"Jeff Latimer" lats@yless4u.com.au wrote:
I looked at that and it required another test. When I discussed this last with Dmitry, I said it would be best to pass machine and he agreed that this would be best. The checking for alloc failures will generate similar amounts of code to what I am removing. I am happy to do it but have a concern that the code will look to be inefficient and not accepted. Logically, the patch is the correct way to proceed as machine should be passed even if not implemented yet. That said, if the patch is acceptable, there is another case that will need to be brought into line.
Hans means something like the following done in PdhOpenQueryA:
WCHAR *sourceW = NULL; if (source && !(sourceW = pdh_strdup_aw( source ))) return PDH_MEMORY_ALLOCATION_FAILURE;
and that's certainly not a large amount of code. Actually all pdh_strdup_aw callers already check for a possible NULL, but that may be sent as a separate patch IMO.