Skip redundant frames in const recursion errors #136649
Conversation
estebank
commented
Feb 6, 2025
estebank
commented
```
error[E0080]: evaluation of constant value failed
--> $DIR/infinite-recursion-const-fn.rs:4:5
|
LL | b()
| ^^^ reached the configured maximum number of stack frames
|
note: inside `a`
--> $DIR/infinite-recursion-const-fn.rs:4:5
|
LL | b()
| ^^^
note: inside `b`
--> $DIR/infinite-recursion-const-fn.rs:7:5
|
LL | a()
| ^^^
note: [... 125 additional calls ...] inside `b`
--> $DIR/infinite-recursion-const-fn.rs:7:5
|
LL | a()
| ^^^
note: inside `ARR::{constant#0}`
--> $DIR/infinite-recursion-const-fn.rs:9:18
|
LL | const ARR: [i32; a()] = [5; 6];
| ^^^
```
|
r? @Noratrieb rustbot has assigned @Noratrieb. Use |
rustbot
added
S-waiting-on-review
|
Some changes occurred to the CTFE machinery cc @rust-lang/wg-const-eval |
| && last_frame.where_ == frame.where_ | ||
| && last_frame.instance == frame.instance => | ||
| { | ||
| Some(last_frame) if !seen.insert(frame.clone()) => { |
There was a problem hiding this comment.
Please add comments explaining what is done and why.
I am not sure I agree with removing duplicates from a backtrace -- that seems like it could be quite confusing. I can't quite follow the logic for printing the "N frames skipped" here, is that always printed when seeing a duplicate?
This new logic abuses the times field, where it no longer means "we saw this function N+1 times" -- I think that's too hacky and confusing.
| // Don't print [... additional calls ...] if the number of lines is small | ||
| if frame.times < 3 { | ||
| let times = frame.times; |
There was a problem hiding this comment.
Is this logic still correct? It seems to be we could now have a situation where we see 2 frames that both have been seen already, so last_frame.times is 2, but these were actually 2 different frames and then this here will print the wrong thing.
| LL | hint_unreachable() | ||
| | ^^^^^^^^^^^^^^^^^^ | ||
| note: inside `hint_unreachable` | ||
| note: [... 124 additional calls ...] inside `hint_unreachable` |
There was a problem hiding this comment.
This would probably look better as 2 separate notes, now that it doesn't represent 124 calls of the same function any more.
Noratrieb
added
S-waiting-on-author
|
☔ The latest upstream changes (presumably #137848) made this pull request unmergeable. Please resolve the merge conflicts. |
