State of the killqt4 branch (Stephan?)

[Stephan Witt]

Am 25.11.2022 um 00:55 schrieb Stephan Witt <st.witt at gmx.net>:
> 
> Am 25.11.2022 um 00:29 schrieb Jean-Marc Lasgouttes <lasgouttes at lyx.org>:
>> 
>> Le 25/11/2022 à 00:18, Stephan Witt a écrit :
>>> This is a profiler run with Apple Instruments - I built the LyX binary for profiling.
>>> https://my.hidrive.com/share/6t6dr1c.4o
>>> The experiment was to open the german tutorial and start scrolling from top to bottom and back to top.
>>> The left side is the profiler run with cache and the opposite side is without cache.

Sorry, I didn’t say what I’ve used for it. The code is the master branch from origin.

I’ve built it with a Xcode project created by cmake and self compiled Qt 6.3.1 frameworks (release mode).

The system is MacBook Pro (15 Zoll, 2019) with macOS 10.14.6 Mojave, 2.4 GHz Intel Core i9, Radeon Pro 560X 4GB graphics.

Stephan

>> 
>> What is the difference between the two images you uploaded.
> 
> The deepness of the open call graph is different. The 2nd shows more details inside of the getTextLayout_helper function.
> 
>> It seems that the cache is needed. Can you make a view where all the versions of getTextLayout are visible? There is a second one that uses a TextLayoutHelper parameter.
> 
> I'm not sure how to do that - this is an excerpt of the heaviest stack trace.
> 
> I’ve changed the filter from „matches all“ to „matches any“ (?) and then I tried to locate examples of these.
> 
> The result is the 3rd image now.
> 
>> It seems anyway that the cache should be kept on a mac.
> 
> Yes. It seems so.
> 
> Stephan
> -- 
> lyx-devel mailing list
> lyx-devel at lists.lyx.org
> http://lists.lyx.org/mailman/listinfo/lyx-devel