RPL32

[David Hayden]

I wanted to give an update on RPL32 - a 32-bit sysRPL implementation in C++. See this thread on the old forum for some more info.

I solved the problem of garbage collection with C pointers by using handles as Marcus suggested. To help ensure I got all the cases, I built a torture test into the test harness. When run in this mode, every call that can result in moving memory causes all memory to move to a new block and the old block is "poisoned" by overwriting it with a pattern. Pointers that aren't handled thus end up pointing to garbage and the program quickly crashes.

I'm currently working on Extended real numbers. After that I can do complex numbers and units. The plan is still to use binary math and "close enough" calculations at this point. I'm paying close attention to the error conditions though. My thinking here is that getting the math right is very hard and if I concentrate on that, the rest of RPL will never happen. Better to try to get the OS working first. If it gets to that point, the math can be rewritten.

I will probably delay GROBs also in favor of some sort of text-based implementation first.

Speaking of GROBs, if anyone has thoughts on how to handle color grobs in RPL, I'm all ears. Hopefully there's a better way than the existing method for gray scale.

I've submitted an article about this to Datafile. I'll post the article here after datafile has been out for a while.

Dave