|
This hits on the biggest problem with projectional languages, and why text has dominated: they don't interoperate well with any existing developer tooling. So the language implementors have to write everything the developer needs: code editor, source control, code review, code browser, etc You're completely isolated from the wider ecosystem of development, and the benefits so far don't seem worth it. As described here, you save on the reference resolution phase of linking, which is nice, it should be faster and eliminate some classes of error. The nice part about text as a common format is that it's common. If every language has to implement every tool then it's an N * M problem. When tools can be reused across languages they become N + M. A user of a projectional language is giving up git, Github, their favorite code editor, and depending on language-specific tooling to replace all of that. And even when the lang-specific tooling is sufficient, it needs to be relearned and recustomized (think all the configuration just in a code editor: prefs, theme, extensions like VIM keybinds and Copilot, etc). Is the benefit worth the cost? |