In this case the fix is for a Null reference as indicated in the accompanying tests. The test prangs in the current build and the patch for pdh_main.c fixes the behaviour on XP. As I said, I can't tell if it
How can a fix to pdh_main.c (which is Wine's implementation) fix something on XP?
The test shows a fault in wine not XP, sorry if I confused the issue. When this test is executed it causes a NULL reference in wine. The fix is for wine.
fixes it in anything else. However, there needs to be a fix for the NULL reference and the case of needing a buffer if size and index > 0 seems to be reasonable behaviour.
When a test fails on some version of Windows something is wrong with the test itself. Possible solutions are to make the test accept the different behavior too or to remove the test altogether when it is sufficiently clear that applications cannot depend on any particular behaviour.
I have no idea whether this test has a problem on one version of Windows or another. I only know the behaviour in XP. This is not a case of XP explicit behaviour it is simply a case of NULL reference. I split the fix into two lines rather than merging it into one because it made the test pass ok in wine and Windows XP. I put a comment to say that as I am sure that someone would have picked that up and queried it. My question was whether this was the same behaviour in other flavours of windows. I suspect it is ok but... your previous posts indicated that there may be problems.
Jeff