#adqe

[adorukiescreve]

Outra poesia nasceu

Sempre (no mínimo)

são dois os que parem a poesia:

Aquele que deseja morrer,

eu lírico;

E aquele que há muito está morto,

o poeta.

[adorukiescreve]

Estética

Estética

sem

métrica Sem

ideia

Vazio

[damonmorgue]

Designing an ASM-like scripting language, Part 2

Obligatory gloating and selecting operation by argument count

As it turns out, forcing brackets was a great idea. The initial tutorial architecture was woefully inadequate, and not a great idea. I think it was meant to be party to both C and C++, which explains some of the design choices. Either way, a little bit had to be refactored with the introduction of the argument stack - curmudgeonly pointer interfaces were the most prominent.

One thing that hit me on the drive home yesterday, and a suspicion which was confirmed today, is that the recursive parser needs a queue rather than a stack. ctrl+f saved the day, thankfully the STL is fairly homogenous and I only had to replace calls to top() with front().

I ran the first operation with a right-most nesting, and after a minor correction with a string index it ran happily. I then tried an operation with a nested first argument, which didn't work. It didn't seem it would be hard to correct though, and this too was a well-founded inquiry. All I had to do was increment that aforementioned string index by 1 after returning from the recursion.

See for yourself!

Well that's the gloating part. What now? Argument-dependent operation types, which I briefly mentioned in the first part of this series. This might be a little bit more tricky. This pertains to the TMR and REG commands - with two arguments, it is considered a write operation and does not return anything. With one argument, it is a read operation and passes it's return down the queue.

So that's today's mission, which after explanation doesn't seem so complex after all.