We actually do track a 'devcontainer.json' file (to keep as a template). We
just don't keep it in the '.devcontainer' dir, for the same reason we don't
track the '.vscode' dir: so devs can tailor it to their specific needs.
Move all lexer/parser files into a single dir, DataStreamGrammar, for
organizational purposes and to avoid dependency conflicts when
generating code. Note that we'll only need to regen cs if we modify the
parser/lexer files, which seems unlikely.
This is a more appropriate name (it is a web app after all), especially
now that we've renamed sparkcmd -> Spark.Console (a literal console
app). The old name implies a parallel between the two apps that doesn't
exist.
The IDE (vscode) generates a 'bin' dir in the Spark.Admin.Console proj.
Presumably, this happens because the out dir is specified as a command
line arg, rather than defined in a proj file, as is the case w/ the
Spark client build.
Not sure why these files were tracked in the first place, but cmake
automatically detects changes to the source file glob. Perhaps they were
originally added because it didn't work on some older version of CMake.
Clean up the web build process a bit while we're at it
* Add 'custom_commands' to run npm install in CMakeLists
* Generate tsconfig files for CMake preset configurations
* Rename local variables to _snake_case in CMakeLists
For some reason, the build fails when the output dir is referenced by the c#
extension in vscode. It produces errors saying various g.cs files (which are
generated for the spark csproj) could not be found. However, examining the
contents of the output directory after the build "fails" show that those files
exist, and are in the proper build subdirectory. Seems like a race.