Consider using an independent process #1
Description
goldfish currently is designed to run as a binary in a subprocess. This means there's a short period indexing the drive on every startup. Adding more data sources, or searching hidden files, will significantly increase indexing time especially since the launcher use case usually means searching all of $HOME.
One possible solution is to run as a separate process and provide a socket (or server?) to connect the running "instance". Changing this would likely mean:
- no more cli parsing. search flags need to be sent over the ipc
- "Proper" data format becomes for important for passing flags to a search and associating a query with the correct application
- all the extra work related to caching and watching the filesystem for changes