16 replies to this topic

[majkinetor]

Thank you Lexikos. I also concluded that WM_PAINT is problematic but I didn't have definitive proof for it. It was happening that OutputDebug starts spamming that it is processing it just before script go freeze.

I will replace my binary of AHK with testing version and let you know if I have any problems.

Each new thread has a period of uninterruptibility, so Critical N might not be necessary if it will always complete quickly (within 15ms?).

It happens with scripts with trivial subclassing procedure, without any kind of lengthy processing which finish within less then 1ms. Thats why it was interesting observation to notice Critical 5000 makes the difference.

Adding critical in wndproc in many cases makes script totally unusable even tho only wndproc thread is affected by it.

How so?

You can't interact with GUI at all, from the moment it is launched.

Cheers.

[majkinetor]

First impressions:

For now, I never managed to break MRS script which always break in certain circumstances randomly when I return latest AHK. (drag & drop files, menu showing over LV, scrolling, col resizing etc...)

I didn't find any changes in behavior in other scripts so far which extensively use sub-classing.

This is definitely an improvement. Time will tell if that is what was needed for sub-classing to work correctly in all circumstances. I will continue testing during next month.