No checking for unbalanced '('

[akkartik]

This program waits on a ( until it encounters a ). That in itself is fine, but it interacts poorly with the way it sequences through multiple files on the commandline. If you give it a filename that contains just a (, then it gets silently to the prompt and waits for you to type the rest of an s-expression except the initial (. Feels quite profane, as a Lisper. Files should be syntactically self-contained, and unbalanced parens within a file should cause an immediate error.

(Tinylisp has this problem as well, but it feels like more of an issue here since this program is in most ways quite usable.)

[Robert-van-Engelen]

Not sure how to work around this. I tried tinkering with checking fin in list() to detect EOF inside ( ) but that isn't a reliable condition as it still waits until user input has responsed with something to trigger a fin change. Another way would be to globally count up open ( to then artificially produce the number of ) at EOF to close them or give an error (probably best).

[akkartik]

Yeah, I've been noodling on this as well. The easy way would be that sort of two-pass approach. Have get always read from a buffer, and write to the buffer in advance based on counting parens.

One open question is whether to also consider open strings. That opens a can of worms. Unbalanced parens inside strings are ok. Would probably end up duplicating some code compared to your current high standards. It's amazing (or terrifying 😂) how your load primitive just opens the file and everything else goes through the same flow.