Uh oh!
There was an error while loading. Please reload this page.
use ObligationCtxt not QueryNormalizer in rustdoc's normalization - #108503
use ObligationCtxt not QueryNormalizer in rustdoc's normalization#108503BoxyUwU wants to merge 2 commits into
ObligationCtxt not QueryNormalizer in rustdoc's normalization#108503Conversation
rustbot
commented
Feb 26, 2023
r? @notriddle (rustbot has picked a reviewer for you, use r? to override) |
jyn514
commented
Feb 26, 2023
r? types |
GuillaumeGomez
commented
Mar 25, 2023
Let's re-roll. r? types |
Uh oh!
There was an error while loading. Please reload this page.
oli-obk
commented
Mar 27, 2023
r=me after rebase @rustbot author |
jyn514
commented
Mar 30, 2023
@oli-obk do you know what's going on here? or who we should ask? |
a2868c5 to
390246cCompareoli-obk
commented
Mar 30, 2023
BoxyUwU
commented
Mar 31, 2023
@bors r=oli-obk |
bors
commented
Mar 31, 2023
bors
commented
Mar 31, 2023
⌛ Testing commit d0c308c with merge a14c185f4330588cd23391993cb86940913a391d... |
bors
commented
Mar 31, 2023
💔 Test failed - checks-actions |
notriddle
commented
Mar 31, 2023
@bors retry LLVM build failed |
bors
commented
Jul 30, 2023
☔ The latest upstream changes (presumably #114264) made this pull request unmergeable. Please resolve the merge conflicts. |
fmease
commented
Sep 5, 2023
Locally I've rebased the commits and run @BoxyUwU Would you be willing to bring this PR up to speed and re |
fmease
commented
Sep 5, 2023
Fyi, your PR seems to fix #112242. If you could add a regression test for it, that would be awesome! :) |
d0c308c to
2ccfd04Comparefmease
commented
Sep 21, 2023
@bors r=oli-obk,fmease |
bors
commented
Sep 21, 2023
bors
commented
Sep 22, 2023
…izer, r=oli-obk,fmease use `ObligationCtxt` not `QueryNormalizer` in rustdoc's `normalization` `QueryNormalizer` doesn't handle not-well-formed projections or ambiguity so should not be used by rustdoc as rustdoc happens on code that is not well formed. This PR replaces the usage of `QueryNormalizer` with `ObligationCtxt::normalize` which is designed to work on not-wf code while not being in typeck. This also removes two uses of `actually_rustdoc` from the compiler which seems good to me. I am somewhat confused as to the "point" of `QueryNormalizer`, it intends to be "the main way of normalizing" in the future and yet ICEs when encountering not wf types or when normalization is ambiguous which seems very incompatible with its stated goal since that makes it only suitable for using after typeck?
rust-log-analyzer
commented
Sep 22, 2023
The job Click to see the possible cause of the failure (guessed by this bot) |
bors
commented
Sep 22, 2023
💔 Test failed - checks-actions |
fmease
commented
Sep 22, 2023
Ah, now I can see that it happens with stage2→stage3, not with stage1→stage2. Let me investigate this ICE… |
ehuss
commented
Sep 23, 2023
@bors r- synchronizing the queue |
bors
commented
Oct 26, 2023
☔ The latest upstream changes (presumably #117193) made this pull request unmergeable. Please resolve the merge conflicts. |
Dylan-DPC
commented
Feb 6, 2024
@BoxyUwU any updates on the CI failure? |
QueryNormalizerdoesn't handle not-well-formed projections or ambiguity so should not be used by rustdoc as rustdoc happens on code that is not well formed. This PR replaces the usage ofQueryNormalizerwithObligationCtxt::normalizewhich is designed to work on not-wf code while not being in typeck. This also removes two uses ofactually_rustdocfrom the compiler which seems good to me.I am somewhat confused as to the "point" of
QueryNormalizer, it intends to be "the main way of normalizing" in the future and yet ICEs when encountering not wf types or when normalization is ambiguous which seems very incompatible with its stated goal since that makes it only suitable for using after typeck?