optimized string compares

compiler/res/prog8lib/prog8_lib.asm

@@ -248,31 +248,25 @@ _arg_s2		.word  0
 
 strcmp_mem	.proc
 		; --   compares strings in s1 (AY) and s2 (P8ZP_SCRATCH_W2).
-		;      Returns -1,0,1 in A, depeding on the ordering. Clobbers Y.
+		;      Returns -1,0,1 in A, depending on the ordering. Clobbers Y.
 		sta  P8ZP_SCRATCH_W1
 		sty  P8ZP_SCRATCH_W1+1
 		ldy  #0
-_loop		lda  (P8ZP_SCRATCH_W1),y
-		bne  +
-		lda  (P8ZP_SCRATCH_W2),y
-		bne  _return_minusone
-		beq  _return
-+		cmp  (P8ZP_SCRATCH_W2),y
-		bcc  _return_minusone
-		bne  _return_one
-		inc  P8ZP_SCRATCH_W1
-		bne  +
-		inc  P8ZP_SCRATCH_W1+1
-+		inc  P8ZP_SCRATCH_W2
-		bne  _loop
-		inc  P8ZP_SCRATCH_W2+1
-		bne  _loop
-_return_one
+_loop           lda  (P8ZP_SCRATCH_W1),y
+		beq  _c1_zero
+		cmp  (P8ZP_SCRATCH_W2),y
+		beq  _equal
+		bmi  _less
 		lda  #1
-_return		rts
-_return_minusone
-		lda  #-1
 		rts
+_less           lda  #-1
+		rts
+_equal          iny
+		bne  _loop
+_c1_zero        lda  (P8ZP_SCRATCH_W2),y
+		beq  +
+		lda  #-1
++		rts
 		.pend
 
 

compiler/res/prog8lib/string.p8

@@ -230,9 +230,9 @@ _str    .word 0
 
     asmsub compare(uword string1 @R0, uword string2 @AY) clobbers(Y) -> byte @A {
         ; Compares two strings for sorting.
-        ; Returns -1 (255), 0 or 1 depending on wether string1 sorts before, equal or after string2.
+        ; Returns -1 (255), 0 or 1, meaning: string1 sorts before, equal or after string2.
         ; Note that you can also directly compare strings and string values with eachother using
-        ; comparison operators ==, < etcetera (it will use strcmp for you under water automatically).
+        ; comparison operators ==, < etcetera (this will use strcmp automatically).
         %asm {{
 		sta  P8ZP_SCRATCH_W2
 		sty  P8ZP_SCRATCH_W2+1

compiler/res/prog8lib/virtual/string.p8

@@ -112,9 +112,9 @@ string {
 
     sub compare(str st1, str st2) -> byte {
         ; Compares two strings for sorting.
-        ; Returns -1 (255), 0 or 1 depending on wether string1 sorts before, equal or after string2.
+        ; Returns -1 (255), 0 or 1, meaning: string1 sorts before, equal or after string2.
         ; Note that you can also directly compare strings and string values with eachother using
-        ; comparison operators ==, < etcetera (it will use strcmp for you under water automatically).
+        ; comparison operators ==, < etcetera (this will use strcmp automatically).
         %ir {{
             loadm.w r65534,string.compare.st1
             loadm.w r65535,string.compare.st2

docs/source/todo.rst

@@ -1,6 +1,8 @@
 TODO
 ====
 
+Move vectors such as USRADD in cx16 to cbm block?
+
 See open issues on github.
 
 Re-generate the skeletons doc files.
@@ -38,6 +40,7 @@ Compiler:
 - do we need (array)variable alignment tag instead of block alignment tag? You want to align the data, not the code in the block?
 - ir: related to the one above: block alignment doesn't translate well to variables in the block (the actual stuff that needs to be aligned in memory)  but: need variable alignment tag instead of block alignment tag, really
 - ir: fix call() return value handling
+- ir: add specialized bit test instructions to support "prog8_ifelse_bittest_xxx" see the check in IRCodeGen
 - ir: proper code gen for the CALLI instruction and that it (optionally) returns a word value that needs to be assigned to a reg
 - ir: idea: (but LLVM IR simply keeps the variables, so not a good idea then?...): replace all scalar variables by an allocated register. Keep a table of the variable to register mapping (including the datatype)
   global initialization values are simply a list of LOAD instructions.

examples/test.p8

@@ -5,42 +5,60 @@
 
 
 main {
+    str large1 = "the quick brown fox jumps over the lazy dog. the quick brown fox jumps over the lazy dog. the quick brown fox jumps over the lazy dog. the quick brown fox jumps over the lazy dog."
+    str large2 = "the quick brown fox jumps over the lazy dog. the quick brown fox jumps over the lazy dog. the quick brown fox jumps over the lazy dog. the quick brown fox jumps over the laxx doggo doggo."
+
     sub start() {
-        str name = "zn.iff.jpg"
+        txt.nl()
+        check("", "", 0)
+        check("", "a", -1)
+        check("a", "", 1)
+        check("a", "a", 0)
+        check("a", "z", -1)
+        check("z", "a", 1)
+        check("irmen", "irmen", 0)
+        check("irmen", "irmen2", -1)
+        check("irmen2", "irmen", 1)
+        check("irmen", "irxen", -1)
+        check("irmen", "irman", 1)
+        txt.nl()
 
-        ubyte index
-        bool found
+        bench()     ; orig: 88   (pet: 713)      optimized:   56  451
+        bench2()    ; orig: 131  (pet: 1066)     optimized:   83  674
+    }
 
-        index, found = string.find(name, '.')
-        if found {
-            txt.print_ub(index)
-            txt.nl()
-        } else {
-            txt.print(". not found\n")
+    sub bench2() {
+        cbm.SETTIM(0,0,0)
+        repeat 1000 {
+            bool compare = large1 != large2
+            cx16.r0L++
+            compare = large1 > large2
+            cx16.r0L++
+            compare = large1 <= large2
         }
+        txt.print_uw(cbm.RDTIM16())
+        txt.nl()
+    }
 
-        index, found = string.find(name, '@')
-        if found {
-            txt.print_ub(index)
-            txt.nl()
-        } else {
-            txt.print("@ not found\n")
+    sub bench() {
+        cbm.SETTIM(0,0,0)
+        repeat 2000 {
+            void string.compare(large1,large2)
         }
+        txt.print_uw(cbm.RDTIM16())
+        txt.nl()
+    }
 
-        index, found = string.rfind(name, '.')
-        if found {
-            txt.print_ub(index)
+    sub check(str s1, str s2, byte expected) {
+        byte result = string.compare(s1, s2)
+        txt.print(s1)
+        txt.print(" & ")
+        txt.print(s2)
+        txt.print(": ")
+        txt.print_b(result)
+        if result!=expected
+            txt.print("  !wrong!\n")
+        else
             txt.nl()
-        } else {
-            txt.print(". not r found\n")
-        }
-
-        index, found = string.rfind(name, '@')
-        if found {
-            txt.print_ub(index)
-            txt.nl()
-        } else {
-            txt.print("@ not r found\n")
-        }
     }
 }