nesdev.com

Loopy has just posted an update to asm6 that incorporates several improvements I made:

• PowerPC (big-endian) and 64-bit compatible.
• Includes Dwedit's error-overload prevention code.
• Supports quiet mode with -q, which prints nothing unless there's an error. Useful for scripts.
• Returns non-zero exit status unless there are NO errors AND output file was successfully generated. This allows scripts to catch any errors, rather than thinking they have a valid file.
• It's now an error to include extra arguments, rather than just warning.
• All file writing is now checked for errors, including the final fclose().
• Generates fewer compiler warnings, and avoids non-portable things like (char*)1.

If you see any problems or differences in output, please post here so I can look into it. If possible, try assembling your project with this version and the previous, and be sure the output files are exactly the same.

I found a bug in ASM6. After adding a "JSR somelabel" to a label which appears later in the source file, some stuff involving the current address $ breaks. I would add ".org $" as a workaround, and that worked.

Edit:

So I'm using a series of INCBINS to make a patch.
When I have a construct like this:

I get an error in the last incbin line. The error goes away if I put ".org $" after the jsr SomeLabel.

_________________
Here come the fortune cookies! Here come the fortune cookies! They're wearing paper hats!

OK, I think I might have found the cause. It seems that the boolean dependant is set to true by eval() when the expression involves a label not yet defined. And apparently, eval() doesn't clear dependant otherwise, instead relying on the caller to do so. So sprinkled throughout the source are dependant=0 lines. I believe that there should be one of those in incbin(). I inserted one after seekpos=0 and it appears to work now. Doing .org $ has the side-effect of clearing dependant, and thus making incbin() work as well. Might be a good idea for me to add an extra fresh_eval() or something that clears dependant first, and then use that wherever dependant is cleared first.

BTW, this bug was in the previous version as well. I guess I was most interested in any bugs I introduced into the new version, but I might as well address others as well...

I'm surprised you were able to fix something in the jumbled spaghetti mess which is ASM6.

_________________
Here come the fortune cookies! Here come the fortune cookies! They're wearing paper hats!