Closed rhendric closed 4 years ago
im sure you'll make it! add more comments into code for the next readers, who knows what will happen to github
@determin1st, anything in particular that doesn't seem clear to you? All of the new code reads as fairly self-explanatory to me, but getting other perspectives on that sort of thing is why I put myself through review.
@rhendric sorry, my comments here are not useful. i'd like to dig into compiler later.. used to explain everything in code comments.. so, really, don't bother :)
This PR contains five bug fixes (two of which are regressions that I caused), a very minor feature, and some housekeeping, all related to or needed by the interactions between
rewrite-shorthand
andunfold-soak
. The title commit reads as follows:This issue was a regression from 1.5.0 first appearing in commit 8cffc57, in which I introduced the
rewrite-shorthand
method and moved slice expanding into it. The problem with my refactoring is that it caused everything inrewrite-shorthand
to happen beforeunfold-soak
, but the code in the issue relies onunfold-soak
happening prior to slices being expanded.This commit prepends a short-circuiting call to
unfold-soak
in therewrite-shorthand
method of every node type that defines a non-trivialunfold-soak
.It seems I made quite the mess when introducing
rewrite-shorthand
! In addition to the regressions tested and fixed here, there are a few other shorthand-rewriting regressions which will be fixed in the next PR. This PR was getting quite large though, so let's get this merged first.These fixes aren't especially complicated, but I'm feeling especially humbled by how much I screwed up, so I'm going to set the review period for this PR at three weeks; I will merge on or after June 29 if there are no comments.
Fixes #1102.