https://bugs.kde.org/show_bug.cgi?id=488828
--- Comment #8 from Blazer Silving <breakingsp...@gmail.com> --- Some notes after modifying values and testing with kdesrc-build: -- Sizing/Padding Adjusting m_relativeMargin variables in expolayout.h works to help window scaling. Less margins = larger window previews. The default of 0.07 on all sides are far too large, even 0.02 has too much space. I think a value of 0.01 looks great, while I'm sure many users will think I am wrong and prefer the default or even larger gaps. User customization would shine here. -- Positioning The other variables defined in expolayout.h don't modify the basic row/column behavior of the grid layout. Had hoped there would be a value to determine the clinginess of a window to it's nearest cell, but it is not possible to achieve anything like a radial layout using this new code, so that wouldn't make any difference. Quoting the reference paper: https://invent.kde.org/plasma/kwin/-/issues/189 In LayoutClosest the algorithm creates a rectangular grid of slots, and assign windows to those slots. In LayoutNatural, the algorithm incrementally adjusts the windows, scaling or moving them, based on nearby available space and overlaps. The current system is akin to LayoutClosest, with dynamic column/rows generated, where windows will travel over others to reach their target slot. Wondering again if it could be possible to generate conditional layouts based on position/overlap, which would of course bring it closer to the previous Natural behavior. -- I've patched my system Kwin package to revert this entire change in the meantime, while I test in a local profile. A non-technical question that I'm genuinely curious about, enough to bother asking: What is the rationale behind deciding not to offer any user customization with this radically changed layout? The bullet points I've read in the general design docs concern overwhelming users with options and potential misconfiguration, those simply don't make sense in this context. I would love to contribute and add small options to the KCM module to improve this new layout (like a slider or integer for Padding size mapped to m_relativeMargin), but I don't want to waste time if they aren't even going to be considered by design. -- You are receiving this mail because: You are watching all bug changes.
