-
Notifications
You must be signed in to change notification settings - Fork 49
Description
These functions each execute a script which loads the previous file list, regenerates the code in the build directory, deletes files which are not in the new file list, then copies new/modified files (including the file list) back to the source directory. These functions assume that you will always build the generated code in a separate sub-directory, and all of the files in that sub-directory are generated by this function.
That deletion step means that if you start building a schema or client target in an existing directory, you may delete some of your implementation files the first time you try to build with update_graphql_..._files
!
I think the scripts need to be more robust to misuse. They should output an error if the directory contains any files but the file list is missing, and they should maybe generate a separate list of headers which can be used to exhaustively compare the directory contents with the previous generation pass. If any unrecognized *.h
or *.cpp
files are in the directory, it should fail fast.