# Library Interface

* Introduce namespaces for command names
    * By making command names a list, e.g. `'(posix quit)`
    * POSIX namespace should be treated as reserved
* Documentation: Clarify interaction between `edward ed addr` and `edward ed editor`
* `(edward ed cli)`: Allow defining custom command-line options
* Provide dedicated types for addresses and ranges
* Add `(chicken type)` annotations for the public API to catch more
  usage errors during compile-time instead of run-time.
* Provide an `(edward)` meta-library which imports all sub-libraries.

# Miscellaneous

* Consider using Declarations and (chicken type) to enable more optimizations
	* https://wiki.call-cc.org/programming-for-performance
	* https://wiki.call-cc.org/man/5/Declarations
	* https://wiki.call-cc.org/man/5/Foreign%20type%20specifiers
* In append/insert/… command: Verify that the address is valid before
  entering insert mode. For example, when performing `2a` on an empty
  file the append will only error out after all text has been entered.
* Figure out why edward is slower when compiled statically with eggs (linkage static)
* Profile larger ed inputs (e.g. as generated by a fuzzer) with
  chicken-profile and figure identify (parsing) procedures which
  could be further optimized for performance
* Improve parser error messages (mostly returns `failed char pred` currently)
    * Consider prefixing parser errors with command parser name
      (how should the `g` command be handled in this regard?)
    * Consider adding column information to parser errors
    * Add parse-with-failure-reason to more parser combinators
* Refactor parse-line in a way that it can consume end-of-file.
  Requires refactoring of parse-input-mode.
* Allow input with spaces in interactive tests
    * Probably requires rewriting the test runner in Python or something
    * Would enable adding tests for edge cases like c0fe7d15700cc1ca03198c05d3762f7516f185fe