http://bugs.winehq.org/show_bug.cgi?id=5139
------- Additional Comments From jasonspiro3+wine@gmail.com 2006-19-05 13:50 ------- I am posting this issue here per mike_m. I suspect the row in the Media table that mentions accessor.cab is being selected when it shouldn't be, and that if it were not selected this bug wouldn't have appeared at all.
Wine-msi is executing the SQL statement "SELECT * FROM `Media` WHERE `LastSequence` >= 1 ORDER BY `LastSequence`". That select statement should not be selecting the row that mentions DiskId="22" LastSequence="0" DiskPrompt="" Cabinet="accessor.cab" VolumeLabel="" Source="".
The only line in a +all log of an America's Army install that mentions the Media table is:
trace:msi:MSI_DatabaseOpenViewW L"SELECT * FROM `Media` WHERE `LastSequence` >= 1 ORDER BY `LastSequence`" 0x7fc2f794
But, because the install fails later, I suspect the row is being selected despite that LastSequence is 0. How else could Wine-msi think that a file named accessor.cab exists?
(Background info: There is a full MSIDIFF dissasembly of America's Army.msi attached to bug 4533; I saw the row (DiskId="22" LastSequence="0" DiskPrompt="" Cabinet="accessor.cab" VolumeLabel="" Source="") in comment#5. I confirmed that that was the only row in the entire MSI file mentioning a file "accessor.cab" using the WiX decompiler. I am using a copy of wine from CVS that I checked out about a week ago built on Debian Etch. wine --version returns 0.9.12.)