The Antlr dependency is not like the others. Is that a problem?

[westonpace]

I believe all of our non-antlr dependencies are obtained via submodules. Antlr, on the other hand, is downloaded from github via custom cmake. I think the reasoning behind this discrepancy is that these are the instructions suggested by Antlr itself. Or perhaps it is that Antlr requires java?

Part of me thinks that having these things be consistent will make it easier to understand and maintain for future developers. Is there any particular reason we can't have it as a submodule?

[EpsilonPrime]

Part of the reasoning is the Java dependency. You don't need to pull in a Java compiler to use the prebuilt JAR tool (just the JRE). (For reference, the tool is what converts the grammar into C++ code. There's also the C++ runtime which does the evaluation at runtime.). As a C++ developer having to install a Java compiler to build a C++ library is a bit of a stretch.

That said it will work as a submodule and we can certainly try it. Now that we have a version in it'll be easier to compare what happens with a PR.

[ingomueller-net]

FWIW, I have come to believe that the combined find_package/FetchContent approach I implemented in #138 is actually superior to the approach with git submodules. It may not matter too much for this project, but downstream projects may run into diamond dependencies that they can't properly resolve if all projects use git submodules. With find_package+FetchContent, the top-level project can decide which version is compatible with all dependencies and how to best get it. If anything, I'd move the other dependencies to that approach as well.

NB: googletest and Protobuf have used some version of that for a long time as well.