{12} All tickets recently changed first (101 matches)

SeaMonkey? crashed while displaying Flash content and produced the accompanying trap file.

Using otrdecoder.exe (​http://www.onlinetvrecorder.com/downloads/otrdecoder-bin-i586-pc-mingw32msvc-0.4.1132.tar.bz2) for decoding large TV-rec-files. If the files are smaller than 2GB everything works fine, with >2GB files the decoder quits with "Fehler: Eingabedatei ist zu klein.". Same file, same decoder with win XP works fine.

The decoder is run by the following command: c:\usr\bin\pec.exe c:\programs\otrdecoder\otrdecoder.exe -e "ID" -p "PASSWD" -m -i file.otrkey -C auto

Probably any other program has same problems in handling large files.

The file flash.txt for 0.4.1 GA states:

• Registry:

This release of the Flash Plugin Wrapper no longer uses the registry to find it's location. You can check with regedit2.exe if the key "Flash10_plugin" exists, but it should be removed by the installer.

At the suggestion of Steven Levine, this could be made clearer, that the registry referred to is the OS/2 profile registry OS2.INI . I assume that the Open32 registry USER.DAT is still in use.

Suggested change:

• Registry:

This release of the Flash Plugin Wrapper no longer uses the OS/2 Profile registry OS2.INI to find its location. [ . . . ]

Odin 0.8.9 locks up Firefox immediately as soon as a video on YouTube is started. Will provide logs tomorrow. Odin 0.8.8 is fine.

Certain text does not display in popups generated by the application running under pe. (Popups generated by odin itself such as about missing dll's display correctly.) Typically the message describing the problem is missing but the labeling of buttons is present. This problem is reproducible and occurs in the odin versions released on 2011/09/30, 2011/07/10, and 2010/10/01. It does not occur in earlier odin versions I have tried. (By the way, printing works for the first time using the most recent odin versions; great progress!)

The application GPSU (Alan Murphy - murphy@…) displays a popup until the first user action showing the GPSU logo, an example map inset, the name and version of the program, and the name of the licensed user. A log file generated by the debug version of odin 0.7.1 (2011/09/30) as well as screen shots of the opening of GPSU 5.12 under windows XP and odin are attached. All of the text is missing in the odin version. The the map inset displays correctly but the program logo is incorrectly positioned in the odin version although the visible part is otherwise correctly displayed; this latter error occurs under all versions of odin I have tried and in all versions of GPSU that I have used. Presumably the map inset and the logo use different graphic formats. I mention this unrelated error here because the same logfile and screen shots exhibit both errors.

I found the played sound in LAOE and Raven very distorted. The same file played in a native OS/2 MMOS2 program is okay, no distortion. All levels are okay. As both Java programs do have distortion then its or Tritonus (both make use of the Java program) or its in Odin. But I can't check. Is there a way of checking in Odin?

[will cross ref this to openJDK bug list]

Using either openJDK GA2 or GA3 with Odin 0.8.5, printing causes the application to exit. A blank page is ejected from the printer. Printing is to an HP LaserJet?, using the Postscript driver, printing via LPR.

Rolling back to Odin 0.8.4 (and using openJDK GA2), printing works as expected.

Note JVM crash log (full log attached):

#
# A fatal error has been detected by the Java Runtime Environment:
#
#  EXCEPTION_ACCESS_VIOLATION (0xc0000005) at pc=0x15429a07, pid=156, tid=10223635
#
# JRE version: 6.0-b24
# Java VM: OpenJDK Client VM (20.0-b12 mixed mode os2-x86 )
# Problematic frame:
# C  [WGSS50+0x9a07]
#
# If you would like to submit a bug report, please visit:
#   http://java.sun.com/webapps/bugreport/crash.jsp
# The crash happened outside the Java Virtual Machine in native code.
# See problematic frame for where to report the bug.
#

The standard file open/save dialog is seriously broken in the current Odin builds: it doesn't allow to navigate through the file system properly and can't be even dismissed with OK or Cancel buttons.

This is in particular seen in Java AWT applications that use this dialog.

When using Odin in SDK mode, the console I/O is not done properly. This means that STDIN/STDOUT handles are implemented using a very restricted set of functions (HMDeviceStandardClass) which only implements reading and writing (regardless of if the handles are the real console or redirected to files etc).

As a result things like GetNumberOfConsoleInputEvents?() and PeekConsoleInput?() don't work in real console mode (although the code implementing them is present in classes HMDeviceConsoleInClass and HMDeviceConsoleOutClass). And things like SetFilePointer?() don't work in redirected mode (although the HMDeviceFile class implements them properly).

Due to these defects, the console I/O got partially broken after r22024 (which fixed a lot of inconsistencies in return codes). In particular, this affects Java which will throw an exception if you e.g. attempt to read something from STDIN.

Hi,

It would be nice if the SDK was provided to match each public Odin build.

I assume that currently, I'll need to compile odin myself to get the libs and headers?

I want to investigate compiling gtk+ using odin to see if I can get a working gimp build.

When playing videos on youtube with a color depth of 16 bits, there are color errors. Screenshot illustrates the problem.

The problem goes away, when switching to 32 bit color depth.

Using eCS 2.1 and Odin 0.8.5 installed using YUM. Apps requiring Odin (Java, Flash) seem to work well.

If I try to install the OdinApp? package (after unzipping) using 'pe setup.exe' in an eCS command window, I get the message: 'File not found: D:\var\temp\_ISMP1.DIR\_INS5176._MP.exe'. I click the 'OK' button and the install terminates.

Attempting to run Instantbird (instant messaging client based on Mozillia's xulrunner) I receive: The application has called the non-existing api KERNEL32.DLL->InitializeSListHead (loaded by MSVCR100.DLL)

Flash 11.4 needs SHGetSpecialFolderPathW, there is no definition in the normal def file but is in the debug. Added and it fixes the issue with Flash 11.4:

Index: src/shell32/shell32.def
===================================================================
--- src/shell32/shell32.def	(revision 22010)
+++ src/shell32/shell32.def	(working copy)
@@ -403,3 +403,4 @@
     SHGetDiskFreeSpaceA               = "_GetDiskFreeSpaceExA@16"             @1222
     SHGetDiskFreeSpaceExA             = "_GetDiskFreeSpaceExA@16"             @1223
     SHGetDiskFreeSpaceExW             = "_GetDiskFreeSpaceExW@16"             @1224
+    SHGetSpecialFolderPathW           = "_SHGetSpecialFolderPathW@16"         @1225

I have a guess that if we build a Win32 application from sources, we don't need to switch the FS register from the OS/2 TIB to the Win32 TIB and back. This is a continuation of the idea from #82 (which already eliminates the need to do this FS switch in order for SEH to work).

If we can get rid of this switch completely, we will get save ourselves from having a milky-way of function wrappers used to call OS/2 functions from the Win32 context (and the other way around). Getting rid of something is always good as it reduces chaos.

The current implementation of SEH (__try/__except) requires Odin to force the Win32 TIB switch mode in which the FS register points to the Win32 thread information block instead of the OS/2 one.

This mode is the default in Odin if it runs a Win32 binary as this is what the binary expects and needs. But in cases when we build Win32 apps from sources there is actually no sense in maintaining this FS context switch from OS/2 to Win32 and back. Even if the Win32 application not only uses __try/__except but also accesses FS directly, it can always be tailored with #ifdef (we can access the Win32 TIB with GetThreadTEB() regardless of the real FS value).

Switching FS may even be dangerous in some cases and this especially relates to situations when Win32 code calls OS/2 code and vice versa: the called party may not expect a wrong context (if e.g. it does its own exception handling since this is what FS is used mostly for). In either case, maintaining the context switch is error-prone per se due its complexity (you may look at e.g. sehutil.s to see it).

In order to get getURL() action script command working, Flash32 requires windows security subsystem to be implemented

I installed 0.8.4-1 with yum

I try to play this game CLUB2000.zip (Same game as in Ticket #54 (closed defect: fixed)

It does nottin, only see this in popup.log

03-19-2012  18:29:01  SYS3175  PID 472a  TID 0004  Slot 00a1
Q:\USR\BIN\PE.EXE
c0000005
1c7cc034
P1=00000001  P2=00000014  P3=XXXXXXXX  P4=XXXXXXXX  
EAX=00000001  EBX=031d0000  ECX=0004472a  EDX=13e8af20
ESI=1c020394  EDI=00004000  
DS=0053  DSACC=d0f3  DSLIM=5fffffff  
ES=0053  ESACC=d0f3  ESLIM=5fffffff  
FS=150b  FSACC=00f3  FSLIM=00000030
GS=0000  GSACC=****  GSLIM=********
CS:EIP=005b:1c7cc034  CSACC=d0df  CSLIM=5fffffff
SS:ESP=0053:032e0eac  SSACC=d0f3  SSLIM=5fffffff
EBP=032e0ef4  FLG=00010206

KERNEL32.DLL 0001:0000c034

Trying to start Jeti/2 (using Java6) crashes persistently during startup.

12-31-2011 01:04:02 SYS3170 PID 0079 TID 001c Slot 0096 L:\JAVA6\BIN\JAVA.EXE 00000000 00000000 EAX=00000000 EBX=00000000 ECX=08edf0ec EDX=2215bbe0 ESI=08edf0b0 EDI=1a618950 DS=0053 DSACC=d0f3 DSLIM=4fffffff ES=0053 ESACC=d0f3 ESLIM=4fffffff FS=4667 FSACC=00f3 FSLIM=00000fff GS=0000 GSACC= GSLIM= CS:EIP=005b:1d326020 CSACC=d0df CSLIM=4fffffff SS:ESP=0053:08edf010 SSACC=d0f3 SSLIM=4fffffff EBP=08edf048 FLG=00010246

# # A fatal error has been detected by the Java Runtime Environment: # # EXCEPTION_ACCESS_VIOLATION (0xc0000005) at pc=0x1cd00d3f, pid=121, tid=7929884 # # JRE version: 6.0-b22 # Java VM: OpenJDK Client VM (19.0-b09 mixed mode os2-x86 ) # Problematic frame: # C [IPHLPAPI+0xd3f] # # If you would like to submit a bug report, please visit: # ​http://java.sun.com/webapps/bugreport/crash.jsp # The crash happened outside the Java Virtual Machine in native code. # See problematic frame for where to report the bug. #

Current thread (0x225b7800): JavaThread? "Thread-9" [_thread_in_native, id=7929884, stack(0x08de0000,0x08ef0000)]

siginfo: ExceptionCode?=0xc0000005, writing address 0x32342e35

Registers: EAX=0x000002f8, EBX=0x00b8d340, ECX=0x00000370, EDX=0x32342e35 ESP=0x08edeab0, EBP=0x08edeae8, ESI=0x225c52c0, EDI=0x00000280 EIP=0x1cd00d3f, EFLAGS=0x00010202

Register to memory mapping:

EAX=0x000002f8 0x000002f8 is pointing to unknown location

EBX=0x00b8d340 0x00b8d340 is pointing to unknown location

ECX=0x00000370 0x00000370 is pointing to unknown location

EDX=0x32342e35 0x32342e35 is an unallocated location in the heap

ESP=0x08edeab0 0x08edeab0 is pointing into the stack for thread: 0x225b7800 "Thread-9" prio=6 tid=0x225b7800 nid=0x79001c runnable [0x08ede000]

java.lang.Thread.State: RUNNABLE

EBP=0x08edeae8 0x08edeae8 is pointing into the stack for thread: 0x225b7800 "Thread-9" prio=6 tid=0x225b7800 nid=0x79001c runnable [0x08ede000]

java.lang.Thread.State: RUNNABLE

ESI=0x225c52c0 0x225c52c0 is pointing to unknown location

EDI=0x00000280 0x00000280 is pointing to unknown location

Top of Stack: (sp=0x08edeab0) 0x08edeab0: 00000000 00000000 00000000 00000000 0x08edeac0: 1ffc5e51 00000041 08edeb50 00000002 0x08edead0: 08edeb10 08edeae4 00000000 08edeb10 0x08edeae0: 08edeafc 08edeb50 08edeb28 1f4a92a4 0x08edeaf0: 00000041 08edeb50 00000002 08edeb10 0x08edeb00: 1ffc5e51 00000041 08edeb90 00000002 0x08edeb10: 00000002 08ed027f 00004667 00000002 0x08edeb20: 00000041 1a618950 08edef68 1f4d4cf4

Instructions: (pc=0x1cd00d3f) 0x1cd00d2f: 75 f7 89 da 81 c2 d4 01 00 00 74 0a 90 83 c0 28 0x1cd00d3f: 8b 12 85 d2 75 f7 89 da 81 c2 fc 01 00 00 74 0a

Stack: [0x08de0000,0x08ef0000], sp=0x08edeab0, free space=1018k Native frames: (J=compiled Java code, j=interpreted, Vv=VM code, C=native code) C [IPHLPAPI+0xd3f]

Java frames: (J=compiled Java code, j=interpreted, Vv=VM code) j 0x23179fa5

Currently, the stack size passed to CreateThread?() is interpreted as the whole (maximum) stack size. Ноwever under real Windows, according to MSDN, this parameter by default specifies the size of the pre-committed area on top of the stack; the whole stack size is defined by the PE header in this case (1 MB by default). In order to override the header, the STACK_SIZE_PARAM_IS_A_RESERVATION flag must be used, so that the stack size in CreateThread?() will define the whole area.

Wiki and readme doesn't really reflect the current state of odin. we should change that sooner or later

​OpenProject 1.4 does not work with GCC-Odin anymore.

It used to work just perfectly with the last VAC-Odin-Builds.

It fails to start because the JVM cannot reserve enough heap for OpenProject to start.

This is more of a question that could lead to an enhancement request. There are a couple of game console emulation applications I have been using for years under ODIN. Kega Fusion and ePSXe. They work very well using GDI softdraw plugins other than fullscreen mode either crashes the emulator or it defaults to 640x480 resolution in a window with the mouse cursor hidden. under SNAP I use the zoom feature to get around it. but under Panorama there is no fix. Both apps use ini files that I can edit the fullscreen resolution but still gets over ridden and defaults back to 640x480. Is there any configuration trick I am missing to tell Odin's GDI support what the proper matching full screen resolution should be? or report back to the application.

Certain icon formats are loaded from the EXE/DLL resources incorrectly. The following problems are seen in WinRAR 4.0 GUI:

1. Shifted AND masks (Help -> About WinRAR... -> the jumping stack of books under the logo).
2. Wrong color depth (the jumping stack of books, if I hack the AND mask manually, looks like a 16-color image while under Win7 it's a full color image -- this may be a consequence of selecting an inappropriate version of the icon from the resource).
3. Wrong icon dimensions reported (the tool button width in the main window doesn't fit the icon so the icons are drawn overriding each other).

These problems seem to affect the CreateIconFromResource?()/LoadImage?() APIs. Since these APIs seem not to be used in e.g. Java or Flash, the above problems don't affect these apps. Instead they affect native Win32 apps (which usually supply the icons in the resources).

New odin applications can dinamically load the runtime libraries, so config.sys libpath changes are no longer required.

To allow finding runtime, a new string value is added to user.ini registry, under KLIBC key (as per ​http://svn.netlabs.org/libc/ticket/225).

config.sys changes are moved to a new package id, to make them optional.

I have update using yum my Odin to version 0.821 after that most of my win32 program function normally but vary other program don't giving this pop up error windoow:

Memory allocation failure while loading D:\PROD\PGMODIN\WINZIP\WINZIP32.EXE

This is a old WinZip? 8.1 that I used many time before using the most valid new archive viewer now default in eCs2.x Other programs: CALC.EXE a former Ms Calculator and other.

I downlaoded odin v0.8.1 and extract it, running a program with pe randomrunner.exe and then get a popup with Internal error while loading randomrunner.exe When i use v0.7.1 it runs ok

While browsing a web page with Flash content, SeaMonkey? crashed and produced the following *.TRP file indicating a trap in CRYPT32.dll.

move Odin to a newer buildsystem like kmk, as the currently system is very old

I can't update to 0.8.2 with "yum update libodin" and hate the idea to mix RPM and manual installations... :-)

berbible ran fine prior to GCC builds but with GCC builds: Memory allocation failure while loading E:\PRGFILES\SOFSTUF\BERBIBLE\BERBIBLE.EXE An Odin popup with the above pops up when trying to launch it with GCC builds.

It is possible that forceful thread termination with TerminateThread?()/ExitThread?() causes resource leaks (i.e. internal handle data).

This is at least what was taking place when multimedia timer threads were terminated this way (see ​http://svn.ecomstation.nl/flash10/ticket/28 for details). After removing these calls in r21930, the leak problem was solved.

However, in theory, these calls should not cause such a bad effect on the rest of the application and this ticket is to check them in this regard.

winrar requires GetComboBoxInfo? for it password box, I cannot test this code as with the GCC/KMK changeover Odin will crashes:

Index: include/win/winuser.h
===================================================================
--- include/win/winuser.h	(revision 21924)
+++ include/win/winuser.h	(working copy)
@@ -1922,6 +1922,7 @@
 #define CB_GETDROPPEDWIDTH       0x015f
 #define CB_SETDROPPEDWIDTH       0x0160
 #define CB_INITSTORAGE           0x0161
+#define CB_GETCOMBOBOXINFO       0x0164
 
 /* Combo box notification codes */
 #define CBN_ERRSPACE        (-1)
@@ -3355,9 +3356,9 @@
 #endif /* defined(_WINGDI_) && !defined(NOGDI) */
 
 HKL         WINAPI ActivateKeyboardLayout(HKL,UINT);
-LONG        WINAPI BroadcastSystemMessageA(DWORD,LPDWORD,UINT,WPARAM,LPARAM); 
-LONG        WINAPI BroadcastSystemMessageW(DWORD,LPDWORD,UINT,WPARAM,LPARAM); 
-#define     BroadcastSystemMessage WINELIB_NAME_AW(BroadcastSystemMessage) 
+LONG        WINAPI BroadcastSystemMessageA(DWORD,LPDWORD,UINT,WPARAM,LPARAM);
+LONG        WINAPI BroadcastSystemMessageW(DWORD,LPDWORD,UINT,WPARAM,LPARAM);
+#define     BroadcastSystemMessage WINELIB_NAME_AW(BroadcastSystemMessage)
 WORD        WINAPI CascadeWindows(HWND, UINT, const LPRECT, UINT, const HWND *);
 BOOL        WINAPI CascadeChildWindows(HWND, DWORD);
 INT       WINAPI CopyAcceleratorTableA(HACCEL,LPACCEL,INT);
Index: src/user32/combo.c
===================================================================
--- src/user32/combo.c	(revision 21924)
+++ src/user32/combo.c	(working copy)
@@ -1885,7 +1885,24 @@
    }
 }
 
+static LRESULT COMBO_GetComboBoxInfo(const HEADCOMBO *lphc, COMBOBOXINFO *pcbi)
+{
+    if (!pcbi || (pcbi->cbSize < sizeof(COMBOBOXINFO)))
+        return FALSE;
 
+    pcbi->rcItem = lphc->textRect;
+    pcbi->rcButton = lphc->buttonRect;
+    pcbi->stateButton = 0;
+    if (lphc->wState & CBF_BUTTONDOWN)
+        pcbi->stateButton |= STATE_SYSTEM_PRESSED;
+    if (IsRectEmpty(&lphc->buttonRect))
+        pcbi->stateButton |= STATE_SYSTEM_INVISIBLE;
+    pcbi->hwndCombo = lphc->self;
+    pcbi->hwndItem = lphc->hWndEdit;
+    pcbi->hwndList = lphc->hWndLBox;
+    return TRUE;
+}
+
 /***********************************************************************
  *           ComboWndProc_common
  *
@@ -2278,7 +2295,8 @@
     case CB_GETEXTENDEDUI16:
     case CB_GETEXTENDEDUI:
         return  (lphc->wState & CBF_EUI) ? TRUE : FALSE;
-
+    case CB_GETCOMBOBOXINFO:
+        return COMBO_GetComboBoxInfo(lphc, (COMBOBOXINFO *)lParam);
     default:
         if (message >= WM_USER)
             WARN("unknown msg WM_USER+%04x wp=%04x lp=%08lx\n",
@@ -2309,3 +2327,13 @@
     if (!IsWindow(hwnd)) return 0;
     return ComboWndProc_common( hwnd, message, wParam, lParam, TRUE );
 }
+
+/*************************************************************************
+ *           GetComboBoxInfo   (USER32.@)
+ */
+BOOL WINAPI GetComboBoxInfo(HWND hwndCombo,      /* [in] handle to combo box */
+                PCOMBOBOXINFO pcbi   /* [in/out] combo box information */)
+{
+    TRACE("(%p, %p)\n", hwndCombo, pcbi);
+    return SendMessageW(hwndCombo, CB_GETCOMBOBOXINFO, 0, (LPARAM)pcbi);
+}
Index: src/user32/user32.def
===================================================================
--- src/user32/user32.def	(revision 21924)
+++ src/user32/user32.def	(working copy)
@@ -769,3 +769,6 @@
 
     "_RunOnAuxThread@20"                                         @2048 NONAME
     "_RunOnAuxThreadAndWait@24"                                  @2049 NONAME
+
+; New but not used by Flash
+    "_GetComboBoxInfo@8"                                         @2050

we should also move TCPIP V4 headers to V4.1 as V4 is a very old. this could give more general code with other ports.

When DosLoadModule? loads a DLL with the name having .DLL it will search for the file minus the DLL in the name. If the name does not have an extension or the extension is in lower case (.dll) then it will load. Odin has a number of obvious uses of DosLoadModule? with .DLL. There are some places where the file name is supplied by variable name... we may need to lowercase some variables if there are any reports of failures with libpathstrict.

My boot parttion is E: . There isn't even a C: partition on this system, yet no error message. Is a config.sys change even necessary? Looks like the .lib files have been installed in usr\lib and the .exe in usr\bin which are in the path already.

Installing : odin-win32k-0.7.1-3.i386 14 18

NOTE:

The file 'C:\config.sys' has been changed. You need to reboot your computer in order to activate these changes.

Installing : odin-exe-tools-0.7.1-3.i386 2/3

Installing : odin-doc-0.7.1-3.noarch 3/3

odin-doc-0.7.1-3.noarch was supposed to be installed but is not! odin-exe-tools-0.7.1-3.i386 was supposed to be installed but is not! odin-win32k-0.7.1-3.i386 was supposed to be installed but is not!

Installed:

odin-doc.noarch 0:0.7.1-3 odin-exe-tools.i386 0:0.7.1-3

odin-win32k.i386 0:0.7.1-3

Complete!

Not sure if this is an odin or java6 problem but Arachnophilia - ​http://www.arachnoid.com/arachnophilia/ - fails to display the file open dialog.

The error log generated by this app when trying to open a file contains the following error:-

java.io.IOException: Could not get shell folder ID list

After installing ODIN by running YUM INSTALL LIBODIN the portable executable PE.EXE resides in .\usr\lib. In the adapted config.sys the subdirectory .\usr\lib should be in the PATH statement to run PE.EXE from anywhere, but it's not.

Odin does not support recording sound. This is something which is not of top importance, but still cool to have.

There is an annoying recursion in the Odin exception handler where it calls DosExit?() while processing the XCPT_PROCESS_TERMINATE and XCPT_ASYNC_PROCESS_TERMINATE (which are themselves the result of DosExit?()).

This recursion in particular leads to a big POPUPLOG.OS2 file (filled up with 0xC0010001 and 0xC0010002 exception records) at Java process termination (either normal or abnormal).

This needs to be solved as it indicates an error in the exception processing logic.

It is known that if there are lots of heavy applications loaded when Odin starts (e.g. SeaMonkey? + FireFox? + OpenOffice?), it may fail to initialize KERNEL32.DLL. Sometimes it fails in DOS calls (e.g. DosQueryModuleName?()), sometimes in even malloc(). See #23 for more information.

For now, we just show a nice error message asking the user to free memory by closing other applications and retry, but it would be also good to track these allocation problems down to see why exactly it fails to get memory.

odin tends to hang the active process when a odinized app is terminated.

it may be due to DosExitList?() hangs.

Odin32 already supports the SystTray? eCenter plugin by MadInt? however we have a better implementation called xsystray (so far, it's part of our Qt4 port). We should add support for xsystray too.

Due to bugs in the DosQuerySysState? API, the returned buffer seems to contain circular references under some (quite random) conditions. This makes the walkModules() worker (used by DosQueryModuleHandleStrict?()) recurse into itself until it eats up all the stack.

From the user perspective, this looks like an application giving 100% CPU load for some tens seconds and then crashing with SYS1808.

Odin contains an emulation library for MSVCRT in src/msvcrt which in particular defines MS-specific functions like _wgetdcwd() or _wfullpath(). Since these functions are frequently used in Windows sources (and in particular, in Java), it makes sense to resurrect this library to avoid fixing each such invocation individually.

A user reports (​http://svn.netlabs.org/java/ticket/41) that he cannot enter umlauts with the recent Odin release under some circumstances.

This is clearly a regression of r21529.

Odin does not use TZ to define the Windows time zone (GetTimeZoneInformation?() API). Instead, it uses WGSS for the initial TimeZoneInformation? setting but WGSS returns GMT on freshly installed systems.

We need to derive TimeZoneInformation? from TZ each time when KERNEL32.DLL gets loaded and update it if differs. This will make sure that if the user changes his system time zone setting (via the Clock Synchronization setting or manually in config.sys), it will get picked up by Odin.

e:\svn-code\odin32[0]svn up . svn: Can't check path 'testapp/encodings/.cmd': SYS0123: A file name or volume label contains an incorrect character. It does not show from the copy and paste but between the / after encodings and the .cmd there are four arrows that point to the right. From the web interface I see the file as: тЅст.cmd

e:\svn-code\odin32[1]help sys0123

SYS0123: A file name or volume label contains an incorrect character.

EXPLANATION: The system does not accept characters displayed using the keyboard combination ALT plus 0-32 or the following characters:

\ / [ ] : | < > + ; = . ? * "

ACTION: Correct the file name or volume label and retry the command.

There is a bunch of constants (such as IP_ADD_MEMBERSHIP) that were redefined by Microsoft in WinSock2 and have values incompatible with WinSock1. We miss these new defines in Odin headers and simply fall back to old ones.

While it seems to work at binary level (because the values for constants supplied to Odin by the application are taken from the real MS SDK when the application is built and get properly recognized by the new WinSock2 APIs in the Odin code), compiling applications at source level drags the old constants in while actually calling the new WinSock2 APIs and as a result these APIs get wrong parameters and return a failure.

In application, heavily uses handlers, overrun handler index causing crash