[MIR] Implement translation of references to various items #29907
Conversation
|
Nits addressed. |
nagisa
force-pushed
the
mir-moar-constants
branch
from
53aa6b3 to
6bb92a5
Compare
| substs: &'tcx Substs<'tcx>, | ||
| did: DefId) | ||
| -> OperandRef<'tcx> { | ||
| match ty.sty { |
There was a problem hiding this comment.
Checking the type of the item like this feels like a horrible hack. Perhaps we should just be including more information in the Literal MIR node? Or a Def (as I think you suggested on IRC)? Also, I haven't gone digging into trans, but is this code here a kind if port of code that exists elsewhere?
There was a problem hiding this comment.
Also, I haven't gone digging into trans, but is this code here a kind if port of code that exists elsewhere?
Nope. The trans_fn_ref below is a simplified “port”, but I haven’t found anything in old trans which would do matching like this (AFAICT it gathers such information from various other sources, esp. HIR).
Perhaps we should just be including more information in the Literal MIR node? Or a Def (as I think you suggested on IRC)?
This approach makes it pretty easy to match whether enum/struct references should be translated as functions or not. I’m not sure other methods would allow checking for this as easily.
I’m willing to explore for better methods to match correct path, but won’t be able to dedicate much time for that in coming weeks.
There was a problem hiding this comment.
Okay, I’m convinced now this is not only a hack but is also wrong (e.g. it will panic compiling a function that’s defined in impl block).
However, I won’t be able to get back to it for at least a week and a half. I’ll go ahead and mark the issue as [WIP] and hopefully work on it later.
|
☔ The latest upstream changes (presumably #29886) made this pull request unmergeable. Please resolve the merge conflicts. |
nagisa
force-pushed
the
mir-moar-constants
branch
2 times, most recently
from
9a2607e to
f013dd2
Compare
|
☔ The latest upstream changes (presumably #29534) made this pull request unmergeable. Please resolve the merge conflicts. |
nagisa
force-pushed
the
mir-moar-constants
branch
from
a3485cc to
250d9e7
Compare
nagisa
changed the title
|
☔ The latest upstream changes (presumably #25570) made this pull request unmergeable. Please resolve the merge conflicts. |
nagisa
force-pushed
the
mir-moar-constants
branch
from
250d9e7 to
3673ee5
Compare
|
I changed my approach to storing relevant kind inside I’m not entirely sure I’m using the correct substs in all the correct locations, but things seem to be working; something to be reviewed more carefully. |
nagisa
changed the title
|
☔ The latest upstream changes (presumably #30301) made this pull request unmergeable. Please resolve the merge conflicts. |
nagisa
force-pushed
the
mir-moar-constants
branch
from
f02c51a to
09033dc
Compare
nagisa
force-pushed
the
mir-moar-constants
branch
from
09033dc to
f95aae4
Compare
|
(I hate how github removed all the comments on force push :() nagisa@ece05d3 and nagisa@5fb2edc were the two commits that had comments. |
nagisa
force-pushed
the
mir-moar-constants
branch
from
f95aae4 to
7dd9579
Compare
|
📌 Commit 7dd9579 has been approved by |
Still will not translate references to items like `X` or `Y::V` where
```
struct X;
enum Y { V }
```
but I must go work on university things so I’m PRing what I have.
r? @nikomatsakis
Still will not translate references to items like
XorY::Vwherebut I must go work on university things so I’m PRing what I have.
r? @nikomatsakis