I've got a rather large piece of ESP32 code broken down into a .ino file and .h header files containing a number of classes (declarations and implementation - yes, we know its not ideal practice but it's a lot easier to manage and work on than a 15,000 line .ino file). The header files are all #include'd into the .ino file.
When I place a breakpoint in the .ino file VisualMicro stops correctly, etc. But when I place the breakpoint in the .h file they never trigger, even with a breakpoint in the .ino file too. If I take the code from the header file and move it to the .ino file, the breakpoints trigger.

Is this expected? Is there a setting I haven't tried, yet, to use the header files as intended even when debugging?
Thank you!

Thank you Simon. I had kind of expected this to be the answer, unfortunately, but was hoping there was a magic switch somewhere I hadn't found.
I'll try with cpp files. The alternative would be to turn all the supporting classes into 1 or more libraries.
Thanks.

Thank you Tim. I re-factored the code properly (I was being lazy). So now I've got a nice bunch of proper header files and CPP with my .ino. The .ino wouldn't compile when it had the same name as one of the .cpp files so I renamed it.

99.9999% of my code is templates. Template interfaces and implementations must reside in the same translation unit. Standard practices are to put everything in a header <.h> file (e.g. the C++ Standard Library). So, without breaking any ANSI/ISO Standards or 30+ years of coding practice, what is your workaround for using the serial debugger in a header?