bmx: allow palette to be loaded into memory buffer instead of vram

compiler/res/prog8lib/cx16/bmx.p8

@@ -1,7 +1,6 @@
 ; Routines to load and save "BMX" files (commander X16 bitmap format) Version 1.
 ; Only uncompressed images are supported for now.
 ; BMX Specification: https://cx16forum.com/forum/viewtopic.php?t=6945
-; TODO: make read_palette() and write_palette() use an optional palette_buffer_ptr to store palette into system ram instead of directly into vram.
 
 %import diskio
 
@@ -19,9 +18,10 @@ bmx {
     ubyte palette_start
     ubyte compression
 
-    uword error_message         ; pointer to error message, or 0 if all ok
+    uword error_message             ; pointer to error message, or 0 if all ok
-    uword max_width = 0         ; should you want load() to check for this
+    uword max_width = 0             ; should you want load() to check for this
-    uword max_height = 0        ; should you want load() to check for this
+    uword max_height = 0            ; should you want load() to check for this
+    uword palette_buffer_ptr = 0    ; should you want to load the palette into main memory instead of directly into vram
 
     sub load(ubyte drivenumber, str filename, ubyte vbank, uword vaddr, uword screen_width) -> bool {
         ; Loads a BMX bitmap image and palette into vram. (and Header info into the bmx.* variables)
@@ -189,11 +189,21 @@ save_end:
 
     sub read_palette() -> bool {
         ; load palette data from the currently active input file
+        ; if palette_buffer_ptr is not 0, the palette data is read into that memory buffer,
+        ; otherwise it is read directly into the palette in vram.
         cx16.vaddr(1, $fa00+palette_start*2, 0, 1)
+        cx16.r1 = palette_buffer_ptr
         cx16.r0L = palette_entries
         do {
-            cx16.VERA_DATA0 = cbm.CHRIN()
+            cx16.r2L = cbm.CHRIN()
-            cx16.VERA_DATA0 = cbm.CHRIN()
+            cx16.r2H = cbm.CHRIN()
+            if cx16.r1 {
+                pokew(cx16.r1, cx16.r2)             ; into memory
+                cx16.r1+=2
+            } else {
+                cx16.VERA_DATA0 = cx16.r2L          ; into vram
+                cx16.VERA_DATA0 = cx16.r2H
+            }
             cx16.r0L--
         } until cx16.r0L==0
         return not cbm.READST()

docs/source/todo.rst

@@ -2,6 +2,8 @@
 TODO
 ====
 
+- optimize 6502 codegen: "for 0 to end" into repeat loops. (only for ubyte and uword, 6502 cpu)
+
 - [on branch: shortcircuit] investigate McCarthy evaluation again? this may also reduce code size perhaps for things like if a>4 or a<2 ....
 
 ...

examples/cx16/showbmx.p8

@@ -1,8 +1,12 @@
-
+; Viewer program for BMX image files.
-; viewer program for BMX image files.
+; This program shows *one* way to do it, by checking the header upfront,
-; see https://cx16forum.com/forum/viewtopic.php?t=6945
+; and loading the palette into system ram first. The simplest way to load
+; a BMX file is to just read everything into vram directly using a single bmx.load() call.
+;
+; BMX file format: see https://cx16forum.com/forum/viewtopic.php?t=6945
 
 %import textio
+%import palette
 %import bmx
 %option no_sysinit
 %zeropage basicsafe
@@ -15,20 +19,29 @@ main {
             txt.print("\nenter bmx image filename: ")
             if txt.input_chars(&filename) {
                 if bmx.load_header(8, filename) {
-                    txt.print("\nwidth: ")
+                    txt.print("\nsize: ")
                     txt.print_uw(bmx.width)
-                    txt.print("\nheight: ")
+                    txt.chrout('*')
                     txt.print_uw(bmx.height)
-                    txt.print("\nbpp: ")
+                    txt.print("  bpp: ")
                     txt.print_uw(bmx.bitsperpixel)
                     txt.nl()
                     sys.wait(100)
 
+                    ; tell the loader to put the palette into system memory instead
+                    ; also make palette black at first, to hide loading (even though it is very fast)
+                    ; (you could do a soft fade-in effect with this for instance)
+                    bmx.palette_buffer_ptr = memory("palette", 512, 0)
+                    sys.memset(bmx.palette_buffer_ptr, 512, 0)
+                    palette.set_rgb(bmx.palette_buffer_ptr, 256)
+
                     ; switch to correct screen mode and color depth
                     void cx16.screen_mode($80, false)
                     cx16.VERA_L0_CONFIG = cx16.VERA_L0_CONFIG & %11111100 | bmx.vera_colordepth
-                    ; actually load
+
+                    ; now load the image
                     if bmx.load(8, filename, 0, 0, 320) {
+                        activate_palette()
                         void txt.waitkey()
                     }
                 }
@@ -44,4 +57,17 @@ main {
             }
         }
     }
+
+    sub activate_palette() {
+        ; copies the pallette data from the memory buffer into vram
+        cx16.r1 = bmx.palette_buffer_ptr
+        cx16.r2L = bmx.palette_start
+        cx16.r3L = bmx.palette_entries
+        do {
+            palette.set_color(cx16.r2L, peekw(cx16.r1))
+            cx16.r1+=2
+            cx16.r2L++
+            cx16.r3L--
+        } until cx16.r3L==0
+    }
 }

examples/test.p8

@@ -2,61 +2,79 @@
 %zeropage basicsafe
 
 main {
+    ubyte counter
+    uword wcounter
+    ubyte end=10
+    uword wend=10
+
     sub start() {
+        cx16.r0=0
+        forloops()
+        txt.print_uw(cx16.r0)
+        txt.nl()
 
-        txt.print("for:\n")
+        cx16.r0=0
-        for cx16.r0L in 10 to 20 {
+        untilloops()
-            txt.print_ub(cx16.r0L)
+        txt.print_uw(cx16.r0)
-            txt.print(" before...")
+        txt.nl()
-            if cx16.r0L > 15
+    }
-                break
+
-            if cx16.r0L ==14
+    sub forloops() {
-                continue
+        end=10
-            txt.print("after\n")
+        for counter in 0 to end {
+            cx16.r0++
         }
-        txt.nl()
+        end=0
-
+        for counter in 0 to end {
-        txt.print("repeat:\n")
+            cx16.r0++
-        cx16.r0L=10
-        repeat 10 {
-            cx16.r0L++
-            txt.print_ub(cx16.r0L)
-            txt.print(" before...")
-            if cx16.r0L > 15
-                break
-            if cx16.r0L ==14
-                continue
-            txt.print("after\n")
         }
-        txt.nl()
+        end=255
-
+        for counter in 0 to end {
-        txt.print("while:\n")
+            cx16.r0++
-        cx16.r0L=10
-        while cx16.r0L<20 {
-            cx16.r0L++
-            txt.print_ub(cx16.r0L)
-            txt.print(" before...")
-            if cx16.r0L > 15
-                break
-            if cx16.r0L ==14
-                continue
-            txt.print("after\n")
         }
-        txt.nl()
 
-        txt.print("until:\n")
+        wend=1000
-        cx16.r0L=10
+        for wcounter in 0 to wend {
-        do {
+            cx16.r0++
-            cx16.r0L++
+        }
-            txt.print_ub(cx16.r0L)
+    }
-            txt.print(" before...")
+
-            if cx16.r0L > 15
+    sub untilloops() {
+
+        end=10
+        counter = 0
+        repeat {
+            cx16.r0++
+            if counter==end
                 break
-            if cx16.r0L ==14
+            counter++
-                continue
+        }
-            txt.print("after\n")
-        } until cx16.r0L>20
-        txt.nl()
 
+        end=0
+        counter = 0
+        repeat {
+            cx16.r0++
+            if counter==end
+                break
+            counter++
+        }
+
+        counter = 0
+        end=255
+        repeat {
+            cx16.r0++
+            if counter==end
+                break
+            counter++
+        }
+
+        wcounter = 0
+        wend=1000
+        repeat {
+            cx16.r0++
+            if wcounter==wend
+                break
+            wcounter++
+        }
     }
 }