[PATCH v2 resend 3/6] comctl32/listview: Send LVN_ODSTATECHANGED only for virtual lists.

[Angelo Haller]

On 10/06/2022 03.13, Zhiyi Zhang wrote:
>
> On 5/26/22 04:00, Angelo Haller wrote:
>> From: Angelo Haller <angelo at szanni.org>
>>
>> The LVN_ODSTATECHANGED notification should only be sent to lists
>> that have LVS_OWNERDATA set.
>>
>> Signed-off-by: Angelo Haller <angelo at szanni.org>
>> ---
>>   dlls/comctl32/listview.c | 1 +
>>   1 file changed, 1 insertion(+)
>>
>> diff --git a/dlls/comctl32/listview.c b/dlls/comctl32/listview.c
>> index 72ade724313..318df0a4093 100644
>> --- a/dlls/comctl32/listview.c
>> +++ b/dlls/comctl32/listview.c
>> @@ -8946,6 +8946,7 @@ static VOID LISTVIEW_SetOwnerDataState(LISTVIEW_INFO *infoPtr, INT nFirst, INT n
>>   {
>>       NMLVODSTATECHANGE nmlv;
>>   
>> +    if (!(infoPtr->dwStyle & LVS_OWNERDATA)) return;
> Make sense. It will be better if you can add a simple test before this patch and remove the todo_wines after the fix.
Is this strictly necessary? The call site has special handling for:
infoPtr->dwStyle & LVS_OWNERDATA

We could guard the call to the function at the call site, if that is 
preferred.

Apart from that I am not sure how to write a test for this. This bug is 
triggered by creating a non ownerdata list and selecting multiple 
entries holding shift+ctrl and clicking with the mouse. This will send 
an LVN_ODSTATECHANGED notification where it is not supposed to (as it is 
not an ownerdata list).

I have not found any test examples with mouse emulation. Maybe I missed 
something.

Or should I open a separate bug report for this line?

>>       if (!item) return;
>>   
>>       ZeroMemory(&nmlv, sizeof(nmlv));