|
MSLU: reported bugs and known issues
(The Microsoft Layer for Unicode on Windows 95, 98, and Me Systems)
(Originally posted 27 April 2002; last updated 10 December 2004)
By popular request, this page is going to list known issues that are either under investigation or fixed for a future release. Most issues start life being reported
in the microsoft.public.platformsdk.mslayerforunicode newsgroup on the msnews.microsoft.com news server and you can often see discussion about the issues there while
they are under investigation.
If you have any questions about or problems with MSLU, whether or not they are related to issues listed below, you should feel free to ask them in the
mslayerforunicode newsgroup (available via both NNTP and
HTML). You can also come by with
good news, if you want to. :-)
The latest Platform SDK contains a UnicoWS.lib file that was build using the VC++ 7.0 compiler, and the format is not 100% compatible with the linker in VC++ 6.0. In order to use the .LIB file, you need to either not use the /DEBUG flag or use VC++ 7.0. This problem will be addressed in the next PSDK update.
On a Windows 95 system with no net installed, no mpr.dll is on the system. This will cause a LoadLibrary() of unicows.dll to fail since it cannot bind to any of the WNet* functions that MSLU needs. The easy workaround is to just install the net on the target Win95 machine -- I have not seen such a machine in years and can't imagine that such machines are installing new software....
If you have a RichEdit control supporting Unicode (using the RichEd20W class name) and you attempt to use the GetWindowTextW API or the WM_GETTEXT message to retrieve the text it contains, the text will be garbled. A fix is not currently under consideration because the EM_GETTEXTEX message is available and will work properly, and it will not convert the text via the default system codepage the way that MSLU would for WM_GETTEXT or GetWindowTextW.
When MSLU's BeginUpdateResourceW and BeginUpdateResourceA APIs are called on binaries with no resources with the bDeleteExistingResources parameter set to FALSE, MSLU's call to LoadLibraryExA with the LOAD_LIBRARY_AS_DATAFILE flag to enumerate existing resources will fail and GetLastError() will return ERROR_BAD_FORMAT. This is a limitation of the Win9x function and there is no workaround for this, other than to simply pass TRUE when there are no resources in the binary upon which you plan to call the resource updating functions (since MSLU cannot afford to risk calling without LOAD_LIBRARY_AS_DATAFILE and causing code to run).
When an EDIT control is subclassed, passing the EM_GETLINE message to it will fail anytime the window has been subclassed.
When MSLU's GetFullPathNameW wrapper is called with a buffer size that cannot fit the returned string, 0 will currently be returned rather than the required size (the documented behavior) and MSLU may attempt to improperly copy to the buffer.
The GetTempPathW API in MSLU has a similar problem, though with GetTempPathW there is no risk to the API improperly copying data to the buffer.
MSLU's CreateEnhMetaFileW wrapper is having problems with a NULL lpszDescription.
When zero length strings are used for either of its string parameters, MSLU's CompareStringW wrapper is returning failure.
MSLU's CopyEnhMetaFileW wrapper is having problems with a NULL lpszfile parameter.
The VerQueryValueW API wrapper provided by MSLU uses a special buffer provideed by increasing the requested size returned by the GetFileVersionInfoSizeW API. After this, calls to VerQueryValueW will reuse that buffer. Obviously this method will not work properly if you use alternate APIs to load the version resource.
There are problems with strings in WM_DEVICECHANGE notifications being NULL, due to incorrect setting of the structure size.
There are times that the syntethic clipboard formats that handle the automatic mappings between CF_UNICODETEXT and CF_TEXT/CF_OEMTEXT that happen when you call the MSLU GetClipboardData API will not work properly, and will leak memory. This does not happen on clipboard operations inside of the application of non-synthetic mappings; it only happens on cross-process synethtic clipboard operatations.
This page is obviously not on the Microsoft web site, but it is maintained by an employee of Microsoft who is the principal developer for MSLU. Certain factors
beyond normal control can cause there to be a difference between what you see here and what is eventually released. As they say, this posting is provided "AS IS"
with no warranties, and confers no rights. With that said, best efforts to keep this page accurate will be made any time such a difference does, in fact, exist.
Problems with this site? Please contact the
webmaster@trigeminal.com
with your comments, questions, or suggestions.
|
|
