On 09/14/2009 03:10 PM, Huw Davies wrote:
programs/winetest/main.c | 87 ++++++++++++++++++++++++++++++++++++++------- 1 files changed, 73 insertions(+), 14 deletions(-)
Hi Huw,
Winetest doesn't do a complete run anymore (tested only on W2K3 and Vista) for now. It ends with a 998 error during enumeration of the test (oleaut32 being the last visible one).
I'm crossbuilding a winetest as we speak but just thought I'd mention it.
On 09/17/2009 08:13 AM, Paul Vriens wrote:
On 09/14/2009 03:10 PM, Huw Davies wrote:
programs/winetest/main.c | 87 ++++++++++++++++++++++++++++++++++++++------- 1 files changed, 73 insertions(+), 14 deletions(-)
Hi Huw,
Winetest doesn't do a complete run anymore (tested only on W2K3 and Vista) for now. It ends with a 998 error during enumeration of the test (oleaut32 being the last visible one).
I'm crossbuilding a winetest as we speak but just thought I'd mention it.
The attached patch fixes the issue.
We are using LOAD_LIBRARY_AS_DATAFILE as we don't really want to load the dll but we need some info from it. The problem is that GetModuleFileName can't cope with LOAD_LIBRARY_AS_DATAFILE loaded dlls.
So will using LoadLibrary introduce issues for this dll (and maybe future com dlls)?
The get_dll_path logic was introduced for .NET as we ourselves don't always know which dll/version is actually loaded. GetModuleFileName proved to be the most logic approach.
As these com dlls are hardcoded in the registry I assume we don't even need to call GetModuleFileName and just stick with that value from the registry?
Paul Vriens wrote:
The attached patch fixes the issue.
We are using LOAD_LIBRARY_AS_DATAFILE as we don't really want to load the dll but we need some info from it. The problem is that GetModuleFileName can't cope with LOAD_LIBRARY_AS_DATAFILE loaded dlls.
So will using LoadLibrary introduce issues for this dll (and maybe future com dlls)?
The get_dll_path logic was introduced for .NET as we ourselves don't always know which dll/version is actually loaded. GetModuleFileName proved to be the most logic approach.
As these com dlls are hardcoded in the registry I assume we don't even need to call GetModuleFileName and just stick with that value from the registry?
Hi Paul,
Thanks for tracking this down. There shouldn't be any issues with using LoadLibrary, so we can either use your patch or just rely on the info from the registry, I don't think it really matters. Shall I leave you to send in a patch?
Thanks again, Huw.
On 09/17/2009 11:39 AM, Huw Davies wrote:
Paul Vriens wrote:
The attached patch fixes the issue.
We are using LOAD_LIBRARY_AS_DATAFILE as we don't really want to load the dll but we need some info from it. The problem is that GetModuleFileName can't cope with LOAD_LIBRARY_AS_DATAFILE loaded dlls.
So will using LoadLibrary introduce issues for this dll (and maybe future com dlls)?
The get_dll_path logic was introduced for .NET as we ourselves don't always know which dll/version is actually loaded. GetModuleFileName proved to be the most logic approach.
As these com dlls are hardcoded in the registry I assume we don't even need to call GetModuleFileName and just stick with that value from the registry?
Hi Paul,
Thanks for tracking this down. There shouldn't be any issues with using LoadLibrary, so we can either use your patch or just rely on the info from the registry, I don't think it really matters. Shall I leave you to send in a patch?
Thanks again, Huw.
I'll sent a patch with the LoadLibrary.