What are we going to do with --ignore-file and --ignore-dir

[Andy Lester]

This is a dump of dev/IGNORE.md. The biggest problem we have is the idea of paths vs. directory names.

Thoughts?


# --ignore-dir and --ignore-file revamp

Users have been dissatisfied with `--ignore-dir` and `--ignore-file`. Here are some common "bug reports"

- `--ignore-dir=foo/bar` doesn't ignore `foo/bar` at all (GH #291)
- `--ignore-dir=foo` ignores all directories named `foo`, not just the one at the root of the search (GH #216)
- `--ignore-dir=foo foo -f` doesn't search `foo/` (GH #492)
- No way to ignore a file in a folder (GH #479)
- `--ignore-dir` doesn't implement all filters (GH #42)
- What's the right behavior for `--ignore-dir=foo`, `--noignore-dir=bar`, `--ignore-dir=baz`, wrt. `foo/bar/baz/file.txt`

Also consider GH #330, and consider that you could be anywhere in a project but still source the `--ignore-dir=./foo` rule
from an ackrc a few directories above. Are we preparing to teach ack about the notion of a project root? Also, consider
`--noignore-dir`

# Future directions

## Scenario 1

--ignore-dir can ignore paths as well as directories. But then relative
to what? Current directory? That means an --ignore-dir in an .ackrc
isn't useful.

Do we make the relativeness based on where the directive is? Is an
--ignore-dir in a project-level .ackrc relative to that file? And an
--ignore-dir from the command-line is relative to Cwd?

## Scenario 2

--ignore-dir only ignores directories, but warns if you try to ignore a path

What's the distinction between --ignore-file and --type/--notype? Does it make sense to allow an "anonymous" type (ex.
--type=is:Something), and make --ignore-file a synonym for --notype? Could --ignore-dir be implemented like this?

If I have `--ignore-dir=test` in my .ackrc, and I do ack $term test, what happens?
--
Andy Lester => www.petdance.com