Brian Cook

I still suspect the line should be entirely removed; that _section_level_ should not be adjusted at that point. What is the intention? Why should _section_level_ be adjusted in _visit_start_of_file_?

@estk, were you able to reproduce the problem?

The latter is really a requirement. The file set can also be damaged if the program is terminated abnormally with or without using a thread. Imagine `SIGKILL` while the files...

I have had trouble waiting on a `join_handle` in a `Drop` implementation. The situation was a bit different. I mention the trouble because it could be a dead end. The...

Agree. Sounds good! I think I have a sequence for rolling the log file that is not much more complicated than what's already in log4rs and is actually thread safe....

I am. The fix requires an interface change to `FixedWindowRollerBuilder::build`. That change allows another as-yet-unreported bug to be fixed. Still interested?

There really are two things at play. The directory and the pattern. While similar (they're both a sequence of characters) they do have different purposes, different constraints, and different behaviours....

[The R&D repository for this issue.](https://github.com/Coding-Badly/pattern-path-buf)

> ...a few other breaking changes... If this is a breaking change then I made a mistake. 😄 > ...maybe we'll keep this open for a bit while we proceed...

The _digitalRead_ can be handled with an operator... ``` bool operator ! (const PinStatus &v) { return v == HIGH ? LOW : HIGH; } ```