regex: enable safe handling of invalid UTF-8 by default

[vtjnash]

No description provided.

[JeffBezanson]

What is this for?

[vtjnash]

Seemed like someone wanted them in a list, so I kept them here

[JeffBezanson]

Any performance cost?

[vtjnash]

We have some really good @nanosoldier runbenchmarks("shootout" || "problem", vs=":master") for those 😬

[nanosoldier]

Something went wrong when running your job:

ProcessExitedException(2)

cc @christopher-dG

[vtjnash]

@nanosoldier runbenchmarks("shootout" || "problem", vs=":master")

[nanosoldier]

Your benchmark job has completed - possible performance regressions were detected. A full report can be found here. cc @christopher-dG

[vtjnash]

Seems safe to say this is well within the noise, as the benchmarks that include a regex explicitly (as opposed to using them from somewhere else, like Base.Filesystem) are:

src/problem/IMDBGraphs.jl
src/problem/JSONParse.jl
src/problem/SpellCheck.jl
src/shootout/k_nucleotide.jl
src/shootout/regex_dna.jl
src/shootout/revcomp.jl

[vtjnash]

Will merge soon, unless any objections

[StefanKarpinski]

I object.

[vtjnash]

Ah, sorry, I assumed you were in favor of this from past discussions. What's your objection?

[StefanKarpinski]

This is an improvement but my understanding is that this treats \x80 in a regex literal as U+80 whereas I think it should either be an error or match the byte 0x80.

[vtjnash]

No, you're thinking of the other issue. This doesn't change the match string interpretation. This simply sets the flag to disable UB in pcre2 when the searched string isn't valid utf8.