MMIX Bug Report Umlaut in Strings and Character Constants
Content
MMIXware Version
mmix-20110831Bug Reported
Initial: 24/4/2013Author
Martin RuckertDescription
mmixal considers character codes greater than 127 in strings and character constants as too big to fit in one byte.Details
The following MMIX programLOC #100 GREG @ a BYTE "pâté",10,0 b BYTE 'p','â','t','é',10,0 c BYTE #70,#E2,#74,#E9,10,0 Main LDA $255,a TRAP 0,Fputs,StdOut LDA $255,b TRAP 0,Fputs,StdOut LDA $255,c TRAP 0,Fputs,StdOut ADD $255,$255,'â' TRAP 0,Halt,0produces the following warnings:
"byte.mms", line 3 warning: constant doesn't fit in one byte "byte.mms", line 3 warning: constant doesn't fit in one byte "byte.mms", line 4 warning: constant doesn't fit in one byte "byte.mms", line 4 warning: constant doesn't fit in one byte "byte.mms", line 13 warning: Z field doesn't fit in one byteThe documentation to mmixal, however, states:
"In the present implementation a character constant will always be at most 255, since wyde character input is not supported." The problem is caused by the assignment
acc.h=0, acc.l=*p;in the sections "Scan a character constant" and "Scan a string constant". Where p points to a char, which might be signed or unsigned depending on the compiler. In case the compiler uses signed characters, the sign extension will fill the low TETRA of acc, making the value too big to fit in one byte.
Proposed Patch
In mmixal.w the lineacc.h=0, acc.l=*p;is replaced by
acc.h=0, acc.l=(unsigned char)*p;in both sections "Scan a character constant" and "Scan a string constant".
