[PATCH 1/2] comctl32/listbox: Don't try to paint non-visible items for multi-column listboxes

[Nikolay Sivov]

On 5/20/19 3:29 PM, Gabriel Ivăncescu wrote:
> On 5/20/19 3:24 PM, Nikolay Sivov wrote:
>> On 4/22/19 3:32 PM, Gabriel Ivăncescu wrote:
>>
>>> Signed-off-by: Gabriel Ivăncescu <gabrielopcode at gmail.com>
>>> ---
>>>
>>> Currently the early break happens only with single-column listboxes. 
>>> This
>>> makes it so that it works properly with multi-column listboxes as well.
>>>
>>>   dlls/comctl32/listbox.c | 1 +
>>>   1 file changed, 1 insertion(+)
>>>
>>> diff --git a/dlls/comctl32/listbox.c b/dlls/comctl32/listbox.c
>>> index bd9cffd..69fa56a 100644
>>> --- a/dlls/comctl32/listbox.c
>>> +++ b/dlls/comctl32/listbox.c
>>> @@ -1149,6 +1149,7 @@ static LRESULT LISTBOX_Paint( LB_DESCR *descr, 
>>> HDC hdc )
>>>               rect.right += descr->column_width;
>>>               rect.top = 0;
>>>               col_pos = descr->page_size - 1;
>>> +            if (rect.left >= descr->width) break;
>>>           }
>>>           else
>>>           {
>> How does this work for RTL case?Existing optimization for single 
>> column case is testing in vertical direction, so it's not the same.
>>
>>
>
> Hi Nikolay,
>
> I think it works fine since the coordinates are mirrored (i.e. 
> increasing means going left). Because the code already does rect.right 
> += descr->column_width; just above that, so that line would be wrong 
> already if it didn't work in RTL. (it would paint items in the wrong 
> order)
Fair enough. I tried it briefly before replying and it showed some 
artifacts when resizing, that does not happen in LTR. Turns out it 
glitches without your patch as well.