[FIX] fix regex validator for std::filesystem::path

[smehringer]

Resolves #2214

[eseiler]

So the problem was that std::filesystem::path models forward_range and that makes it use the wrong overload because it is more constraint than the option_value_type one?

And then weird stuff happened because it validated by iterating over the path and checking each increment?

[codecov]

Codecov Report

Merging #2216 into master will increase coverage by 0.00%.
The diff coverage is 100.00%.

@@           Coverage Diff           @@
##           master    #2216   +/-   ##
=======================================
  Coverage   98.08%   98.08%           
=======================================
  Files         262      262           
  Lines       10679    10680    +1     
=======================================
+ Hits        10474    10475    +1     
  Misses        205      205           

Impacted Files	Coverage Δ
include/seqan3/argument_parser/validators.hpp	90.45% <100.00%> (+0.04%)	⬆️

---

Continue to review full report at Codecov.

Legend - Click here to learn more
Δ = absolute <relative> (impact), ø = not affected, ? = missing data
Powered by Codecov. Last update 8140b10...a78526a. Read the comment docs.

[smehringer]

So the problem was that std::filesystem::path models forward_range and that makes it use the wrong overload because it is more constraint than the option_value_type one?

It's fine that it models forward_range but unfortunately the requires clause is also true, namely that the value_type of the range is convertible to std::string. The range value type of a std::filesystem::path is again a std::filesystem::path 🙄

And then weird stuff happened because it validated by iterating over the path and checking each increment?

Yes, somehow it creates an infinite loop 🤔

[eseiler]

Yes, somehow it creates an infinite loop 🤔

Shouldn't the loop happen on compiletime, because that's where the types must be resolved.
I thought it will try to iterate, create some string, then something goes out of scope and then it segfaults...

[smehringer]

Yes, somehow it creates an infinite loop thinking

Shouldn't the loop happen on compiletime, because that's where the types must be resolved.
I thought it will try to iterate, create some string, then something goes out of scope and then it segfaults...

No it compiles fine but the infinite loop is created when iterating over the file path itself. I don't know what the specs say but I think this will just return the path itself for ever and ever

[eseiler]

No it compiles fine but the infinite loop is created when iterating over the file path itself. I don't know what the specs say but I think this will just return the path itself for ever and ever

Ah yes, it returns the path itself. I thought that you meant that the function call is recursive before.

[smehringer]

@marehr actually your proposed changes to the for_each call fixes the error on its own because when it casts the type to string it won't call the function recursively 👍 . I removed the addition to the requires clause again.

[marehr]

thank you 👍

[rrahn]

Some minor suggestion. Otherwise looks good to me.

[rrahn]

thanks