Am Freitag, 2. Januar 2004 17:28 schrieb Mike Hearn:
Hi Stefan,
The attached patch is necessary for IE installer to work again. It seems the magic value is not always correct: advpack derefs this value so it probably points to a struct. I'll try and find out what it contains when using the native DLL (if anything).
It seems to be optional, making it 0 allows IE to continue.
Hi Mike, your are right. Its a pointer to a structure.
By running native cabinet.dll i found that the dll allocates two memory areas per file inside the cabinet that are not freed within the extract function. The first memory area is always 12 bytes. The size of the second one is the length of the filename without path plus 1.
The pointer points to a structure like this:
struct ExtractFileList { LPSTR filename; struct ExtractFileList *next; DWORD unknown; // always 1L } ;
Did your start modifying/fixing the extract function for this item? If not i will do it.
Bye Stefan
On Tue, 2004-01-06 at 17:59, Stefan Leichter wrote:
The pointer points to a structure like this:
struct ExtractFileList { LPSTR filename; struct ExtractFileList *next; DWORD unknown; // always 1L } ;
Could the unknown be a boolean?
Did your start modifying/fixing the extract function for this item? If not i will do it.
No, I didn't, please do it :)
thanks -mike
On Tuesday 06 January 2004 12:10 pm, Mike Hearn wrote:
On Tue, 2004-01-06 at 17:59, Stefan Leichter wrote:
The pointer points to a structure like this:
struct ExtractFileList { LPSTR filename; struct ExtractFileList *next; DWORD unknown; // always 1L } ;
Could the unknown be a boolean?
Did your start modifying/fixing the extract function for this item? If not i will do it.
No, I didn't, please do it :)
thanks -mike
Just a wild guess -- might this unknown boolean be whether or not (1 meaning "not," I guess) to "run" the extracted file? This is one part of the "standard" that is rather unclear (for example, what does "run" mean, exactly, and who does the running?), so one never knows where it might pop up, and it would make sense that real-world instances were "always" the same...
Thanks for all these investigations; wish I was able to provide more help right now, but I have very little time ATM.
-gmt
Am Dienstag, 6. Januar 2004 19:10 schrieb Mike Hearn:
On Tue, 2004-01-06 at 17:59, Stefan Leichter wrote:
No, I didn't, please do it :)
Hi Mike
does the patch work for you, or does it break your installer again?
Bye Stefan