Hi
I have been trying to nail what is causing this behaviour, the resizing of standard controls and custom controls are much slower in trunk by a factor of at least 4.
also i have controls that are shifiting their position.
I have attached a project which will demonstrate this if you run it, it creates a multiple copies of the tbutton and timage ( this was done to visually show the problem),
I have created my own form.resize routine that iterates through the button and images ( from number 2, which is the ones created at form create, and positions them acording to the position of the initial tbutton/timage.
If you res-ze the form you will see odd effects, the app allows you to turn on app.processmessages, use formbrgin/endupdate and also allows you to set the creates controls anchors to either none [] or akTop+akLeft.
if you do this again other odd things happen, what is concerning is that the timage (master actually moves it position and theother follow it), which is odd as the app does not change its ordinates.
On my devel machine
using laz 1.6.5 and fpc 2.6.5 I get a reasonable rapid response of aroun 11ms, using trunk I get 66ms.
On a lower spec laptop trunk can give a figure of over 500ms to execute the resize, in laz 1.6 the same params i get about 60ms.
Running on a laptop highlighted a strange effect, its as though the components and panels are being re-drawn multiple times at slightly different positions then finally being drawn at the correct co-ordinates.
I cannot use anchors in many of my apps as the postion, size of componentsare dependent on what user has selected and what best fits the screen ration, so some time controls may or may not not be visible, and dependent on the screen ratio some controls can be under others or to their side, and this has to be smooth and fast, it has previously, but something in trunk has changed.
If their is something happening in trunk, that is trying to auto alighn, rescale is their away to turn it off so the app can control what it wants to.
The app also keeps a track of how many times events are fired, this is interesting also as it appears trunk is somehow filtering some 'Duplicate' events from the teh message queue.
Apologies for the wall of text, but if you were to run this on various versions of lazarus I think you will get similar result and questions?
I have tested mainly on Windows, but the issue is also on Mac.
firev1 laptop trunk
firev2 laptop 1.6.5 2.6.5