liam-ilan / crumb Goto Github PK

The following code has O(n^2) complexity (n is the length of in), accidentally quadratic:

Compile the interpreter with no optimizations and try to execute the output of the following Python program:

print("(print\"" + "a" * int(1e6) + "\")")

It will take about 10^12 operations, seconds at the very least. -O2 is typically smart enough to optimize reading strlen() on each iteration, but I would not rely on that.

The following snippet can access array out of bounds:

Consider a string that terminates with ". The program enters the first if statement, increases i, now code[i] is \0. Then it enters the while loop, increases i one more time, now code[i] is out-of-bounds. It does not cause crash currently for me, though. You can also see a hint by passing -fsanitize=address to gcc/clang.

Another issue is with isdigit function that expects a non-negative number, but you give it a char that can be (and typically) is signed:

Currently the project has no open source license, and neither do any contributions in the pull requests. Will this project be open sourced?

Is (3*5)^(1/2) not 3.87 instead of 5.83 like you have shown in your example?

It seems to me that the lexer is ad-hoc with no specific grammar for what it's trying to lex:

In particular, it seems to assume that \" always terminates the string literal. It's not the standard behavior for other programming languages; one can escape the backslash with another backslash, e.g. "\\".

Just as the title says

I only started learning C early this year (late Jan early Feb) and reading through this code I'm impressed by how much I understand obviously from the fact that you wrote such concise and easy to understand code.

I've picked up on a few memory handling tricks along the way that I didn't think of and will actually be rewriting a simple interpreter-ish project I did a few weeks back (shameless plug here) based on some of the practices I've picked up in this codebase

Thank you for sharing this project and writing neat concise code like this. I'm not done reading through everything but it's like a very good book you can't put down. I'm taking notes as I read in case I forget something

Anyway keep being awesome Liam 👍🏾

There seems to be a read-past-buffer issue in the lexer. Let's say that you have a file that ends with /, then here you peek behind the end of the buffer. This pattern is used several times. I did not look closer into the potential effects of this pattern but it might lead to crashes if by chance the lexer continues to find valid patterns, I assume.