Does anyone use the mame source code drivers memory map and hardware info to help with writing the scripts?
No better resource that I know of. There are some gotchas in MAME but I haven't run into any problems yet.
Do you use RKSIC.EXE or 9LC.EXE to compile it? or either? What are the pros and cons of each one? (I'm trying to work out what I can get away with with comments and script format)
I use RKSIC.EXE, since it is has the transfer program too & I like some of the features.
I see a SETUP MESSAGES code called SET-EXERCISE ERRORS? on the reference card that has this description: "Determines whether error messages and prompts for looping on errors are displayed"
You could also use REG8 or REG9 to keep user entered script controls and counts: a) Have a user input yes/no to a "stop on errors" prompt at the start of the script, or is the setup code only allowed at the script start?
I don't know if you can do that or not? I belive that Set-Exercise errors is just a prompt to loop. I don't think you can supress error messages.
b) Count the number of errors? and display it at the end of the tests if non-zero.
If there were 8 tests, you could store a bit pattern in the register,
and
then report which tests failed at the end based on which bits were on. (if this thing will do the appropriate math on the register?)
Troubleshooter Volume 2 Number 1 2nd page has the 232 dongle & code approach.
The other problem I've noticed when coding I've strung a bunch of RAMs together & check each pair (2114s on Pole), but when it gives you an error message it just says RAM error at $4401, so now you've got to go back to the memory map & figure out which ram it is. Be nicer if it said VIDEO RAM 7F 8F BAD.
Isn't it possible to split the ram tests up into memory mapped pieces that match the chips or chip pairs?
Yes I generally do that, ROMs are checksumed by chip, RAM are checked by chip or block (2x2114s).
If there's a fault could it branch to a sub-test program that poked 0x00 and 0xFF into the failed address and reported the faulty read-back result? This would show which of the 4bit 2114's may be at fault.
The error message usually tells you what failure you have.
If the ram (or rom) test failed couldn't the script go to a sub-test that just poked that address only, continuously, so you could find the chip enable on the faulty ram/rom ? (or is that the looping function anyway :-) )
Yes you can do looping for that.