| |
| 
|
|
Version 1.80 Known bugs. Ok, Version 1.80 beta is a bit buggy but I froze the code anyways. Even though there are a few bugs most are quite easily to avoid. I ran out of room long long ago and had to use some very agressive space optimization techniques to get it to fit into $1000 asembler statements. This renders the code a bit hard to modify, when you need to find space for each modification. I think the next step is to run the code through a custom spagetitti optimizer. Freezing the code allows me to test against a fixed target, and create a list of bugs that can be fixed all at once in a more controlled fashoned. (BUG) Lcd port is buggy, it needs some pause statements to work right in this version. The earlier 1.0 version worked much better. (REASON) A fifo was added in V1.8 to improve execution speed, unfortunately it can overflow. (BUG) If you try to list a long basic program, or disasembly, a watchdog reset occurs. (REASON) dog needs to be touched more often. (BUG) It seems that some valid basic statements do not asemble when there structure ends in a sufix like 'return' 'CRLF' or 'SPACE'. If you use a contraction such as 'ret' , 'crl' , 'sp' , 'spa' , or 'spac' it works fine. If you add a ';' at the end you are ok too. But if you list to store a program to disk and try to ^C the program back, you need to look for this error. (REASON) pookies. (BUG) If you perform a math op using a left or right shift by zero, the result is 0 or -1 not the identity. (REASON) Shift by 0 actually does a shift by 256. (BUG) If you ^C break out of a disassembly, the program may get stuck until a reset. (REASON) TBD, It may be an overflow of the hardware stack (not the basic stack). (BUG) The divide does not give correct results if the divisor is greater than 156. (REASON) TBD. (WORKAROUND) none. (BUG) If the CONFIGURATOR LCD_MIRROR bit is set the serial port stream/ fifo may get corrupted. (REASON) TBD. (WORKAROUND) none. (BUG) The multiply upper 8 bits MSB is not properly sign extended if the first variable is negative. (REASON) Due to the Multiply's optimization technique. (WORKAROUND) Only use an unsigned first variable if you are using a .* or if you need to use MSB later. (NOTE) A normal multiply works fine on signed data. This will not be fixed. (BUG) The values 0, 10, 11, 12, 13 can not be printed because they are interpreted as _CONTROL_CHARACTERS_. If you are creating a custom character, you may need to use these values. (FIX) The workaround is to set an unused bit, EG. $00 becomes $80. This will not be fixed. |