ICE Fix #16562 #16912
ICE Fix #16562 #16912
Conversation
Detect ty_err in fixup_substs and return None instead of failing. In search_for_vtable return vtable_error in both early and non-early if fixup_substs returns None. This was asserting previously. I don't know anything about the early vs. non-early stages and whether this is significant or not. Reduced the amount of data being printed to the debug log in tydecode.rs. It was printing several megabytes of metadata. Added license text to test
wickerwaka
force-pushed
the
ice-16562
branch
from
c83c67b to
6ef79b0
Compare
|
Added the license header, thanks. |
|
@pnkfelix Does this look good to you? |
|
@wickerwaka Nice fix! Sorry for the delay. |
|
ping @pnkfelix |
| _ => fail!("t_f should be a trait") | ||
| } | ||
| }) | ||
| match fixup_ty(vcx, span, t, is_early) { |
There was a problem hiding this comment.
This seems like you manually inlined the effect of map ... am I correct? Was that necessary?
|
Ah, the codebase has changed significantly since this was written. I think this ICE is gone now; the test you have here now does not iCE. I checked with nmatsakis; he says the line-limit in It seems like a rebase is going to amount to:
If you are willing to rebase, (and I hope you are), then that can be considered r+'ed by me. Bonus points if you actually split the rebased PR into two commits (since the two items in my list above are 100% orthogonal to each other, in my opinion). |
|
@wickerwaka (sorry for the delay in the review. obviously things changed in the meantime.) If you want to do the rebase suggested above, great. If you don't feel like bothering, please just drop me a line and I'll try to salvage what I want from this. |
|
@wickerwaka Any news on this? @pnkfelix You might just have to take point here. |
|
@gankro, Sorry, real life getting in the way. I'll take a look at this. |
|
Closing this, everything other than the metadata debug output has been addressed by other changes. Opened a new PR with the metadata stuff: PR #19043 |
internal: Remove span trait
Fixes #16562
Detect ty_err in fixup_substs and return None instead of failing.
In search_for_vtable return vtable_error in both early and non-early
if fixup_substs returns None. This was asserting previously. I don't
know anything about the early vs. non-early stages and whether this
is significant or not.
Reduced the amount of data being printed to the debug log in
tydecode.rs. It was printing several megabytes of metadata.