Saulius Krasuckas saulius2@ar-fi.lt writes:
Hello folks,
- On 2022-01-28 10:29, Nikolay Sivov wrote:
--- a/dlls/dwrite/tests/analyzer.c +++ b/dlls/dwrite/tests/analyzer.c
...
@@ -155,7 +152,7 @@ static inline void flush_sequence(struct call_sequence **seg, int sequence_index { struct call_sequence *call_seq = seg[sequence_index];
- HeapFree(GetProcessHeap(), 0, call_seq->sequence);
- free(call_seq->sequence); call_seq->sequence = NULL; call_seq->count = call_seq->size = 0;
}
may I ask: what's about all the HeapAlloc() => malloc() huss inside the test (or any tests) ? 15y ago the idea was to use as much of WinAPI as possible (at least in DLLs implementation). I say the project is splitting modules into PE-part and the *nix-part these days. Looks nice.
But why to reverse the HeapAlloc() usage in tests?
For all modules that have been converted to PE, malloc() is now also going through the Windows API, so there's no longer a reason to avoid it. And it's more standard and more readable than HeapAlloc().