Uh oh!
There was an error while loading. Please reload this page.
Move coverage-increasing void out of define-values. - #201
Conversation
mfelleisen
commented
Sep 18, 2023
@samth Do you know why the tests fail? |
Also copy properties appropriately to support `'errortrace-annotate`. Fixesracket#200.
samth
commented
Jan 7, 2024
The test failure is in |
mikesperber
commented
Jan 14, 2024
@samth I can confirm this breaks the stepper. Here's a sample program: #lang htdp/isl
(define-struct pare (kar kdr))
(define (add-pare pare)
(+ (pare-kar pare)
(pare-kdr pare)))
(add-pare (make-pare 2342))This completes immediately with the PR in place, with no output whatsoever. |
mikesperber
commented
Jan 14, 2024
samth
commented
Jan 14, 2024
Does that fix the problem in #200? I would still like to actually do the right thing here, which I assume involves changing the stepper to cope with this change. |
mikesperber
commented
Jan 14, 2024
Yes. |
rfindler
commented
Jan 14, 2024
I can't remember the details that I apparently unearthed a while ago, but I do remember enough to remain confident that the change I suggested in the comment that @mikesperber links to (starting with the comment "Ah. This is probably the right fix:") is a good change. Maybe other changes are also good, but that one seems to me to pretty clearly be right. It seems likely to me that that code was trying to get something about coverage to work, but it was copying over only the source locations, where it needs to actually copy all the properties, not just the source locations. |
Fixes#200.