Journal entries from August 2022¶
What if?¶
Choose your path.
Our assembler gives us a lot of convenience for testing features of our VM. So let us start doing interesting stuff with it. We do have support for jumps already, but as it is now, save of an endless loop, there is absolutely no reason to do it, yet. All our programs run their predetermined way. If you look again at label.lva, you can see that none of those gotos introduce any dynamic. We could just ditch them and reorder the rest. It would do the same, only more efficient. They simple tangle up our linear code, without removing its linearity.
You labeled me, I'll label you¶
We add a feature to our assembler that we overlooked before.
Over the last few entries we created ourselves a really useful little assembler program. I hope you played around with it and enjoyed not having to write bytecode directly. If you did, you should have noticed that I left out a really important detail. Remember when I was complaining about how bad writing bytecode is? And that it got even worth, when we introduced jumps? Yeah, I did not solve that problem at all. If anything, I made it worse, because you still have to count the relative bytes to your destination, but you do not see those bytes any longer. You just have to know, how many bytes each instruction will produce.
Running assembler programs¶
We will extend our assembler to do something useful, finally: execute our programs on lovem.
We have created ourselves an assembler in ~300 lines of code. And it has a command line interface, an API to be used in a program, and even useful error reporting. That is cool! But what do we do with the bytecode? It just dumps them to the console. That is not very useful. We could copy/paste that into one of our example binaries... This is not what we wanted. So let us enhance our assembler.
Assembling bytes¶
Our new assembler is almost done assembling. Over the last entries we learned how the program parses the assembly sourcecode and produces a list of parsed instructions. What we now need to do, is turn that into bytes.
Handling instructions¶
We took care of all the dirty work inside the assembler during the previous posts. We now have a cleanly parsed instruction with an optional argument that we can evaluate. Let us dive into parse_instruction():