Thanks for testing. All the compile errors are in hodl.cpp yet that file hasn't changed in
several releases. This is c++ and it involves templates, way over my head.

The trigger was a change in algo-gate by moving some code from the default target to the hodl
target. A simple cut/paste from cpu-miner.c into hodl-gate.c (and a few other algos' targets).
hodl-gate.c is the only file in the hodl dir that was modified. The only connection to hodl.cpp
is hodl.cpp includes hodl-gate.h.

Hodl.cpp contains a number of hacks already to get it working, perhaps the new compiler doesn't like
them.

The only difference I can see from moving code from cpu-miner.c to hodl-gate.c is that code now sees
different #includes. But how does that affect compiling a third file?

I'm pretty much stuck. All I can do is tinker with the code so see if I can fix it as magically as I broke
it. But that tinkering would have no direction and would be like throwing darts blindfolded.

The unfortunate part of this is it prevents using any of the algos in that release.

I am considering dropping support for non-aes hodl in the next release and ripping out all those hacks.
I knew it was fragile code and was afraid to touch it once it was working.
Non-aes hodl will still available in the legacy 3.4.12 version. There have been no changes to hodl
since then and not likely any changes in the future, espescially to the non-aes version.