Wimp window flags bit 10 - GCOL colours
Chris Dewhurst (1709) 165 posts |
https://www.riscosopen.org/wiki/documentation/show/Window%20Flags Bit 10 – window colours as GCOL numbers 0-254 Does ‘colours’ refer to the colours in the window block +32 to +38 ? I can’t seem to get it to work, e.g. set window flags bit 10 and set window block + 35 to 129. Should give dark red background but actually shows up as grey, as though the system is only picking up the low nybble (standard Wimp palette). Am I doing something wrong here or is bit 10 not actually implemented ? thanks |
Rick Murray (539) 13806 posts |
in a 256-colour mode an untranslated colour is given as %cccccctt, i.e. bits 0-1 give bits 6-7 of the TINT and bits 2-7 give bits 0-5 of the GCOL. That’s a 256 colour mode, mind you. Does it do the same thing in 16M colour modes? |
Chris Dewhurst (1709) 165 posts |
Setting bit 10 of window flags seems to be ignored (or at least low nybbles of the colours taken as standard Wimp colours) by the system in 256, 64K and 16M colour desktops on Raspberry Pi RISC OS5.28. On an earlier RISC OS 3.19 (admittedly under emulation) dektop mode 15 (256-colours): setting bit 10 of the window flags seems to result in the system using bits 0-1 of the colours. So only black and dark shades of red possible. I wonder what’s going on. |
Rick Murray (539) 13806 posts |
It looks like true colour support might have broken this. window_bg ALTENTRY MOV R3,#&80 MOV R4,#&00 ; GCOL action 0 [ TrueIcon3 B settruecolour | LDR R14,[handle,#w_flags] TST R14,#wf_realcolours ; need translation? BEQ settranslate ; call suitable routine ; fall through ] https://gitlab.riscosopen.org/RiscOS/Sources/Desktop/Wimp/-/blob/master/s/Wimp04#L4379 |
Rick Murray (539) 13806 posts |
This change was introduced when the Ursula branch was brought in, but there doesn’t appear to be any mention of this in the various bits of documentation and/or the BlackLog, the other log, or the change history. Should this be filed as a bug? |
Steve Pampling (1551) 8155 posts |
RO3.8, so theoretically RO4.02 ought to have the same bug or a fix listed in it’s changelog? |