Add import-symbol-at-point to package-fu

[high-ego]

I wrote a function analogous to export-symbol-at-point. Unfortunately, while writing it, I rewrote a large part of the contrib. I would write tests to check if I broke something, but I'm not sure how to test this.

I did a little manual checking, and the only thing I think I broke is unexporting when the symbols are written as strings. The problem is that slime-sexp-at-point returns nil when it's called from a position like ("foo"|), with the point after a string that's the last element in a list. I don't know if this is expected behaviour, and I should write a wrapper that does what I need (return "foo"), or if I should change the function.

The messages that the new version produces are a little less informative. They are the same regardless of whether the defpackage form was changed. I'll change this if anyone thinks the distinction is important.

[joaotavora]

high-ego <notifications@github.com> writes:

I wrote a function analogous to export-symbol-at-point. Unfortunately, while writing it, I rewrote a large part of the contrib. I would write tests to check if I broke something, but I'm not sure how to test this. I did a little manual checking, and the only thing I think I broke is unexporting when the symbols are written as strings. The problem is that slime-sexp-at-point returns nil when it's called from a position like ("foo"|), with the point after a string that's the last element in a list. I don't know if this is expected behaviour, and I should write a wrapper that does what I need (return "foo"), or if I should change the function. The messages that the new version produces are a little less informative. They are the same regardless of whether the defpackage form was changed. I'll change this if anyone thinks the distinction is important.

FWIW, someone added this to SLY some years ago: joaotavora/sly#147 and then I cleaned it up. As far as I remember, didn't take a (big) rewrite and nothing was broken. Also, the author is fond of ASDF's "package-inferred system" and wrote a sly-import-symbol-package-transform-function to deal with that, which I don't use.

[luismbo]

@high-ego can you have a look at João's commit and try to port things over, maybe, or double down on the need to rewrite.

[high-ego]

I decided to stick with the rewrite, and wrote some tests for the parts I rewrote. To fix the case broken before, I modified slime-bounds-of-sexp-at-point.