Re: Changed behavior of ::SHFileOperation() in Vista
- From: David Wilkinson <no-reply@xxxxxxxxxxxx>
- Date: Tue, 03 Oct 2006 10:55:22 -0400
Christopher Davis [MSFT] wrote:
Thanks David. I looked these up and there are customer bugs tracking your issues - so they are being looked into.
Christopher Davis [MSFT]
"David Wilkinson" <no-reply@xxxxxxxxxxxx> wrote in message news:usPj4gj4GHA.2144@xxxxxxxxxxxxxxxxxxxxxxx
Christopher Davis [MSFT] wrote:
Thanks David. I've debugged the issue and this appears to be a valid bug in Vista. It only seems to materialize when you specify FOF_SILENT and the hwnd member of the SHFILEOPSTRUCT is NULL. I'll follow up with any further progress.[snip]
Thanks Christopher!
If you are interested in other possible developer-related bugs in Vista, I have put up a web page at
http://www.effisols.com/vista/
Several of the issues described therein were resolved in build 5728, but a few remain, in particular two where Vista incorrectly reports that a setup or uninstall failed. Another is that Vista does not appear to virtualize the HKCR registry hive, despite the fact that writing to HKCR is normally equivalent to writing to HKLM; this causes failure of the MFC CWinApp::RegisterShellFileTypes() method.
David Wilkinson
Visual C++ MVP
Thanks, Chris.
I have just noticed that the error "Hidden message box on uninstall" which I listed as fixed in build 5728 is only fixed in the x86 version. In x64 the problem remains. I will file this with the Microsoft beta client (unrewarding though it is).
I have to say IMHO that the problems described on my web page related to erroneous reports that an application did not install or uninstall correctly are extremely serious, and will create a negative experience for many users. The example given in MSDN is that of Microsoft Intellitype (a Microsoft application, no less!) that was hard wired only to install under Windows 2000 or XP. Such a setup program deserves to fail, and should not be a reason to create a negative experience for users of correctly written setups that will run on any version of Windows. I think using lack of change in the program database to indicate failure is a fundamentally flawed idea, and the ability to turn this feature off using a manifest will not help the great number of legacy installations that will be negatively impacted (ironically, these are the ones that are supposed to "helped" by this feature).
I would welcome the opportunity to discuss these issues with the team involved with this aspect of Program Compatibility.
David Wilkinson
Visual C++ MVP
.
- Prev by Date: Re: error MFC80UD.dll not found when running in debug mode
- Next by Date: Re: Embedding Simple MFC GUI app into website
- Previous by thread: Re: Changed behavior of ::SHFileOperation() in Vista
- Next by thread: Re: Changed behavior of ::SHFileOperation() in Vista
- Index(es):
Relevant Pages
|
|
