Sukera

One possible test could try to call `do_invoke` with a very large `nargs` so that the OOB access would write across a page boundary and cause a segfault.

> Access is UB, but indexing is generally valid Deref when `i` is OOB is UB. I don't follow what you mean with "access" and "indexing". > and, as a...

> one alternative would be to have codepoint(c) return a value for overlong encodings. In that case we can remove hascodepoint and should probably export ismalformed and document that codepoint...

> The only case I can see where there would be a problem is code that checks for a specific character and then later assumes that character's code point cannot...

> it will mean string1 == string2 is no longer equivalent to all(collect(string1) .== collect(string2)). (It will also slow down character comparisons.) Yeah, that's certainly a bad tradeoff. Let's just...

Yep, also breaks on 1.7-rc1: ``` julia> versioninfo() Julia Version 1.7.0-rc1 Commit 9eade6195e* (2021-09-12 06:45 UTC) Platform Info: OS: Linux (x86_64-linux-gnu) CPU: Intel(R) Core(TM) i7-6600U CPU @ 2.60GHz WORD_SIZE: 64...

Kind of? The MWE by @martinholters at least throws a sensible `StackOverflowError` and not an internal error for me: ``` julia> xs = tuple(("a" for _ in 1:2000)...); julia> foo(xs)...

Kind of tangential, but having full-on Julia source code as the configuration file format seems VERY dangerous, especially if people start sharing these. Would using a TOML be an option...

> TOML would be good, we just don't yet have a parser available at this stage. It will be awkward, since we don't even have arrays or dicts at this...

Failures are real - one consequence of this PR is that e.g. `Base.Iterators` is truthfully printed as just `Iterators`, similar for `Threads`, because the name is exported: ``` julia> Base.isexported(Base,...