011/5/1 Joachim Eibl wrote:

> http://kdiff3.sourceforge.net/kdiff3.zip
>
> contains the most recent version from svn trunk (which is about to be
> released soon), statically linked against Qt4.7.2.

Program requires libgcc_s_dw2-1.dll, I managed to provide a copy locally.

> For the start you can test if the problem is reproducable.

Yes, the problem is still here, when comparing directories F:/ and G:/

> If yes, then it is probably best just to let it crash while running from gdb
> and send me the stack.

gdb session and crash backtrace follows...

--------8<--------8<--------8<--------8<--------8<--------8<--------8<--------

C:\kdiff3>gdb kdiff3.exe
GNU gdb (GDB) 7.2
Copyright (C) 2010 Free Software Foundation, Inc.
License GPLv3+: GNU GPL version 3 or later <http://gnu.org/licenses/gpl.html>
This is free software: you are free to change and redistribute it.
There is NO WARRANTY, to the extent permitted by law.  Type "show copying"
and "show warranty" for details.
This GDB was configured as "i686-pc-mingw32".
For bug reporting instructions, please see:
<http://www.gnu.org/software/gdb/bugs/>.
Reading symbols from C:\kdiff3/kdiff3.exe...done.
(gdb) run
Starting program: C:\kdiff3/kdiff3.exe
[New Thread 156.0xf5c]
warning: BTMMHOOK 02.05.2011 03:35:46 Thread<0F5C> Hook DLL loaded

[New Thread 156.0xc60]
[New Thread 156.0xe8c]
[New Thread 156.0x7e8]
[New Thread 156.0x968]
[New Thread 156.0x164]
[New Thread 156.0xfd8]
[New Thread 156.0x534]
BFD: C:\WINDOWS\system32\WMVCore.dll: Warning: Ignoring section flag IMAGE_SCN_M
EM_NOT_PAGED in section .reloc
[New Thread 156.0xe6c]
BFD: C:\WINDOWS\system32\WMVCore.dll: Warning: Ignoring section flag IMAGE_SCN_M
EM_NOT_PAGED in section .reloc
warning:

 *** A stack buffer overrun occurred in C:\kdiff3/kdiff3.exe :


warning: This is usually the result of a memory copy to a local buffer or struct
ure where the size is not properly calculated/checked.

warning: If this bug ends up in the shipping product, it could be a severe secur
ity hole.

warning: The stack trace should show the guilty function (the function directly
above __report_gsfailure).

warning:  *** enter .exr 0022E9D0 for the exception record

warning:  *** then kb to get the faulting stack



Program received signal SIGTRAP, Trace/breakpoint trap.
0x7c91120f in ntdll!DbgUiConnectToDbg () from C:\WINDOWS\system32\ntdll.dll
(gdb) bt
#0  0x7c91120f in ntdll!DbgUiConnectToDbg ()
   from C:\WINDOWS\system32\ntdll.dll
#1  0x7c977ede in ntdll!RtlUnhandledExceptionFilter2 ()
   from C:\WINDOWS\system32\ntdll.dll
#2  0x0022ea20 in ?? ()
#3  0x7c9785f1 in ntdll!RtlUnicodeStringToAnsiSize ()
   from C:\WINDOWS\system32\ntdll.dll
#4  0x7c870ef6 in KERNEL32!BeginUpdateResourceA ()
   from C:\WINDOWS\system32\kernel32.dll
#5  0x7c85a6f2 in SetClientTimeZoneInformation ()
   from C:\WINDOWS\system32\kernel32.dll
#6  0x7c85a788 in SetClientTimeZoneInformation ()
   from C:\WINDOWS\system32\kernel32.dll
#7  0x7c85a7bd in SetClientTimeZoneInformation ()
   from C:\WINDOWS\system32\kernel32.dll
#8  0x7c85a834 in SetClientTimeZoneInformation ()
   from C:\WINDOWS\system32\kernel32.dll
#9  0x7c83b11c in ValidateLocale () from C:\WINDOWS\system32\kernel32.dll
#10 0x00b340ca in fileTimeToQDateTime(_FILETIME const*) ()
#11 0x00b3651d in QFSFileEngine::fileTime(QAbstractFileEngine::FileTime) const
    ()
#12 0x00af6639 in QFileInfoPrivate::getFileTime(QAbstractFileEngine::FileTime) c
onst ()
#13 0x00af6799 in QFileInfo::created() const ()
#14 0x00461d44 in FileAccess::setFile (this=0x22f8f8, name=...,
    bWantToWrite=false) at fileaccess.cpp:138
#15 0x00460e0d in FileAccess::FileAccess (this=0x22f8f8, name=...,
    bWantToWrite=false) at fileaccess.cpp:55
#16 0x00405108 in SourceData::setFilename (this=0x2343b80, filename=...)
    at diff.cpp:187
#17 0x0044608d in KDiff3App::slotFileOpen (this=0x2343a08) at pdiff.cpp:1101
#18 0x00418525 in KDiff3App::completeInit (this=0x2343a08, fn1=..., fn2=...,
    fn3=...) at kdiff3.cpp:454
#19 0x0047429b in KDiff3Shell::KDiff3Shell (this=0x2324f90,
    bCompleteInit=true) at kdiff3_shell.cpp:69
#20 0x00404803 in qMain (argc=1, argv=0x21a2430) at main.cpp:249
#21 0x0048a4cf in WinMain@16 ()
#22 0x0048a198 in main ()
(gdb)

--------8<--------8<--------8<--------8<--------8<--------8<--------8<--------

HTH
-- 
-=[Yang]=-

Re: [Kdiff3-user] kdiff3 Ver 0.9.95 stack buffer overrun

[Yang T. <yan...@gm...>]

2011/4/30 Joachim Eibl wrote:

> If you can't build yourself, I could create an executable that could be
> debugged with gdb.

Yes, I can use gdb on an executable with debug info that you provided.
Feel free to attach it directly to me or tell me where to download it
from.

Ah, if you could tell me the name of the function that the 'Ok' button
calls from the window that allows the files or directories selection.
It would save some debugging time, I would set a break point there and
single step from there onwards.

Thanks,
-- 
-=[Yang]=-

Re: [Kdiff3-user] kdiff3 Ver 0.9.95 stack buffer overrun

[Joachim E. <joa...@gm...>]

Hi,

I tried to reproduce, but on my Windows-Vista machine there seems to be no 
problem.

It would help me most, if you were able to debug this yourself.
If you can't build yourself, I could create an executable that could be 
debugged with gdb.
Would this be ok?

Joachim



> Hi,
> 
> - System Windows XP SP3 (Spanish version)
> 
> - KDiff3 version kdiff3 Version 0.9.95 installed using
> KDiff3Setup_0.9.95-2.exe
> 
> When I mount two iso CD images on two virtual CD drives and attempt a
> directory comparison of the root of this two drives, KDiff3 bombs out
> generating an error report that is supposedly intended to be sent to
> MS.
> 
> Application event viewer records the following error:
> 
> Aplicación con errores: kdiff3.exe, versión: 0.0.0.0, módulo con
> error: kernel32.dll, versión 5.1.2600.5781, dirección de error
> 0x0005a6f2.
> 
> Application Failure  kdiff3.exe 0.0.0.0 in kernel32.dll 5.1.2600.5781
> at offset 0005a6f2
> 
> The same comparison running KDiff3 from withing WinDbg reports a stack
> buffer overrun, as the following info shows:
> 
> ---8<------------------8<------------------8<---------------
> 
>  *** A stack buffer overrun occurred in "C:\Archivos de
> programa\KDiff3\kdiff3.exe" :
> 
> This is usually the result of a memory copy to a local buffer or
> structure where the size is not properly calculated/checked.
> If this bug ends up in the shipping product, it could be a severe security
> hole. The stack trace should show the guilty function (the function
> directly above __report_gsfailure).
>  *** enter .exr 0023E8A0 for the exception record
>  *** then kb to get the faulting stack
> 
> (100.350): Break instruction exception - code 80000003 (first chance)
> eax=00000000 ebx=00020600 ecx=7c978568 edx=0023e62b esi=00000001
> edi=0023ebc8 eip=7c91120e esp=0023e844 ebp=0023e880 iopl=0         nv up
> ei pl zr na pe nc cs=001b  ss=0023  ds=0023  es=0023  fs=003b  gs=0000    
>         efl=00000246 ntdll!DbgBreakPoint:
> 7c91120e cc              int     3
> 0:000> .exr 0023E8A0
> *** ERROR: Symbol file could not be found.  Defaulted to export
> symbols for C:\WINDOWS\system32\kernel32.dll -
> ExceptionAddress: 7c85a6f2
> (kernel32!SetClientTimeZoneInformation+0x00000921) ExceptionCode: c0000409
> (Stack buffer overflow)
>   ExceptionFlags: 00000000
> NumberParameters: 0
> 0:000> kb
> ChildEBP RetAddr  Args to Child
> WARNING: Stack unwind information not available. Following frames may be
> wrong. 0023e880 7c9785f1 0023ebc8 7c9785f6 0023ebd0 ntdll!DbgBreakPoint
> 0023e890 7c870ef6 0023ebc8 00000001 c0000409
> ntdll!RtlUnhandledExceptionFilter+0x12
> 0023ebd0 7c85a6f2 00000000 00000024 7c85a6f8
> kernel32!BeginUpdateResourceA+0x13b 0023ee60 7c85a788 0023ee78 0023eed8
> 0000c9a8
> kernel32!SetClientTimeZoneInformation+0x921
> 0023eea4 7c85a7bd 0023eed8 0000c9a8 0023f12c
> kernel32!SetClientTimeZoneInformation+0x9b7
> 0023eebc 7c85a834 0023eed8 0000c9a8 0023f12c
> kernel32!SetClientTimeZoneInformation+0x9ec
> 0023f0e4 7c83b11c 0000c9a8 00000000 0023f12c
> kernel32!SetClientTimeZoneInformation+0xa63
> *** ERROR: Module load completed but symbols could not be loaded for
> C:\Archivos de programa\KDiff3\kdiff3.exe
> 0023f288 00cf6cfc 00000000 0023f410 0023f420 kernel32!ValidateLocale+0x18f4
> 0023f458 00cacbb4 0023f4c0 018acb40 00000001 kdiff3+0x8f6cfc
> 0023f508 00cad3a1 0182b018 00000001 0023f528 kdiff3+0x8acbb4
> 0023f528 004a7493 0023f6b0 0023f650 ffffffff kdiff3+0x8ad3a1
> 0023f6e8 004a8b1e 0023f7b0 0023f8d0 00000000 kdiff3+0xa7493
> 0023f758 0040bbad 0023f7b0 0023f8d0 00000000 kdiff3+0xa8b1e
> 0023f818 00475078 0176cfd4 0023f8d0 0023f8e0 kdiff3+0xbbad
> 0023f9a8 0041c089 0176ce60 00000001 0176cfc0 kdiff3+0x75078
> 0023fb58 004be40e 0176ce60 0023fbd0 0023fbe0 kdiff3+0x1c089
> 0023fc18 00405efb 0173f7f8 00000001 0023fd50 kdiff3+0xbe40e
> 0023fe28 004e5078 00000001 01604f18 01604ed0 kdiff3+0x5efb
> 0023fef8 004e4d18 00400000 00000000 002623a4 kdiff3+0xe5078
> 0023ff78 0040124b 00000001 01603ed8 01602988 kdiff3+0xe4d18
> 
> ---8<------------------8<------------------8<---------------
> 
> More info... It doesn't matter if virtual CDs are Daemon Tools',
> Alcohol 52%'s or Microsoft Virtual CD Control Tool's, the problem
> persists with any of them.
> 
> Mounted ISO images being compared are valid ones, Windows explorer
> sees correctly into mounted drives, and WinMerge works properly.
> 
> Searching for a SetClientTimeZoneInformation function call in sources
> available in kdiff3-0.9.95.tar.gz shows that there is no such call, in
> there at least.
> 
> If you need any further info to help debug/fix this issue, or you want
> me to run whatever test, please don't hesitate to ask it.
> 
> Cheers,

Re: [Kdiff3-user] kdiff3 Ver 0.9.95 stack buffer overrun

[Valentin R. <kd...@ru...>]

Hello,

Could you please file a report on bugs.kde.org, please?

Thanks,
Valentin

On 04/29/2011 06:25 PM, Yang Tse wrote:
> Hi,
>
> - System Windows XP SP3 (Spanish version)
>
> - KDiff3 version kdiff3 Version 0.9.95 installed using KDiff3Setup_0.9.95-2.exe
>
> When I mount two iso CD images on two virtual CD drives and attempt a
> directory comparison of the root of this two drives, KDiff3 bombs out
> generating an error report that is supposedly intended to be sent to
> MS.
>
> Application event viewer records the following error:
>
> Aplicación con errores: kdiff3.exe, versión: 0.0.0.0, módulo con
> error: kernel32.dll, versión 5.1.2600.5781, dirección de error
> 0x0005a6f2.
>
> Application Failure  kdiff3.exe 0.0.0.0 in kernel32.dll 5.1.2600.5781
> at offset 0005a6f2
>
> The same comparison running KDiff3 from withing WinDbg reports a stack
> buffer overrun, as the following info shows:
>
> ---8<------------------8<------------------8<---------------
>
>  *** A stack buffer overrun occurred in "C:\Archivos de
> programa\KDiff3\kdiff3.exe" :
>
> This is usually the result of a memory copy to a local buffer or
> structure where the size is not properly calculated/checked.
> If this bug ends up in the shipping product, it could be a severe security hole.
> The stack trace should show the guilty function (the function directly
> above __report_gsfailure).
>  *** enter .exr 0023E8A0 for the exception record
>  *** then kb to get the faulting stack
>
> (100.350): Break instruction exception - code 80000003 (first chance)
> eax=00000000 ebx=00020600 ecx=7c978568 edx=0023e62b esi=00000001 edi=0023ebc8
> eip=7c91120e esp=0023e844 ebp=0023e880 iopl=0         nv up ei pl zr na pe nc
> cs=001b  ss=0023  ds=0023  es=0023  fs=003b  gs=0000             efl=00000246
> ntdll!DbgBreakPoint:
> 7c91120e cc              int     3
> 0:000> .exr 0023E8A0
> *** ERROR: Symbol file could not be found.  Defaulted to export
> symbols for C:\WINDOWS\system32\kernel32.dll -
> ExceptionAddress: 7c85a6f2 (kernel32!SetClientTimeZoneInformation+0x00000921)
>    ExceptionCode: c0000409 (Stack buffer overflow)
>   ExceptionFlags: 00000000
> NumberParameters: 0
> 0:000> kb
> ChildEBP RetAddr  Args to Child
> WARNING: Stack unwind information not available. Following frames may be wrong.
> 0023e880 7c9785f1 0023ebc8 7c9785f6 0023ebd0 ntdll!DbgBreakPoint
> 0023e890 7c870ef6 0023ebc8 00000001 c0000409
> ntdll!RtlUnhandledExceptionFilter+0x12
> 0023ebd0 7c85a6f2 00000000 00000024 7c85a6f8 kernel32!BeginUpdateResourceA+0x13b
> 0023ee60 7c85a788 0023ee78 0023eed8 0000c9a8
> kernel32!SetClientTimeZoneInformation+0x921
> 0023eea4 7c85a7bd 0023eed8 0000c9a8 0023f12c
> kernel32!SetClientTimeZoneInformation+0x9b7
> 0023eebc 7c85a834 0023eed8 0000c9a8 0023f12c
> kernel32!SetClientTimeZoneInformation+0x9ec
> 0023f0e4 7c83b11c 0000c9a8 00000000 0023f12c
> kernel32!SetClientTimeZoneInformation+0xa63
> *** ERROR: Module load completed but symbols could not be loaded for
> C:\Archivos de programa\KDiff3\kdiff3.exe
> 0023f288 00cf6cfc 00000000 0023f410 0023f420 kernel32!ValidateLocale+0x18f4
> 0023f458 00cacbb4 0023f4c0 018acb40 00000001 kdiff3+0x8f6cfc
> 0023f508 00cad3a1 0182b018 00000001 0023f528 kdiff3+0x8acbb4
> 0023f528 004a7493 0023f6b0 0023f650 ffffffff kdiff3+0x8ad3a1
> 0023f6e8 004a8b1e 0023f7b0 0023f8d0 00000000 kdiff3+0xa7493
> 0023f758 0040bbad 0023f7b0 0023f8d0 00000000 kdiff3+0xa8b1e
> 0023f818 00475078 0176cfd4 0023f8d0 0023f8e0 kdiff3+0xbbad
> 0023f9a8 0041c089 0176ce60 00000001 0176cfc0 kdiff3+0x75078
> 0023fb58 004be40e 0176ce60 0023fbd0 0023fbe0 kdiff3+0x1c089
> 0023fc18 00405efb 0173f7f8 00000001 0023fd50 kdiff3+0xbe40e
> 0023fe28 004e5078 00000001 01604f18 01604ed0 kdiff3+0x5efb
> 0023fef8 004e4d18 00400000 00000000 002623a4 kdiff3+0xe5078
> 0023ff78 0040124b 00000001 01603ed8 01602988 kdiff3+0xe4d18
>
> ---8<------------------8<------------------8<---------------
>
> More info... It doesn't matter if virtual CDs are Daemon Tools',
> Alcohol 52%'s or Microsoft Virtual CD Control Tool's, the problem
> persists with any of them.
>
> Mounted ISO images being compared are valid ones, Windows explorer
> sees correctly into mounted drives, and WinMerge works properly.
>
> Searching for a SetClientTimeZoneInformation function call in sources
> available in kdiff3-0.9.95.tar.gz shows that there is no such call, in
> there at least.
>
> If you need any further info to help debug/fix this issue, or you want
> me to run whatever test, please don't hesitate to ask it.
>
> Cheers,

[Kdiff3-user] kdiff3 Ver 0.9.95 stack buffer overrun

[Yang T. <yan...@gm...>]

Hi,

- System Windows XP SP3 (Spanish version)

- KDiff3 version kdiff3 Version 0.9.95 installed using KDiff3Setup_0.9.95-2.exe

When I mount two iso CD images on two virtual CD drives and attempt a
directory comparison of the root of this two drives, KDiff3 bombs out
generating an error report that is supposedly intended to be sent to
MS.

Application event viewer records the following error:

Aplicación con errores: kdiff3.exe, versión: 0.0.0.0, módulo con
error: kernel32.dll, versión 5.1.2600.5781, dirección de error
0x0005a6f2.

Application Failure  kdiff3.exe 0.0.0.0 in kernel32.dll 5.1.2600.5781
at offset 0005a6f2

The same comparison running KDiff3 from withing WinDbg reports a stack
buffer overrun, as the following info shows:

---8<------------------8<------------------8<---------------

 *** A stack buffer overrun occurred in "C:\Archivos de
programa\KDiff3\kdiff3.exe" :

This is usually the result of a memory copy to a local buffer or
structure where the size is not properly calculated/checked.
If this bug ends up in the shipping product, it could be a severe security hole.
The stack trace should show the guilty function (the function directly
above __report_gsfailure).
 *** enter .exr 0023E8A0 for the exception record
 *** then kb to get the faulting stack

(100.350): Break instruction exception - code 80000003 (first chance)
eax=00000000 ebx=00020600 ecx=7c978568 edx=0023e62b esi=00000001 edi=0023ebc8
eip=7c91120e esp=0023e844 ebp=0023e880 iopl=0         nv up ei pl zr na pe nc
cs=001b  ss=0023  ds=0023  es=0023  fs=003b  gs=0000             efl=00000246
ntdll!DbgBreakPoint:
7c91120e cc              int     3
0:000> .exr 0023E8A0
*** ERROR: Symbol file could not be found.  Defaulted to export
symbols for C:\WINDOWS\system32\kernel32.dll -
ExceptionAddress: 7c85a6f2 (kernel32!SetClientTimeZoneInformation+0x00000921)
   ExceptionCode: c0000409 (Stack buffer overflow)
  ExceptionFlags: 00000000
NumberParameters: 0
0:000> kb
ChildEBP RetAddr  Args to Child
WARNING: Stack unwind information not available. Following frames may be wrong.
0023e880 7c9785f1 0023ebc8 7c9785f6 0023ebd0 ntdll!DbgBreakPoint
0023e890 7c870ef6 0023ebc8 00000001 c0000409
ntdll!RtlUnhandledExceptionFilter+0x12
0023ebd0 7c85a6f2 00000000 00000024 7c85a6f8 kernel32!BeginUpdateResourceA+0x13b
0023ee60 7c85a788 0023ee78 0023eed8 0000c9a8
kernel32!SetClientTimeZoneInformation+0x921
0023eea4 7c85a7bd 0023eed8 0000c9a8 0023f12c
kernel32!SetClientTimeZoneInformation+0x9b7
0023eebc 7c85a834 0023eed8 0000c9a8 0023f12c
kernel32!SetClientTimeZoneInformation+0x9ec
0023f0e4 7c83b11c 0000c9a8 00000000 0023f12c
kernel32!SetClientTimeZoneInformation+0xa63
*** ERROR: Module load completed but symbols could not be loaded for
C:\Archivos de programa\KDiff3\kdiff3.exe
0023f288 00cf6cfc 00000000 0023f410 0023f420 kernel32!ValidateLocale+0x18f4
0023f458 00cacbb4 0023f4c0 018acb40 00000001 kdiff3+0x8f6cfc
0023f508 00cad3a1 0182b018 00000001 0023f528 kdiff3+0x8acbb4
0023f528 004a7493 0023f6b0 0023f650 ffffffff kdiff3+0x8ad3a1
0023f6e8 004a8b1e 0023f7b0 0023f8d0 00000000 kdiff3+0xa7493
0023f758 0040bbad 0023f7b0 0023f8d0 00000000 kdiff3+0xa8b1e
0023f818 00475078 0176cfd4 0023f8d0 0023f8e0 kdiff3+0xbbad
0023f9a8 0041c089 0176ce60 00000001 0176cfc0 kdiff3+0x75078
0023fb58 004be40e 0176ce60 0023fbd0 0023fbe0 kdiff3+0x1c089
0023fc18 00405efb 0173f7f8 00000001 0023fd50 kdiff3+0xbe40e
0023fe28 004e5078 00000001 01604f18 01604ed0 kdiff3+0x5efb
0023fef8 004e4d18 00400000 00000000 002623a4 kdiff3+0xe5078
0023ff78 0040124b 00000001 01603ed8 01602988 kdiff3+0xe4d18

---8<------------------8<------------------8<---------------

More info... It doesn't matter if virtual CDs are Daemon Tools',
Alcohol 52%'s or Microsoft Virtual CD Control Tool's, the problem
persists with any of them.

Mounted ISO images being compared are valid ones, Windows explorer
sees correctly into mounted drives, and WinMerge works properly.

Searching for a SetClientTimeZoneInformation function call in sources
available in kdiff3-0.9.95.tar.gz shows that there is no such call, in
there at least.

If you need any further info to help debug/fix this issue, or you want
me to run whatever test, please don't hesitate to ask it.

Cheers,
-- 
-=[Yang]=-

Re: [Kdiff3-user] Decoding a cp1251 file saved as utf8?

[Angel T. <fn...@fm...>]

On 02/16/11 21:54, Joachim Eibl wrote:
> Hi Angel,
>
> No good news: When looking at the output in hex
> (e.g. via  od -t x cp1251.saved.as.utf8.txt )
> you see that there is no useful information left anymore.
>
> 0000000 efbdbfef bfefbdbf bdbfefbd efbdbfef
> 0000020 bfefbdbf bdbfefbd efbdbfef bfefbdbf
> 0000040 bdbfefbd efbdbfef bfefbdbf bdbfefbd
> 0000060 efbdbfef bfefbdbf bdbfefbd efbdbfef
> 0000100 bfefbdbf bdbfefbd efbdbfef bfefbdbf
> 0000120 bdbfefbd efbdbfef 000abdbf
>
> So in your concrete situation I can't do much for you.
>
> But I must admit, that I was not aware of that problem. As I mentioned before
> I expected no irreversible conversion loss, but now I think that Qt internally
> converts to 16 bit  although UTF8 allows 32 bit characters. So most random
> combinations will result in a "invalid" character.
>
> I will try to detect this and display a warning in KDiff3 for such situations.
>
> Thanks for telling! I really do hope you find some backup.

Hello Joachim,

And don't worry.  The file I lost in this incident was not too long and 
I was able to remember its contents.

Despite this little problem, I am still a happy user of KDiff3. I also 
think that the suggested warning (or smth similar) is highly desirable 
in this situation.

Best regards,
Angel Tsankov

Re: [Kdiff3-user] Decoding a cp1251 file saved as utf8?

[Joachim E. <joa...@gm...>]

Hi Angel,

No good news: When looking at the output in hex 
(e.g. via  od -t x cp1251.saved.as.utf8.txt )
you see that there is no useful information left anymore.

0000000 efbdbfef bfefbdbf bdbfefbd efbdbfef
0000020 bfefbdbf bdbfefbd efbdbfef bfefbdbf
0000040 bdbfefbd efbdbfef bfefbdbf bdbfefbd
0000060 efbdbfef bfefbdbf bdbfefbd efbdbfef
0000100 bfefbdbf bdbfefbd efbdbfef bfefbdbf
0000120 bdbfefbd efbdbfef 000abdbf

So in your concrete situation I can't do much for you.

But I must admit, that I was not aware of that problem. As I mentioned before 
I expected no irreversible conversion loss, but now I think that Qt internally 
converts to 16 bit  although UTF8 allows 32 bit characters. So most random 
combinations will result in a "invalid" character.

I will try to detect this and display a warning in KDiff3 for such situations.

Thanks for telling! I really do hope you find some backup.
Joachim




> On 02/15/11 22:41, Joachim Eibl wrote:
> > Hi Angel,
> > 
> > I'm not quite sure that I understand the problem correctly.
> > 
> > When reading as UTF8 and writing the same data as UTF8 then I would not
> > expect many changes, because except for a few places everything should
> > stay the same, regardless of what codec is really used as input.
> 
> I forgot to mention that the original file contains cyrillic characters.
> 
> [...]
> 
> > Could you repeat this with a test file and send the original and modified
> > versions?
> 
> See attached archive.  It contains 2 files: the first one lists the
> lowercase letters (30 in total) of the Bulgarian alphabet plus a LF
> character and the second one was generated by merging the first one with
> a copy of itself (both opened as UTF8 files) and saving the output as
> UTF8.  A binary editor shows that the output file contains the same
> character duplicated 30 times followed by a LF character.
> 
> 
> Regards,
> Angel Tsankov

Re: [Kdiff3-user] Decoding a cp1251 file saved as utf8?

[Angel T. <fn...@fm...>]

On 02/15/11 22:41, Joachim Eibl wrote:
> Hi Angel,
>
> I'm not quite sure that I understand the problem correctly.
>
> When reading as UTF8 and writing the same data as UTF8 then I would not expect
> many changes, because except for a few places everything should stay the same,
> regardless of what codec is really used as input.

I forgot to mention that the original file contains cyrillic characters.

[...]

> Could you repeat this with a test file and send the original and modified
> versions?

See attached archive.  It contains 2 files: the first one lists the 
lowercase letters (30 in total) of the Bulgarian alphabet plus a LF 
character and the second one was generated by merging the first one with 
a copy of itself (both opened as UTF8 files) and saving the output as 
UTF8.  A binary editor shows that the output file contains the same 
character duplicated 30 times followed by a LF character.


Regards,
Angel Tsankov

Re: [Kdiff3-user] Decoding a cp1251 file saved as utf8?

[Joachim E. <joa...@gm...>]

Hi Angel,

I'm not quite sure that I understand the problem correctly.

When reading as UTF8 and writing the same data as UTF8 then I would not expect 
many changes, because except for a few places everything should stay the same, 
regardless of what codec is really used as input. 

Yet you wouldn't write if there were no problems.

Could you repeat this with a test file and send the original and modified 
versions?

Cheers,
Joachim





> Hello.
> 
> I'm using KDiff3 0.9.92 (with KDE 3.5.10 on Debian Lenny) with encoding
> set to UTF-8 for everything (A, B, C, merge output and saving, etc.) and
> "Auto Detect Unicode" checked.  With these settings I accidentally
> merged two CP-1251-encoded files and saved the result.  Now I cannot
> figure out how to "decode" the result properly and, unfortunately, the
> original files are gone.  Does anyone have an idea on decoding the result?
> 
> Thanks in advance,
> Angel Tsankov
> 
> ---------------------------------------------------------------------------
> --- The ultimate all-in-one performance toolkit: Intel(R) Parallel Studio
> XE: Pinpoint memory and threading errors before they happen.
> Find and fix more than 250 security defects in the development cycle.
> Locate bottlenecks in serial and parallel code that limit performance.
> http://p.sf.net/sfu/intel-dev2devfeb
> _______________________________________________
> Kdiff3-user mailing list
> Kdi...@li...
> https://lists.sourceforge.net/lists/listinfo/kdiff3-user

[Kdiff3-user] Decoding a cp1251 file saved as utf8?

[Angel T. <fn...@fm...>]

Hello.

I'm using KDiff3 0.9.92 (with KDE 3.5.10 on Debian Lenny) with encoding 
set to UTF-8 for everything (A, B, C, merge output and saving, etc.) and 
"Auto Detect Unicode" checked.  With these settings I accidentally 
merged two CP-1251-encoded files and saved the result.  Now I cannot 
figure out how to "decode" the result properly and, unfortunately, the 
original files are gone.  Does anyone have an idea on decoding the result?

Thanks in advance,
Angel Tsankov

Re: [Kdiff3-user] crash

[Valentin R. <kd...@ru...>]

On 02/09/2011 09:00 PM, Joachim Eibl wrote:
> Hi,
>
> I can't reproduce this. (But I don't use Fedora.)
It won't reproduce for me neither (I'm using OpenSuSE 11.4).
Cheers,
Valentin

Re: [Kdiff3-user] Questions/Requests

[Joachim E. <joa...@gm...>]

Hi Tom,

Sorry for the very delayed response.

"Stats per file" are there to a certain extent when activating "Full Analysis" 
in the Directory settings.

Keybindings are only adjustable in the KDE version. There it was provided by 
the system. For other platforms more work is needed.

Session saving, directory view output/report are often asked for. Because the 
directory view is quite independant of the rest, this might be a job for a 
volunteer. "Hiding" also affects only this part.

Direct editing, Undo, line wrapping: I have these on my list too. These and 
some other wishes are related and I've started work on a improved 
viewer/editor architecture but am not able to do much lately.

Context menus/Configurable Alerts/Shelling out: Probably not very difficult. 
(Any volunteers?)

I foresee that I will have much less time for development in the future.
So if anyone wants to help, you might want to join the team.

Cheers,
Joachim



> Various things I wish I knew how to do (or features I miss):
> 
> - Hiding - is there a way to manually exlude files or directories?
>   I mean, from the directory browser, not by configure-anti-pattern?
>   It is common for me to want to say "this and this and this don't
>   matter just ignore them."  This could just be "right click on a
>   file or directory, context menu gives option to hide, which then
>   adds anti-pattern on the fly..."
>
> - Session saving - especially important if there is a way to hide.
>   I'd like to be in the middle of a merging session with certain
>   directories hidden and visible "done" marks - and save the session
>   for later, to be reloaded in a few days.
>
> - Directory view output / report - Is there a way to capture the
>   diretory difference list, any kind of "output" mode?
>
> - Stats per file, i.e., put the number of differences next to each
>   file name in the directory pane - it would be useful to see that
>   file A has 3 differences for 30 lines, file B has 19 differences
>   for 2,342 lines...
> 
> - Keybindings - are they customizable?  F6 and CTRL-1 are _NOT_ the
>   ones I would pick, by the way.  And I really _don't_ want "a" or
>   "b" to jump to the file starting with that name...
> 
> - Direct editing - Is there a way to edit without merging?  Often,
>   I'll double-click a file, and then I wish I could edit either or
>   both sides manually... but those panes are read only...
> 
> - Context-menus - I would love it if right-click menus were supported.
> 
> - Undo - well - it would be nice to save history and allow undo...
> 
> - Alerts - is it possible to configure whether some pop-ups pop-up?
> 
> - Shelling out - I'd like to be able to open gvim or filter through
>   something else or shell into an IDE or whatever ... some files are
>   just not that usable in the built-in editor.
> 
> - Line wrapping - 500 character lines are a problem.  Could there be
>   an option for it wrap the left and right panes in sync?
> 
> -Thanks,,, -Tom

Re: [Kdiff3-user] crash

[Joachim E. <joa...@gm...>]

Hi,

I can't reproduce this. (But I don't use Fedora.)
Would you please be so kind to debug (and send me the stack / more info.)
Or at least send the files. Does this happen reproducably with any files?

Best regards,
Joachim


> --- Comment #2 from Diego <pa...@in...> 2011-02-08 04:25:34 EST ---
> Package: kdiff3-0.9.95-5.fc14
> Architecture: x86_64
> OS Release: Fedora release 14 (Laughlin)
> 
> 
> How to reproduce
> -----
> 1. Compare 3 files
> 2. Open "Diffview" menu and select "A vs. B Overview"
> 3. crash
> 
> https://bugzilla.redhat.com/show_bug.cgi?id=659736
> 
> ---------------------------------------------------------------------------
> --- The ultimate all-in-one performance toolkit: Intel(R) Parallel Studio
> XE: Pinpoint memory and threading errors before they happen.
> Find and fix more than 250 security defects in the development cycle.
> Locate bottlenecks in serial and parallel code that limit performance.
> http://p.sf.net/sfu/intel-dev2devfeb
> _______________________________________________
> Kdiff3-user mailing list
> Kdi...@li...
> https://lists.sourceforge.net/lists/listinfo/kdiff3-user

[Kdiff3-user] crash

[Neal B. <ndb...@gm...>]

--- Comment #2 from Diego <pa...@in...> 2011-02-08 04:25:34 EST ---
Package: kdiff3-0.9.95-5.fc14
Architecture: x86_64
OS Release: Fedora release 14 (Laughlin)


How to reproduce
-----
1. Compare 3 files
2. Open "Diffview" menu and select "A vs. B Overview"
3. crash

https://bugzilla.redhat.com/show_bug.cgi?id=659736

[Kdiff3-user] Questions/Requests

[Tom {T. O. <to...@to...>]

Various things I wish I knew how to do (or features I miss):

- Hiding - is there a way to manually exlude files or directories?
  I mean, from the directory browser, not by configure-anti-pattern?
  It is common for me to want to say "this and this and this don't
  matter just ignore them."  This could just be "right click on a
  file or directory, context menu gives option to hide, which then
  adds anti-pattern on the fly..."

- Session saving - especially important if there is a way to hide.
  I'd like to be in the middle of a merging session with certain
  directories hidden and visible "done" marks - and save the session
  for later, to be reloaded in a few days.

- Directory view output / report - Is there a way to capture the
  diretory difference list, any kind of "output" mode?

- Stats per file, i.e., put the number of differences next to each
  file name in the directory pane - it would be useful to see that
  file A has 3 differences for 30 lines, file B has 19 differences
  for 2,342 lines...

- Keybindings - are they customizable?  F6 and CTRL-1 are _NOT_ the
  ones I would pick, by the way.  And I really _don't_ want "a" or
  "b" to jump to the file starting with that name...

- Direct editing - Is there a way to edit without merging?  Often,
  I'll double-click a file, and then I wish I could edit either or
  both sides manually... but those panes are read only...

- Context-menus - I would love it if right-click menus were supported.

- Undo - well - it would be nice to save history and allow undo...

- Alerts - is it possible to configure whether some pop-ups pop-up?

- Shelling out - I'd like to be able to open gvim or filter through
  something else or shell into an IDE or whatever ... some files are
  just not that usable in the built-in editor.

- Line wrapping - 500 character lines are a problem.  Could there be
  an option for it wrap the left and right panes in sync?

-Thanks,,, -Tom

-- 
"Let us do our duty in our shop or our kitchen, the market, the street, the
office, the school, the home, just as faithfully as if we stood in the front
rank of some great battle, and we knew that victory for mankind depended upon
our bravery, strength, and skill.  When we do that the humblest of us will be
serving in that great army which achieves the welfare of the world."
 --Theodore Parker
The little girl expects no declaration of tenderness from her doll.
She loves it -- and that's all.  It is thus that we should love.
 -- DeGourmont

Re: [Kdiff3-user] (no subject)

[owzar001 <owz...@du...>]

Hi Joachim.

Thanks for your response. I did not realize that I could edit the output
window from within kdiff3.

In response to my question, a colleague of mine had suggested that I
use MS track change instead. I think he meant to be funny.

Thanks for contributing kdiff3.

Take care, Kouros

On Sat, 31 Jul 2010, Joachim Eibl wrote:

> Hi Kouros,
>
> If there are multiple changes in a line and you want to select them
> individually, then you have to edit that line in the merge output window. For
> the moment there is no special feature in the GUI allowing a more direct
> approach. I admit, that is is not ideal if the lines are very long.
>
> Best regards,
> Joachim
>
>
>> Hi.
>>
>> I am relatively new user of kdiff3. I have found to be very useful
>> in conjunction with mercurial.
>>
>> I would like to pose a potentially trivial question on merging
>> multiple changes within a single line using the following toy example.
>>
>> Suppose that an author provides the following document for
>> review
>> foo@kolmogorov:/tmp$ cat author/foo.tex
>> This is a simple randomm sample.
>>
>> A co-author proceeds by editing the file by correcting the typo and
>> by deleting the word "simple"
>>
>> foo@kolmogorov:/tmp$ cat coauthor/foo.tex
>> This is a random sample.
>>
>> It is possible fo author to use kdiff3 to accept the typo but
>> not the deletion of "simple" to get
>>
>> foo@kolmogorov:/tmp$ cat author/foo.tex
>> This is a simple random sample.
>>
>> I am using kdiff3 0.9.92 on Debian Lenny AMD64.
>>
>> Sincerely,
>>
>> Kouros
>>
>> foo@kolmogorov:/tmp$ apt-cache policy kdiff3
>> kdiff3:
>>    Installed: 0.9.92-2
>>    Candidate: 0.9.92-2
>>    Version table:
>>   *** 0.9.92-2 0
>>          650 http://ftp.us.debian.org lenny/main Packages
>>          100 /var/lib/dpkg/status
>>
>>
>> ---------------------------------------------------------------------------
>> --- The Palm PDK Hot Apps Program offers developers who use the
>> Plug-In Development Kit to bring their C/C++ apps to Palm for a share
>> of $1 Million in cash or HP Products. Visit us here for more details:
>> http://ad.doubleclick.net/clk;226879339;13503038;l?
>> http://clk.atdmt.com/CRS/go/247765532/direct/01/
>> _______________________________________________
>> Kdiff3-user mailing list
>> Kdi...@li...
>> https://lists.sourceforge.net/lists/listinfo/kdiff3-user
>>
>

Re: [Kdiff3-user] (no subject)

[Joachim E. <joa...@gm...>]

Hi Kouros,

If there are multiple changes in a line and you want to select them 
individually, then you have to edit that line in the merge output window. For 
the moment there is no special feature in the GUI allowing a more direct 
approach. I admit, that is is not ideal if the lines are very long.

Best regards,
Joachim


> Hi.
> 
> I am relatively new user of kdiff3. I have found to be very useful
> in conjunction with mercurial.
> 
> I would like to pose a potentially trivial question on merging
> multiple changes within a single line using the following toy example.
> 
> Suppose that an author provides the following document for
> review
> foo@kolmogorov:/tmp$ cat author/foo.tex
> This is a simple randomm sample.
> 
> A co-author proceeds by editing the file by correcting the typo and
> by deleting the word "simple"
> 
> foo@kolmogorov:/tmp$ cat coauthor/foo.tex
> This is a random sample.
> 
> It is possible fo author to use kdiff3 to accept the typo but
> not the deletion of "simple" to get
> 
> foo@kolmogorov:/tmp$ cat author/foo.tex
> This is a simple random sample.
> 
> I am using kdiff3 0.9.92 on Debian Lenny AMD64.
> 
> Sincerely,
> 
> Kouros
> 
> foo@kolmogorov:/tmp$ apt-cache policy kdiff3
> kdiff3:
>    Installed: 0.9.92-2
>    Candidate: 0.9.92-2
>    Version table:
>   *** 0.9.92-2 0
>          650 http://ftp.us.debian.org lenny/main Packages
>          100 /var/lib/dpkg/status
> 
> 
> ---------------------------------------------------------------------------
> --- The Palm PDK Hot Apps Program offers developers who use the
> Plug-In Development Kit to bring their C/C++ apps to Palm for a share
> of $1 Million in cash or HP Products. Visit us here for more details:
> http://ad.doubleclick.net/clk;226879339;13503038;l?
> http://clk.atdmt.com/CRS/go/247765532/direct/01/
> _______________________________________________
> Kdiff3-user mailing list
> Kdi...@li...
> https://lists.sourceforge.net/lists/listinfo/kdiff3-user
> 

[Kdiff3-user] (no subject)

[owzar001 <owz...@du...>]

Hi.

I am relatively new user of kdiff3. I have found to be very useful
in conjunction with mercurial.

I would like to pose a potentially trivial question on merging
multiple changes within a single line using the following toy example.

Suppose that an author provides the following document for
review 
foo@kolmogorov:/tmp$ cat author/foo.tex
This is a simple randomm sample.

A co-author proceeds by editing the file by correcting the typo and
by deleting the word "simple"

foo@kolmogorov:/tmp$ cat coauthor/foo.tex
This is a random sample.

It is possible fo author to use kdiff3 to accept the typo but
not the deletion of "simple" to get

foo@kolmogorov:/tmp$ cat author/foo.tex
This is a simple random sample.

I am using kdiff3 0.9.92 on Debian Lenny AMD64.

Sincerely,

Kouros

foo@kolmogorov:/tmp$ apt-cache policy kdiff3
kdiff3:
   Installed: 0.9.92-2
   Candidate: 0.9.92-2
   Version table:
  *** 0.9.92-2 0
         650 http://ftp.us.debian.org lenny/main Packages
         100 /var/lib/dpkg/status

[Kdiff3-user] Fwd: Docbook issues

[Rohan G. <roh...@ub...>]

---------- Forwarded message ----------
From: Rohan Garg <roh...@ub...>
Date: Thu, Jul 15, 2010 at 1:56 AM
Subject: Docbook issues
To: kdi...@li...


Hi
With the 0.9.95 release of kdiff3,the documentation which was released seems
to be outdated and causes FTBFS error on launchpad servers.
Build error :
http://people.ubuntuwire.org/~lucas/ubuntu-nbs/64/kdiff3_0.9.95-6_lubuntu64.buildlog
It would be highly appreciated if there were a new release with the proper
docbook.

Regards
Rohan Garg

[Kdiff3-user] [patches!] docbook fixes

[Neal B. <ndb...@gm...>]

Hi.  I maintain kdiff3 for Fedora.  Can you please add these patches for 
docbook?  

http://cvs.fedoraproject.org/viewvc/devel/kdiff3/kdiff3-0.9.95docbook_fixes.patch

Re: [Kdiff3-user] [arey] dir merge - overwrite new with old

[Joachim E. <joa...@gm...>]

Hi David,

The best trick is to modify directorymergewindow.cpp, line 1409 from 
originally:
               if ( mfi.m_ageA == eNew )
to
               if ( mfi.m_ageA != eNew )

Then recompile. This just inverts the behaviour for choosing the proposed 
merge operation, but otherwise you still choose "Copy newer instead of 
merging".

Implementing this as a feature wouldn't be much work, but currently I have no 
time to make a new release.

Cheers,
Joachim



> hi Joachim,
> 
> On 7/6/2010 12:10 PM, Joachim Eibl wrote:
> > David M. Besonen wrote:
> >> how can i configure kdiff3 to merge two directories such
> >> that when duplicate file names are found the oldest file is
> >> the one that is retained?
> >>
> >> assume that dir A: is the location to which the older files
> >> from directory B: will be copied.
> >
> > This is currently not supported.
> 
> might there be a way in which i could "trick" kdiff3 into
> executing this behavior?
> 
> > I never had a requirement for this, and also don't understand
> > in which kind of situation this might be needed. If you think
> > this is a feature others might also need, please provide some
> > more information.
> 
> the situation is one in which many files have been duplicated and
> modified and located in a directory different than the original
> files.  this other directory was already populated with a
> different set of files.  i need to merge these two directories
> and i don't want the duplicated, modified files to overwrite the
> older files in the first directory with the same file names.
> 
> how much work would it be to add this functionality to kdiff3?
> 
> 
> thanks,
> david
> 
> 
> 
> ---------------------------------------------------------------------------
> --- This SF.net email is sponsored by Sprint
> What will you do first with EVO, the first 4G phone?
> Visit sprint.com/first -- http://p.sf.net/sfu/sprint-com-first
> _______________________________________________
> Kdiff3-user mailing list
> Kdi...@li...
> https://lists.sourceforge.net/lists/listinfo/kdiff3-user
> 

Re: [Kdiff3-user] [arey] dir merge - overwrite new with old

[David M. B. <da...@pa...>]

hi Joachim,

On 7/6/2010 12:10 PM, Joachim Eibl wrote:

> David M. Besonen wrote:
>
>> how can i configure kdiff3 to merge two directories such
>> that when duplicate file names are found the oldest file is
>> the one that is retained?
>> 
>> assume that dir A: is the location to which the older files
>> from directory B: will be copied.
>
> This is currently not supported.

might there be a way in which i could "trick" kdiff3 into
executing this behavior?


> I never had a requirement for this, and also don't understand
> in which kind of situation this might be needed. If you think
> this is a feature others might also need, please provide some
> more information.

the situation is one in which many files have been duplicated and
modified and located in a directory different than the original
files.  this other directory was already populated with a
different set of files.  i need to merge these two directories
and i don't want the duplicated, modified files to overwrite the
older files in the first directory with the same file names.

how much work would it be to add this functionality to kdiff3?


thanks,
david

Re: [Kdiff3-user] [arey] dir merge - overwrite new with old

[Joachim E. <joa...@gm...>]

Hi David,

This is currently not supported. I never had a requirement for this, and also 
don't understand in which kind of situation this might be needed.
If you think this is a feature others might also need, please provide some 
more information.

Joachim


> hello,
> 
> how can i configure kdiff3 to merge two directories such that
> when duplicate file names are found the oldest file is the one
> that is retained?
> 
> assume that dir A: is the location to which the older files from
> directory B: will be copied.
> 
> 
> 
> thanks,
> david
> 
> 
> ---------------------------------------------------------------------------
> --- This SF.net email is sponsored by Sprint
> What will you do first with EVO, the first 4G phone?
> Visit sprint.com/first -- http://p.sf.net/sfu/sprint-com-first
> _______________________________________________
> Kdiff3-user mailing list
> Kdi...@li...
> https://lists.sourceforge.net/lists/listinfo/kdiff3-user
> 

[Kdiff3-user] [arey] dir merge - overwrite new with old

[David M. B. <da...@pa...>]

hello,

how can i configure kdiff3 to merge two directories such that
when duplicate file names are found the oldest file is the one
that is retained?

assume that dir A: is the location to which the older files from
directory B: will be copied.



thanks,
david

[Kdiff3-user] ShowIdenticalFiles=0

[Fred L. H. I. <Fre...@ho...>]

My company builds some of our applications based on an in-house 
library.  These materials go through versions/updates.  Those of us who 
use laptops/portables commonly pull a local copy from the server so that 
we can work disconnected.  I built a simple Windows *.cmd file to 
automate this process (and to distinguish between creating a new 
directory and updating/refreshing an existing one).

While trying to use KDiff3.exe (the latest version - as of this date) 
within this *.cmd file, I appear to be unable to get the 
ShowIdenticalFiles= configuration entry to work (or I don't understand 
what it's expected to do).

I placed the following command in the *.cmd file:

"[path to KDiff3]\kdiff3" [server directory] [local directory] --cs 
"FilePattern=*.h;*.hpp;*.c;*.cpp;*.xsd;*.xml;" --cs "ShowIdenticalFiles=0"

The comparison takes place, but/ all/ of the *.h, *.hpp, *.c, *.cpp, 
*.xsd, and *.xml files are displayed, even when the content(s) of the 
local directory are identical in all respects - except for location - to 
the content(s) of the server directory.

I have also checked the --confighelp output and it displays 
"ShowIdenticalFiles=0" as the current persistent setting.

If I deselect the <ShowIdenticalFiles> button on the menu bar of the 
KDiff3 GUI, all of the identical files immediately disappear.  BTW, the 
<ShowIdenticalFiles> button on the menu bar of the KDiff3 GUI always 
shows *active* when I load KDiff3, even though the "persistent" setting 
shows '0'  (which I thought should indicate* inactive* or* false*).

Is there something about this that I do not understand?

FYI, my system is Windows 7 Pro on a DELL Vostro 1520.

Thanks in advance for any help/suggestions,
Fred L. Hoff II