As R7 has dragged out to much more of a prolonged affair than I ever intended, motivation has been steadily leaking away. There are times when it's hard to sit down and look at the code because I've spent so much time crawling towards the release. At the same time, though, the tantalizing nearness of completion makes it hard to think about anything else.
Epoch is unquestionably both the most difficult and most enjoyable project I've ever worked on. The only thing that's ever come close was the old Freon 2/7 realtime raytracing/global illumination system that I worked on several years ago. Epoch is a controlled substance. It provides a hell of a rush, a bewildering high, and only a slightly annoying crash at the end. Hacking through the midnight hours on this language is just ridiculously fun.
There's a lot of things I wish I had done differently in my life, and one of the big ones was abandoning the Freon project. Back in those days, I had predicted that within a few years we would see the end of major advances in scanline rasterization technology, and people would start exploring raytracing on the GPU to gain those next few shiny features. My plan at the time was to design a hardware system that could accelerate raytracing and global illumination, and use it alongside existing GPU systems as a sort of video coprocessor card.
I dropped that project for a variety of reasons; but the big one was that I had gone from working contract jobs whenever I felt like it, to holding not just one but two full-time programming jobs. I barely had time in the day to eat and sleep, let alone hack on a major project.
And I've regretted dropping Freon ever since.
So in a very real way, I'm trying to avoid repeating old mistakes. I don't want to burn out on Epoch and then discover in another year that someone else has accomplished what I set out to do. I don't necessarily have to be the first one (I'll settle for being the best one [wink]) but I'm completely committed to getting this language out there. If nothing else I think it will be an important benchmark of what kinds of features and power are needed in the next generation of programming languages.
But first things first. I have to get R7 done before I can get into the giant messy blob of awesome that will be R8. And in order to get R7 done, I have to do all of this jazz:
- Integrate new features with the assembler and bytecode systems
- Pop the parser stack correctly to improve error responses
- Implement support for named lists
- Implement allocators properly, remove hacked code
- Fix some bugs in nested response map support
- Improve syntax for initializing and accessing nested structures
- Type validation with lists
- Stack sanity checks when reducing infix expressions
- Type validation for operate-assign ops (e.g. += and --)
- Extend support for operate-assign ops to all numeric types
- Implement concatenate-and-assign operator ;=
- Improve validation of forked tasks (esp. when tracking the task name)
- Perform complete code review for exception safety, documentation, code cleanliness, error handling robustness, and elimination of hardcoded strings/magic numbers
- Complete code review of all example programs to ensure they use the latest syntax and features
As you can see, although a few things are marked off, there's a few newcomers as well. I have no idea how long I will continue in this pattern, but hopefully I'll start making net forward progress very soon.
Unfortunately for Epoch, my weekend involves having a Real Life, so I won't be able to hack straight through until Monday. I will be doing as much as I can cram into my schedule, though [smile]
