Another fix for merge_imports
#3769
Conversation
811f06f to
872427a
Compare
| @@ -988,7 +1000,12 @@ mod test { | |||
| test_merge!(["a::b::{c, d}", "a::b::{e, f}"], ["a::b::{c, d, e, f}"]); | |||
| test_merge!(["a::b::c", "a::b"], ["a::{b, b::c}"]); | |||
| test_merge!(["a::b", "a::b"], ["a::b"]); | |||
| test_merge!(["a", "a::b", "a::b::c"], ["a::{self, b::{self, c}}"]); | |||
| test_merge!(["a", "a::b", "a::b::c"], ["a::{self, b, b::c}"]); | |||
There was a problem hiding this comment.
Hmm, looking back to this test, merging use a; into use a::{self}; does not seem to be correct w.r.t. #3750.
@valff IIUC this has existed before this PR, but while you are at it, would you mind having a look if you have time? This won't be a block for this PR - if this is not easy to fix then we can leave it for the future work.
There was a problem hiding this comment.
@topecongiro I have kept use a::{self} as a special case where self is at depth 2, because the first path segment is always a crate name or a special keyword (at least in edition 2018?). So if the first segment is always a module (crate), it should be correct w.r.t. #3750. I find this special case useful because this way we still have a single use statement per crate.
There was a problem hiding this comment.
Ah, I misunderstood this. Thanks!
Sorry, my previous PR #3753 didn't fix #3750 for all cases. This should finally fix it.