Support ignored files as a parameter

[erikh]

This closes #74, see comments for design work.

Still unfinished as of this writing.

[erikh]

This has been changed to live entirely in the spec, as requested.

There are two fields, the "string form" (which is easier for end-users in many cases) and the "list form" (the more programmable version). I figured both of these would be nice, but feel free to tell me otherwise. I would vastly prefer the list form if I had to choose.

There's still some docs and testing to do but this is functioning as intended with initial hand testing.

[erikh]

Another thing we could do, is coalesce these into a subtype that is used by the spec.

It would also make the logic simpler in the loadExcludedFiles code by keying off a single attribute.

LMK how you'd like me to press forward.

[stefanprodan]

@erikh i would opt for a single field that represents a .sourceignore file embedded like showed in the issue, with one expression per line.

[erikh]

Adjusted to support .sourceignore format, PTAL at impl, I think you'll be happy with this round.

Will hack on some tests tomorrow hopefully.

[erikh]

ok, sorry about that, some messy comments from the earlier patches. I've also adjusted a few small bits around the handling of the .gitignore scanner that simplifies things since I made the last comment.

Should be reviewable now.

[stefanprodan]

I just noticed that the defaults are broken in master, I'm getting the .git dir in the artifact. Tests would be awesome

[stefanprodan]

The API looks good now, thanks!

[erikh]

I can't reproduce (the git directory is not in the tarball), but I will try to make a test for that separately.

[stefanprodan]

I can't reproduce (the git directory is not in the tarball), but I will try to make a test for that separately.

I had a .sourceignore file in one of my test repo 🤦 all good, my messy tests were at fault

[erikh]

This is ready. PTAL please.

[stefanprodan]

LGTM

Thanks @erikh