bug#63187: 30.0.50; Tail of longer lines painted after end of nearby lines on macOS

[Kai Ma]

> On Jun 8, 2023, at 13:40, Kai Ma <justksqsf@gmail.com> wrote:
> 
> 
> Hi everyone, I'm also bothered by this bug recently.  And it is
> happening quite frequently for me.
> 
> Unfortunately, the proposed patch and putting
>  [[NSGraphicsContext currentContext] flushGraphics];
> in copyRect didn't work for me.
> 
> See the screencast at https://www.youtube.com/watch?v=uLYPqsUFK5A .

Upon further debugging, the following change seems to fix the problem for me.

--- a/src/nsterm.m
+++ b/src/nsterm.m
@@ -10433,7 +10433,7 @@ @implementation EmacsLayer
    cache.  If no free surfaces are found in the cache then a new one
    is created.  */
 
-#define CACHE_MAX_SIZE 2
+#define CACHE_MAX_SIZE 1
 
 - (id) initWithColorSpace: (CGColorSpaceRef)cs
 {


This probably implies that there is another bug somewhere else in the 
management of IOSurface caches.


> 
> I find that this bug happens more frequently if
> 1) the frame's alpha < 1.0, and
> 2) the frame is in the full screen mode, and
> 3) emacs is built with native compilation enabled.
> 
> (Not doing one or two of these does not prevent the bug from happening,
> but seems to reduce the likelihood.)
> 
> FWIW, some version numbers:
> - emacs 30.0.50 master branch
> - macOS 13.4 (22F66), Intel
> 
> 
> Aaron Jensen <aaronjensen@gmail.com> writes:
> 
>> Thank you. I'll report back if I start seeing it again, but so far so good.
>> 
>> Aaron
>> 
>> On Thu, May 18, 2023 at 7:21 AM, Eli Zaretskii <eliz@gnu.org> wrote:
>> 
>> From: Aaron Jensen <aaronjensen@gmail.com> 
>> Date: Sat, 13 May 2023 10:23:28 -0400 
>> Cc: luangruo@yahoo.com, alan@idiocy.org, 63187@debbugs.gnu.org 
>> 
>> On Sat, May 13, 2023 at 9:54 AM Eli Zaretskii <eliz@gnu.org> wrote: 
>> 
>> Ping! Should the patch be installed now on the emacs-29 branch? 
>> 
>> I think it's worth it. I may have seen 1 glitch (but it was different 
>> -- it looked like an entire blank line was duplicated. Other than that, it's 
>> been solid, which is quite a bit better than it had been
>> before. 
>> 
>> Now installed on the emacs-29 branch, and closing the bug.
>>