Scheme Programming

I've finally got source code references for Picobit byte code. It's proved enormously frustrating. My first approach attached the original syntax references to the structs used in the AST. After I'd done all this, I found only 2/3 of the tests passed.

I made what seemed like a reasonable assumption: the front end uses lots of incomplete pattern matching on the structs. I'd tried to update all the patterns, but perhaps I missed some. I went back and replaced the front end approach with a dictionary from AST nodes to syntax.

The result was exactly the same!

This was worthwhile though: the pretty printing of the AST now matches in the original and new versions, so I can capture that and prove the front and is unchanged. I have a script which captures the AST and the assembler output for each test file. Now I can just run meld with the two output directories as arguments...

There's one instruction I'm just not emitting: goto-if-false. Turns out that's just because I'd factored out a method to deal with re-emitting those instructions, but it actually just didn't emit the new instruction. After that, a branch reduction optimisation needed fixing. Down to one unexplained error now.

Racket points out when pattern matches for structs have the wrong number of arguments, so my first change was perfectly safe. It was also better in an important way: it let me improve error reporting across the board: given an AST node I could report file and position. In the end I've gone with the first approach. I'll hand code the pretty printing of the AST when I next find I need it.

Most of the rest of my pain comes from the fact that Racket stack traces refer to the start of the top level closure, rather than the call site, which is pretty crap compared to what I'm used to (which is the JVM or CLR). The more nested closures the greater the frustration. Anyway, that's made me think I'd better have a rotating buffer of the the last N call sites in the picobit interpreter.

I'm really yet to work out if I can work around this using the Racket debugger.

The branch optimisation code I mentioned above used cadar which means the second item in the first list! When I encounter code like this it reduces my surprise that few people stick with scheme. I'm beginning to think if I get much further with this I'm going to invent a new language with a syntax for index refs.

I did all this to have some sort of debugging capability, but when I finished all my work on the compiler, the program I wanted to debug didn't crash any more, so I guess I inadvertently fixed a compiler bug. Certainly there are a lot more contracts and guards in the compiler, and I've just blindly fixed anything they complain about, but I can't see a fix in the diffs. I guess I'll have to go back and check.

Popular posts from this blog

3D Printer ramblings

 I have a Prusa i3 mk2.5 in storage. I used to use it in my shed I used not to worry about it's harmful emissions: I just didn't share the space with it. I had, however, added an enclosure, and was planning to add a HEPA filter too it. Once it's in an enclosure, it's massive. Even if I had added the filter, there's no where to put it. I'd like a small printer that has an enclosure or is easy to enclose. I figure I could get by with 150x150x150 build area: I was rarely patient enough to wait for my i3 to print bigger models. 200x200x200 would probably actually replace my current printer, so that would be a bonus. Not many printers have HEPA filters, which is really a shame. It seems just to be the result of a misconception about HEPA filters, and about filtration in general. Read this . E.g. I was looking at a used Lulzbot mini, and there are ready made enclosure s for it, however . In this particular case, it's no big deal: opportunities to buy a Lulzbot min
Read more

Balance Bot V2

Image
My current balance bot sort of works, but it's surprising it does: it's centre of gravity is actually only about 60mm above the motor axis. The whole thing ways 320g. The 6 AAA batteries only weigh 70g, so despite being right at the top, the don't affect the centre of gravity much. 6 AA batteries would weigh about 180g. That would add a 34% to the overall weight, and move the centre of gravity much higher. Seems like a dramatic increase in weight: it would be nice to be a bit more incremental. More AAA batteries isn't really worth considering: they only weight 12g, vs 30g for an AA. 4 AA would deliver much less power/torque, so that's no good. There's no middle ground, in other words. I'm also going to make the whole thing a bit taller: I need more space to add more sensors in any case. Last time I made the frame, I needed an elaborate jig to get it square and it actually didn't work that well. This time, I decided to use some short bits of a
Read more

Robot arm

Image
I 3d printed this robot arm I original found on Thingiverse, but the Instructable is recommended. There's a MK2, and also a delta style arm, both of which look like good alternatives, but this will do for now. Here's a video of it moving: I've attached it to an STM32F4 Nucleo board, via a prototype shield: In the video, I'm sending commands from a python script that runs on my desktop, communicating via USB serial. I've written an implementation of canonical s-expressions in c++ that doesn't require memory allocation, as well as a python version to talk to it. It's the main reason I haven't posted for so long, and of course overkill for just controlling some servos, but should prove more generally useful. Here's the code . It uses platformio for the build, and mbed rtos.
Read more