---
tags: steering, rustc
---

# P-high review for 2022-01-28 meeting

At the bottom of this document is a graph. Rectangles are issues (with numbers and titles). Circles are labels. An arc from a label (circle) to an issue (rectangle) mean that that issue has that label.
* Only some labels were chosen to be shown as circles, based on whether they seemed to represent a *category* of bug where we want an easy way to see their neighbors in the graph.
* Some other labels, like `I-unsound`, are instead represented as rectangle colors; there's a legend at top of the graph explaining this.
* Some arcs were removed after the fact to keep the diagram manageable.
* Some arcs are missing arrow-heads. This is just an artifact how pnkfelix tried to keep the layout under control.

## High-Level Findings

* Are some things misprioritized? What is risk associated with each, if left unaddressed?
* (so far, when pnkfelix skims stuff, they are usually able to either justify the P-high priority (i.e. does not feel comfortable downgrading), or, as soon as they find one they would consider deprioritizing, they go look at the specific issue and end up closing it as fixed. :laughing:)
* Sometimes an A-label corresponds to a subteam that will actively engage in triage (as in, actively reconsider the P-labelling, and/or see if anyone is working on resolving the issue, either in a dedicated meeting (e.g. wg-async) or during the T-compiler triage (e.g. wg-llvm). But its hard to tell that from the labels alone.
* In any case, pnkfelix is pretty sure that things that are solely labelled with A-inference are not getting triaged. Consider e.g. [rust#80816][].
* likewise, somethings end up with just the F-const_generics label, but not the A-const-generics label. Does that mean things sometimes slip under the rug? The F-labels are useful metadata, but I don't know if we can rely on them for categorization purposes...
* pnkfelix would like us to enter a state where its easy to divide issues into two buckets:
* bucket:delegated "this has some dedicated team that *does* do triage of P-high stuff at some cadence"
* bucket:remainder "this falls on T-compiler itself to eventually delegate"
* Big picture ideas:
* Some issues had no A-label, so I resorted to picking an F-feature label for them in this picture instead.
* For some reason, a *lot* of A-labels have at most one or two P-high issues associated with them: A-frontend, A-lint, A-inference, A-testsuite, A-pin, A-ffi, A-NLL, A-macros, A-edition-2021, A-debuginfo, A-const-eval, A-proc-macros, A-const-generics, A-dst, A-simd, A-cross, A-runtime, A-generators. (A couple pairs of these are even mapping to the same issue.)
* That may be an artifact of the labels (maybe some get rarely used; e.g. are things being put into A-diagnostics when A-lint could be more appropriate?)
* In any case, with so few issues in each of those buckets, it seems okay to just try to map each of those A-labels to at most one subteam and ping them about whether they have a plan of attack for the issue. (And the A-labels with no associated subteam are the ones that fall into bucket:remainder noted above.)
* 17 issues have no categorization tag at all (see "untagged" at bottom)
* several of those seems to be about performance issues, either with compilation time or with object code quality.
* Maybe wg-performance could try to divvy up some work there (though much of wg-performance's efforts are about measuring and tracking the performance of rustc, not resolving performance issues with output code...)
* There are many issues (more than 5) in each of A-traits, A-linkage, A-llvm, A-lifetimes categories here.
* pnkfelix thinks wg-traits is actively working on changes that will address many of the A-traits issues. It would be nice to know which, if any, have no current plan of attack.
* pnkfelix thinks WG-llvm, rightfully, focuses its energy on P-critical and other true show stopper A-llvm issues. I don't think WG-llvm has a formal structure where it has plans to address these P-high things. (Please correct me if I'm wrong here.)
* Should we be looking into setting up some kind of membership drive and/or mentorship to try to get more people engaged with fixing LLVM issues that are blocking their own work locally?
* pnkfelix thinks A-linkage is an example of an area that fundamentally has no dedicated working group, and won't have one with a groundswell of talent dedicated to resolving those issues. It probably falls upon T-compiler members (presumably ones who are paid to work on rustc) to just chomp on each of those bugs when they have available time. We should nonetheless talk about coordinating effort here; it is a sort of scary set of issues.
* pnkfelix forgot to say anything about A-lifetimes. (They only noticed that it was in this bucket of "A-label with large number of issues" late in the game.) Is there a team that we think is dedicated to this? (Maybe the remnants of wg-nll?)

[rust#80816]: https://github.com/rust-lang/rust/issues/80816

## Meeting Question/Topic queue

* jack huey: IMO every P-high issue should have a WG-* label. Most everything there labelled A-traits should probably be WG-compiler-traits labelled (too)
* pnkfelix: Rust's proposition as a "safe language" is somewhat underminded by every I-unsound issue. It would be neat to eventually get to a point where *every* I-unsound issue is P-critical.
* pnkfelix: what to do about A-lifetimes?
* wwiser: We have a huge number of A- labels and most of our P-high issues are concentrated in a few areas so the rest of the P-high issues are probably dispersed over the large set of area labels. I'm not sure what this points to in terms of getting those issues resolved though. There are so many areas of the compiler that it's far too large a problem for only a few people to tackle.
* pnkfelix: the ongoing pinging to P-high owners could arguably be driven by WG-prioritization. "workload is not the issue, rather how to frame some of the issues that are prioritized"

## Meeting Notes discussion

### Every P-high+T-compiler label should have a subteam (WG)

* furthermore, we need regular double-check that every P-high issue has *a team* in the first place

### I-unsound

* *new* I-unsound issues get P-critical. The P-high I-unsound issues are odd cases that have been grandfathered in, sort of.
* pinging process should hopefully yield a set of plans of attack, and/or uncover cases that have no such plan

### A-lifetimes

* lets classify those as WG-nll by default for now

### ...

* *everything* gets solved by Ownership. :laughing:

## Picture

[(link to separately hosted rendering of original diagram that may be easier to zoom in and out of)](https://mermaid.live/view/#pako:eNqFWutuI8exfpWOEqwkmCPzfvMlkLW79gbybrBaOwfwBkpzpkm2NTdPz4iSDQP5dR7g_Mvr5UnyVVXPhVwyZwFbNcNh173qqxr-dhZmkTlbnm0KnW_V7fuPqVKuWsllbDYmjegW_sV6ZeKLiz_dXn_z6vby8mP6sVTWucr89Kc3d3c_vPo73-GnVBB8LZ_5774JqtRlFZ2FZ14FqTGRC0rjyr0bSfho9m6srDNhSYewHERcB-siS0tcX1y09OWleoGPYpuWdJv-4haLMevPB4ufPp79kYmlOl9nxcpGKns0xa6wJb7-5ar4Wq2eIXxpCqXjONupLFXl1iinE_wvzHIDazya-PzjGSnqRbHp2hQmDQ0xbS5qznMwnBJnJsC5MJvCOGdx9uBqDFnyIlvFJmH-O1tula5K-KMwa6XTSJWFtqVKTLnNImGsVM04LIIwS3Jh7C9qxpP5aDonxkyA8bVa2Y3qsLfQLnuwWbAzq7xy28DZJI9NkOVlrSGxyW1KDPCn0WnSX7A1meCj_4rDvHtTHI-zH7P40aYb6KDoWPWSNPq-KllPmJ--8uWL6DlVt1mo4w-k5tf7ll2vLft3bRulhtPJkJUigjivsgIhmLrSaLhzDSF2No2Ica5dqV6_fqPWVRqWUNjVp18Hb29v6WT8aU4eDQZjPpkInPxTGsd_V_-4_wcOQnwUqVNum1VxpNKsVCE0LZV2CvyLbOdag72GFyDOPbLGFDZ0FxeHd5qgHM8GIw5KIsAyMqtqE2jnTEHyqjc3r5ZsrvbWWtvYRMsv1R8edVyZq61298aFGt7Z3K_I_PePunAXl12B4Bj7q4nuYzI0C7R_pxFoMJZYZQIC7T_GzB1ihmWq8m0GY-jYbtLEwBiF-aWyhSHadeMn0WGROTK3UE0UDYaDAUcREeD2Pcc4znFZXLG2BrYt6qC_Dkxk6XYw7A8HdF732ie_sBAGi-FgNCMGTLA6hdHhViPd_nDx4pcqK7_47Xf5e6l2WfHgKOPfV4gbOrKrBLsGyZ0R3-aiVmXRH02EExHg9PJv1-9fK3qEQ92VurShIs-w7ZAssV2psECtcQo-xBNZnptIrQyHbpFFVWjIFrGFuM_0jdsP79SqsnHk9rOEgyswiAeSrb3yFoms3qSZKzkU9y7bdB5MJpLOICD9mwQCPBrVeZbFRnrRwZVm5-CKmWn4noPRB6d4zbVuc9VmgypPnjoQ5vgnnwquajMP51MxM4i9WrpUBdwWKn-iyHuF5EDRubhUmmpEiFRFJ1EUzawBldcMBR4OynVKjoDS_kwTtQp0JfHFYjHyxQIEZa5Fj0B5hWkMfFjujElVUsWlRUFVrZqOeWrl4OXYsJD8VEACdJ77QmVFT_3l7t1bsSZFUUJKOK4-BSV6UBY2zymY_4uks8lswUnGBEvaPLFEtBtVPqOtWTFYkhVGcZnSMQduucuUjZDRFgWAH3V_Jm44POBWuzalTYz7lDt5EFEUdtK_c9lUnOlo2mfxiIB4r1jbwuQo65QJ6O0JvAdVVxVM69O78RHlBbro9KrfTVbJgbYAH95pAp_-ceDTP3BHMK9sbMtntUKleKDTKazAYgIv-6-rOxOv2VxsD6Vhs7BCaKUlEpXck5siIUzB4uk0S5-TrHJNtriOBdFyGcPQX8ZTPq1JDVT9LLSIqCiwpUlYkcN7jSHn84X0EiKgytvMR47YErLstohJ3WiR6wKwBo2NVYky49JzwhmG7U6wp1PQKdkRoC131l1sfh0wRGHxhGqEWgwEgjABod5zjm51uuFKi2ysCmeRdPK9No4P1WywVL_vsVSfooVBg8pWP1NaN95I__3Pf5UAiwjgELqwflBFs2PZCfga3EAiHGjkSFFXwYFBKxK5inzFDzwDaSSsanNFflNeOu_WyDEExR_58IhGL7r54_UbT0bsRCag3-saufRUGGeuQtD3uIJ89_7DNyJhINpL_oYebInPRAOoE27hTy3IC5XaRmQI8uih8lReAHALFW5N-GCiq06ktvqy6LU8nF2ergu3K_-Lzto9p2Ggd-S5fRvURhhNBTATASOc--5ZcxGQjKwK_CesV2HKqkilmlFGejCq_n8NRFZvrDaZPr3NDj5ym5x17DbErytSR2nJ4ubS20yKbom-SZ-3V_7j7sHT8UQwMBMw0K23oGL857joYITwkwPFgwOigF9TCQG5z5VV4EYTMggE5ETLvQUaXJPsdDYfcQIyQUjh5hV94zu7QRcN3usUQaMkK78hYRqwzyOAfEBFVZyVZ44BXNdJ9KCoy1NNc1UXRymNcclADH-aWj5aeDQ5XDCaJJlQztDSnlXlTIPsUMsHV_122iq3HltRXjXIri1F4OHRx2A6Yw5MUCvNQkXjq7qhsYCqJ81fgEBcPfkTqjC6CLfTsfqMmHTLye3tj9-TEvS3rhKYWR_0xvjxlch6qK2lmA_GfelZRNAEdHPz6u7u_sc3726vP7wBYKA6LzWdJi6eRsnJGYbS-9DE8XJ5q39lROkzQOzgpduTow640Vi8TgQHXPrAcUPTvXzbm3Y0qzvxp6dMFzJcMEGtPnEhEAxGbw9r0XSzB1zBaCmch276Z9bjouYnwJIkXyNvt-p8C3UyQu5xdC5o-vIE99lkNJ8IDgLhdfADjQzdKLRPNAlLZFCvrAilqW_f_qDoKFOcO5h0rYHX_A0lCpxiOZ8KWmWiYzYx9-1Lz_4C-v4NMyumyEtZQlQUOAL6R8NghZy5iGBohABgYmjoPmbS7UldF9O-LDuIAOO_ohAT4-AG30tt-JXmqRlALY4JXJkkL2muY7U6A4rM0mKMUhcbgQYW2gR5GOxE5mCTVuSyR-Cw8WedTCakJu2BCB_HBcZmlCq67cmD0K9VqGHrQmDrrRicI-DzvR1GWc9KZFZpBYwyIM78qt_JQAy9D2m2S4lT9RQklYu9VifsOB8OJr6kgOD5Ef63Kc9lfs1TW83DKIFcneiJnmFHd4oB6ghXciY8PlIsGGeEcmbD8ebHr5QGyaKscuHAfKM2nAbDU6mHMYUVYYLmJvAJYmJgSQfk7ZxXYbHOnboqzVMpp17BtrrUTRU3ycpEEXjqIunYjitaXdL8tq_zz876g7GAUyLIm7JP4N0BgQ_ZKMCpvHsDVqEAh_S0fWBQE2oq4XVPXxlMalY2A-qA9WEgjacymTMB1uKsII-pznNZrUdvWgBwfDeNYjyRanbIA9WkP_bVBAROveHCT9_mKqTW9gkn8gS1rZLV4zQJAFhMYDBftK0ny-EFWih-BZSWW0yHJo1iRLbkfhw_Jsslpg3DIPDl9bfLJRitEIFHpMIsNRyLqkSQlclBNIzw8EiOjoos58IvfQqWtKkunuuUp_EH3atotB-qR0fUWBLpaT69Ryfj_DkqwWI294UHBCS48_HrxzQ0B-6yqdmZwnNryy2HPM98GDpgxpASvk3f4wznU_EDEbRCsk50a1YU1z_-z1DRfltR9hY4oYBBtTOdSkXHcXWiKcxj-ZGUUCaW6uDcNsFF_ELnvIBz1erz5kJHjKbyDDXKFEfCiArAoC9YlwgyGAbddLNcwkIXlwTHXZzt6pSQTZX8_6rM7h0_fH785LoHzH0PuHmzVJFF2rvRov8kNVBMLirQwMCQcNxPbHpFdW0wJdLDbbNTbKYcvQRjeIoMO5J_dQZ6SC84xl80SG086k88wO-T097l-he0tw5E7IB5KX0-8ykvNmlGaqv63O5Edxrst0DyAPsfSOz3O-P5gkVkwuPcTlWXmd7-KtFg067sbC6az46C2k_GAV-h5wNxFxO-QofSrZ1wDtsSg2gWII-hjrCtkz0-8AkX1rtuR_uE1Ww4l60oE3u1S_b_6oIASJylG4EjnDysVZyVTuZGnPm5nFnqB1phZlV349e8jKi3PH3Pkghaj1r3gJqOKVOxZelwejmhi0I_t8es17aevAbDmaQjiMY8tDjdFRlLvkl1SViWau5Ns_Jva-1gGqyeUZnrnoPqEz6A4abiWdk21e22rm5HVZmPFzNZQRABSX5IfdUEMEBK5AVvFkoqXw1E7rbl_YF_Oph5AACi8YURHLGs98ZNqNNyyeGWW1sq3nATzVMdw3dWbM3KdFKvTCfTZgVDWQwh4MeSgC-ChlsOLwd8N8hxOorA4e5tfySs_1Pdi0-MthjM-7IeJ4KE2HsJxW-GXsTlFx8YUaivFL0Vep1efLh8sSm_WNbwg8cYVAoMdALZYwCG6FmZJ9Q1-uylyd8iWLtvyQ6n8ulwNpbhmQjJ7KX61u_EZHsQW81o2NtD7qIhua2vRlHFFUju06aE6jy9N3Sl72q1CPs-f_Hp5kBwGRM0s6JsMMyrVlKr2rpHG3kbeQG--_DeP4fuXrZlZ2c0NU1KYw4Q3gcc3WAd21fMBSkxAWE6OwRAsMhGBGxcjUaoOXnpzk9ykP_o_8D6erMx0cVFTdXtYDqdSztgAnzLLbnV5_i51EDgXHkRVqpzClYo9944IIvlEkMKOi66JS_01auikLcRS_WSwyIs71CcQtNkSS2AZz9DIstKAwS1YFgy3LJ_sw5C-1X5asLjTz37JO4RUVNYnIfB9O7D9Ycf7u4PJ_ITjOej4VhKPgiah59yQETMRDpuFwmU9QIjEZgJv3REAeadztXxc2f9_nAoYBsEYwrNG_kooxUFQob2wWLN-t2dQ_Qi9TzOIF4N_D7OYzFZyNqeCPBgiJO2ghd1BatS85QjYvBNduRxmef9hZ-0iDh8ST4Bh5cVQjrkCHxOVllcT-B7bwgPDh3MfdMgomkaksjRYVMlpfHRWrb2VPxOnDruL3ya9BdNL6sc2WsYjJ6UyxKDswtaQ_K0QoPoqbPmM7_vBdEp_-T2z_zuqjWFLAf8yyX67YCH02Shdu1ywGI68WsxIsAi0YhuShRXRstlmOTL5bsCwLjGsn75h2pFgwglgE3oHbDAANusZk6wm2BWlleJIPaXPDvtktFQdXG0LHxOHLUYCAZjojGOf3OHiU0XsrPIeKbhydyW5qRki-FYJCOCliKduWhvocBmEGytJK4RZysM-9tEFw_qAuYewfL__t__I8tjRLw8xXHaH9QQnDh6y_Hr6qC_CPpw-U2dL_RSpJ7DUOv5jYZ0OYJdjLqeqsSPlwis8lRBW_TRZKXjgwDbb3iVsMvqBWVsSqkpMvzJkR5l04RUr08m06tTHEYDjylGgxpT8Bcg_M-S7JqK8AY2rOEEcYbv_Os1Ys9A6_Pra78fPcFrsBgvBDos6Mcye1WShiOU4bS2W6h95UZ40zpjTb_h4Qpx1b4eDGOaYJofI6nmR0n1R90fIikm74m8T7N7-bHS4YPc9-XB4w_IL5j8I_XPmZrHXpp1s9lY2zhe_nG9mLaHyG9eetKoelLee7JS6ckvNnoyB_cEJvdkeO3J64qeLIx7MnX15K1ET95Z9CzBjh73-578qKe1xp54rXJewvG635WQfjLTkyGmZU9VtycOPLDOkbO7Fq6tsJ53ePDWuLEC98yebHd6snfoydKwJ82kJy8EelJGepL-PcnJE049IpV3nMhjwq7Os8l82j9w6VnvLDEoKTY6W579xmeeoa8kaKJLkH55_PHsY_o7Hq1yQlSvIltmxdmyLCrTOyPMdYfh6my51rEz9UMvrd4UOvF3f_8Pd27AsA)

graph LR
  subgraph legend
     label(($LABEL))
	 issue[$ISSUE]
	 label --> issue
     I-unsound
	 E-needs-test
	 E-needs-mcve
	 E-needs-bisect
  end

  A-frontend((A-frontend)) & A-lint((A-lint)) --> i70819["#70819: 'forbid overwritten<br> by later allow on the same scope level'"]


  A-inference((A-inference)) --> i80816["#80816: 'regression 1.49: problem<br> with autoderef and trait method'"]
  
  A-incr-comp((A-incr-comp)) --> i58368["#58368: 'A big regression in tokio-webpush-simple-opt'"]

  A-pin((A-pin)) --> i85099["#85099: 'A Pin unsoundness involving an impl DerefMut<br> for Pin<&dyn LocalTrait>'"]


  A-ffi((A-ffi)) --> i52652["#52652: 'Abort instead of unwinding past FFI functions'"]

A-NLL((A-NLL)) --> i53114["#53114: '[nll] `_` patterns should not count as borrows'"]

  F-const_generics((F-const_generics)) --> i74713["#74713: 'debug-assertion ICE:<br> assertion failed:< !value.has_escaping_bound_vars()'"]

  F-unsized_locals((F-unsized_locals)) --> i71416["#71416: 'unsized_locals fails to<br> uphold alignment requirements'"]

  A-macros((A-macros)) --> i81211["#81211: 'Method resolution error'"]
  A-edition-2021((A-edition-2021)) & A-macros --> i92137["#92137: 'unreachable!(&quot;{}&quot;) works on Rust 2021'"]

  A-debuginfo((A-debuginfo)) --> i90357["#90357: 'DWARF info for static vars<br> in lib crates has stopped being produced reliably in LTO builds'"]


  A-const-eval((A-const-eval)) & A-diagnostics((A-diagnostics)) --> i85155["#85155: 'Improve diagnostics<br> of evaluation of constant value failed errors'"]
  A-suggestion-diagnostics((A-suggestion-diagnostics)) & A-diagnostics  --> i90286["#90286: 'regression: rustc suggests<br> .as_ref() at incorrect location and other spans have regressed'"]
  A-diagnostics --> i53934["#53934: 'disambiguate between multiple suggestions and a single<br> multi-span suggestion; or, JSON error format is not round-trippable'"]
  A-diagnostics --> i75791["#75791: 'diagnostics: one type is<br> more general for two identical types?'"] --- A-lifetimes
  A-diagnostics & A-proc-macros((A-proc-macros)) --> i76360["#76360: 'Error reporting from attribute macros regressed in 1.46.0'"]

  A-const-generics((A-const-generics)) --> i88888["#88888: 'stability break in rust 1.54: generic Self<br> types are currently not permitted in anonymous constants'"] --- A-simd((A-simd))



    A-associated-items((A-associated-items)) --> i78893["#78893: 'No error reported when a generic parameter<br> doesn't meeting the requirement of an associated type'"]
A-traits((A-traits)) --> i79168["#79168: 'Rustc hangs on recursive traits'"]
  A-associated-items --> i80800["#80800: 'Trait object types aren’t enforcing<br> equality constraints on associated types of super-traits'"] ---- A-typesystem((A-typesystem))
  i80800 --- A-dst((A-dst))
  A-associated-items & A-lifetimes --> i84533["#84533: 'Functions, closures, and HRTB-trait-objects<br> can implement traits such that<br> validity of associated types is never checked.'"] --- A-typesystem & A-closures((A-closures)) & A-dst
  A-associated-items & A-async-await & A-lifetimes  --> i84366["#84366: ''static closures with non-'static<br> return type are unsound '"] --- A-typesystem & A-closures
  A-traits --> i78893
  A-traits --> i80800
  A-traits --> i84533
  A-traits --> i84366

  A-async-await((A-async-await)) & A-generators((A-generators)) & A-traits --> i64552["#64552: 'Lifetime bounds in auto trait impls prevent<br> trait from being implemented on generators'"]



   i67830["#67830: 'ICE on Higher-Ranked Trait Bound<br> for impl Trait in return position '"] --- A-impl-trait((A-impl-trait))


    A-lto((A-lto)) --> i83911["#82911: 'High memory use on Rust 1.51.0<br> with thin LTO and debuginfo'"]
  A-lto --> i91671["#91671: 'doc test Couldn't compile the test on aarch64 + LTO'"] ---- A-LLVM((A-LLVM))
  A-linkage((A-linkage)) & A-lto --> i81408["#81408: 'ACCESS_VIOLATION when<br> dereferencing once_cell::Lazy in closure with LTO'"]
  A-linkage --> i64340["#64340: 'Linking issue with Rust 1.37.0'"]
  A-linkage --> i66916["#66916: 'Emscripten builds broken on nightly?<br> (Linking errors in fresh 'hello world' crate)'"]
  A-linkage --> i75385["#75385: 'Link fails with actix-web<br> when using GNU linker's default linker script'"]
  A-linkage --> i78686["#78686: 'Linking with LLD fails (on Windows)<br> but only in 32-bit (due to space in path)'"]
  A-linkage --> i79609["#79609: 'Passing -C panic=abort still attempts to link<br> in libunwind when targeting i686-pc-windows-gnu on v1.44+'"] --- A-cross((A-cross)) & A-runtime((A-runtime))
  A-linkage --> i79791["#79791: 'Linker error/regression introduced with<br> rustc 1.48.0 on aarch64-unknown-linux-musl target'"]
  A-linkage --> i82151["#82151: 'undefined reference to linker<br> error when using dylibs'"]
  A-linkage --> i86712["#86712: 'Rust musl build segfaults<br> on startup when linked with LLD 12'"]
  A-linkage --> i65391["#65391: 'rust-lld since 1.38 overlaps .text with .rodata<br> for embedded arm target'"] ----- A-LLVM
                i70143["#70143: 'Locals aligned to<br> greater than page size can cause unsound behavior'"] ---- A-LLVM
  A-linkage --> i74657["#74657: 'linker-plugin-lto stopped working in Rust 1.45.0'"] ---- A-LLVM
  i75045["#75045: 'Compiling crate fixed for thumbv6m-none-eabi<br> with opt-level=s spins endlessly in llvm::SelectionDAG::Combine'"] ---- A-LLVM
 i76247["#76247: 'Large performance drop in compiled binary<br> in stable rust 1.45.2 vs 1.44.0 on x86_64 linux'"] ---- A-LLVM
 i79789["#79789: 'Segfault in 1.47.0 and newer stable<br> when building firecracker on aarch64'"] ---- A-LLVM
 i79865["#79865: 'Miscompilation of AVX2 code under --release'"] --- A-LLVM & A-simd
  i80309["#80309: Miscompilation when using<br> wrapping_sub/wrapping_add on pointer"] ---- A-LLVM
   i86106["#86106: 'String::new() is slower than &quot;&quot;.to_string'"] ---- A-LLVM
   i89609["#89609: 'CI: dist-s390x-linux build<br> went from 40min. to 160min with new LLVM pass manager'"] ---- A-LLVM

  A-lifetimes((A-lifetimes)) --> i84305["#84305: 'Opaque impl Trait return types<br> unsoundly ignoring lifetime constraints'"] --- A-typesystem & A-impl-trait & A-closures

  A-lifetimes --> i54895["#54895: 'ICE when using anonymization in impl Trait with HRTB'"] --- A-impl-trait
  A-async-await --> i68109["#68109: 'rustc panics when compiling code<br> that uses tokio's LocalSet'"]
  A-async-await --> i72837["#72837: 'Compiling simple (but long)<br> code with lots of async/await takes hours'"]
  A-incr-comp --> i76037["#76037: 'Disk cache ICE with big array'"]
  A-ffi --> i80127["#80127: 'rustc has wrong signature for C function<br> with 16-byte aligned stack argument in x86_64 Linux'"]
  A-incr-comp --> i84970["#84970: 'Unstable fingerprints tracking issue'"]
  A-lifetimes --> i86172["#86172: 'Compile error: static lifetime not satisfied but it is'"]
  A-proc-macros --> i90256["#90256: 'Rustc passes syntactically invalid<br> input to attribute macros'"]



  
  
   
  
  
  A-incr-comp --> i91807["#91807: 'Regression in Deref&lt;Target = dyn Fn(T)&gt;:<br> forcing query with already existing DepNode'"]
  
    A-traits --> i62742["#62742: 'ICE: Generic type alias to invalid type crashes<br> during type check on latest stable'"]
    A-lifetimes & A-traits --> i84591["#84591: 'HRTB on subtrait unsoundly provides<br> HTRB on supertrait with weaker implied bounds"] ---- A-typesystem
  A-traits --> i84857["#84857: 'auto trait candidate selection is unsound'"] ---- A-typesystem
  
  

  untagged((untagged)) --> i66805["#66805: 'thread 'rustc' panicked<br> at 'called Result::unwrap() on an Err value: DistinctSources'"]
  untagged --> i67497["#67497: 'Switching to opt-level=z on<br> i686-windows-msvc triggers STATUS_ACCESS_VIOLATION'"]
  untagged --> i68324["#68324: 'Exponential compile time for the amount of impls.'"]
  untagged --> i70022["#70022: 'Statics don't support<br> alignments larger than the page size"]
  untagged --> i79590["#79590: 'internal compiler error: unexpected panic'"]
  untagged --> i80951["#80951: 'regression 1.50: Duplicate symbol linker error'"]
  untagged --> i81827["#81827: 'rustc crashed when compiling the crafted input '"]
  untagged --> i84097["#84097: 'Disk usage 2-3x somewhere around 1.48'"]
  untagged --> i84873["#84873: 'Compile time+memory regression<br> between 1.49.0 and 1.50.0'"]
  untagged --> i86511["#86511: 'match an std::cmp::Ordering<br> generates less optimized code in nightly'"]
  untagged --> i85821["#85821: 'Emscripten wasm32 compilation broken'"]
  untagged --> i89195["#89195: 'Compilation appears to loop indefinitely'"]
  untagged --> i89241["#89241: 'Performance regression in<br> string interner benchmark (1.53.0 → 1.55.0)'"]
  untagged --> i89601["#89601: 'nightly-2021-09-03: Compiler hang<br> in project with a lot of axum crate routes'"]
  untagged --> i90180["#90180: 'Build won't complete for binary crate<br> under rustc 1.56.'"]
  untagged --> i90316["#90316: 'Rust 1.56 injected a valgrind<br> incompatibility for Linux/AAarch64 '"]
  untagged --> i91949["#91949: 'Exponential slowdown<br> in catching an overflow error.'"]

  class I-unsound unsound
  class E-needs-mcve needs_mcve_no_test
  class E-needs-test needs_test
  class E-needs-bisect needs_bisect

  classDef unsound fill:#f96
  class i52652,i67497,i70022,i70143,i71416,i79865,i80127,i80309,i80800,i81408,i84305,i84366,i84533,i4591,84857,i85099 unsound

  classDef needs_test fill:#4f0
  class i53114,i68109,i80309,i81827,i91949 needs_test
  
  classDef needs_mcve_no_test fill:#ff8
  class i66916,i67497,i68324,i74657,i79789,i79791,i80951,i83911,i89195,i89241,i89601 needs_mcve_no_test
  
  classDef needs_bisect fill:#ec0
  class i75860 needs_bisect