Pierre

Are you certain of your benchmark? The benefits seem to small compared to what it should be. Could you also compare with a basic constant unfolding version? Aka: ``` return...

> 3 times performance improvement is small? Well, yes. Compared to 15×2 = 30 dereferences, I was expecting one or two orders of magnitude, not just 1/3 of the time....

Your benchmark has indeed a bias. Since you are calling the function with a fixed value, you don't see the full picture. `check_by_bitwise_test` has a constant execution time, always 100ns...

Hi Do you have a simple archive to download that would contain everything needed to reproduce that issue ? In order to discard the «python is slow» argument, I suggest...

Note : the HAVE_X11 wrapping code is wrong. Qt can be compiled for both X11 and Wayland, thus HAVE_X11=1 while running without X11.

FYI, from a database point of view, this is not a silly optimization : conflict_target will always contain indexed fields, and doing an update on these fields will thus prevent...

Hello @petere Any news regarding merging fixes for the two other issues? To be honest, the total_wait_time was no the one bothering us the most… Regards

I've done a rebase so only the remaining parts are shown here.

This may be related to #391.

I just reported a compilation issue when using MSVC++ under windows. Current master does not compile under windows so far. Could you try an older revision, ``4201c6a`` ? This could...