Brandon Chinn pushed to branch wip/T21730-import at Glasgow Haskell Compiler / GHC
Commits:
a1d92378 by Brandon Chinn at 2025-01-25T15:11:54-08:00
Fix for alex-3.5.2.0 (#25623)
This INLINE pragma for alexScanUser was added in 9.12, but then I
ported the change to alex in 3.5.2.0
(https://github.com/haskell/alex/pull/262).
I didn't realize that GHC errors on duplicate INLINE pragmas, so
this ended up being a breaking change.
This change should be backported into 9.12
- - - - -
62760367 by ARATA Mizuki at 2025-01-27T16:23:06-05:00
x86 NCG: Make MOVD's output format explicit
The old design led to inference of a wrong format,
losing upper bits of a vector register.
Fixes #25659
Co-authored-by: sheaf
DeprecationWarning: 'count' is passed as positional argument
- - - - - ef5470a2 by Ben Gamari at 2025-02-19T16:30:53+00:00 Cmm/Parser: Add surface syntax for Mul2 MachOps These are otherwise very hard to test in isolation. - - - - - 59b9307b by Cheng Shao at 2025-02-19T20:24:40-05:00 testsuite: fix InternalCounters test with +debug_ghc The `InternalCounters` test case fails when ghc is built with `+debug_ghc`. This patch skips it in that case and allows the testsuite to pass for the `+debug_ghc` flavour transformer. - - - - - aa69187d by M Farkas-Dyck at 2025-02-19T20:25:31-05:00 Scrub a use of `head` in `GHC.Driver.Make.downsweep_imports.checkDuplicates`. - - - - - 1c8e608a by Cheng Shao at 2025-02-21T20:18:41-05:00 compiler: use fromAscList when applicable This patch uses fromAscList (with O(n) complexity) instead of fromList (with O(nlogn) complexity) in certain Binary instances. It's safe to do so since the corresponding serialization logic is based on toList (same as toAscList). - - - - - 549e0aff by Ben Gamari at 2025-02-21T20:19:18-05:00 rts/linker/MachO: Mark internal symbols as static There is no reason why these should have external linkage. - - - - - fbf3d020 by Cheng Shao at 2025-02-22T07:41:01-05:00 wasm: bump dyld v8 heap size limit This patch sets `--max-old-space-size=65536` as wasm dyld shebang arguments to lessen v8 heap pressure in certain workloads that load the full ghc package. It doesn't really commit 64G memory but it does help reduce v8 gc overhead. - - - - - cb60da24 by Cheng Shao at 2025-02-22T07:41:01-05:00 wasm: fix dyld for shared libraries created by llvm 20.x This patch fixes wasm dyld script for shared libraries created by llvm 20.x. The __wasm_apply_data_relocs function is now optional and may be omitted for shared libraries without any runtime relocatable data segments, so only call __wasm_apply_data_relocs when it's present. - - - - - 7eea38c8 by Cheng Shao at 2025-02-22T07:41:37-05:00 driver: fix wasm backend sysroot lookup logic when -flto is passed For the wasm target, the driver calls `wasm32-wasi-clang --print-search-dirs` and finds the sysroot directory that contains libc.so etc, then passes the directory path to dyld. However, when GHC is configured with -flto as a part of C/C++ compiler flags, the clang driver would insert a llvm-lto specific directory in the --print-search-dirs output and the driver didn't take that into account. This patch fixes it and always selects the non-lto sysroot directory to be passed to dyld. This is one small step towards supporting building all cbits with lto for wasm. - - - - - f3bfe31e by Cheng Shao at 2025-02-23T14:06:25-05:00 wasm: add Note [Variable passing in JSFFI] as !13583 follow up This patch adds a note to explain how the magic variables like `__ghc_wasm_jsffi_dyld` are brought into scope of JSFFI code snippets, as follow up work of !13583. - - - - - c318be56 by Cheng Shao at 2025-02-23T14:07:02-05:00 libffi: update to 3.4.7 Bumps libffi submodule. - - - - - 33aca30f by sheaf at 2025-02-25T08:58:46-05:00 LLVM: account for register type in funPrologue We were not properly accounting for the live register type of global registers in GHC.CmmToLlvm.CodeGen.funPrologue. This meant that we could allocated a register at type <4 x i32> but try to write to it at type <8 x i16>, which LLVM doesn't much like. This patch fixes that by inserting intermerdiate casts when necessary. Fixes #25730 - - - - - 0eb58b0e by sheaf at 2025-02-25T08:59:29-05:00 base: make Data.List.NonEmpty.unzip match Data.List This commit makes Data.List.NonEmpty.unzip match the implementation of Data.List, as was suggested in approved CLC proposal #107. - - - - - f4da90f1 by Matthew Pickering at 2025-02-25T14:11:21-05:00 interpreter: Fix underflow frame lookups BCOs can be nested, resulting in nested BCO stack frames where the inner most stack frame can refer to variables stored on earlier stack frames via the PUSH_L instruction. |---------| | BCO_1 | -<-┐ |---------| ......... | |---------| | PUSH_L <n> | BCO_N | ->-┘ |---------| Here BCO_N is syntactically nested within the code for BCO_1 and will result in code that references the prior stack frame of BCO_1 for some of it's local variables. If a stack overflow happens between the creation of the stack frame for BCO_1 and BCO_N the RTS might move BCO_N to a new stack chunk while leaving BCO_1 in place, invalidating a simple offset based reference to the outer stack frames. Therefore `ReadSpW` first performs a bounds check to ensure that accesses onto the stack will succeed. If the target address would not be a valid location for the current stack chunk then `slow_spw` function is called, which dereferences the underflow frame to adjust the offset before performing the lookup. ┌->--x | CHK_1 | | CHK_2 | | | |---------| |---------| | └-> | BCO_1 | | UD_FLOW | -- x |---------| |---------| | | ...... | | |---------| | PUSH_L <n> | BCO_ N | ->-┘ |---------| Fixes #25750 - - - - - c3f2d284 by Vladislav Zavialov at 2025-02-25T14:11:58-05:00 Remove ArgPatBuilder ArgPatBuilder in Parser/PostProcess.hs became redundant with the introduction of InvisPat (36a75b80eb). This small refactoring removes it. - - - - - 0f2241e9 by sheaf at 2025-02-25T19:23:21-05:00 Propagate long distance info to guarded let binds This commit ensures that we propagate the enclosing long distance information to let bindings inside guards, in order to get accurate pattern-match checking warnings, in particular incomplete record selector warnings. Example: data D = K0 | K1 { fld :: Int } f :: D -> Int f d@(K1 {}) | let i = fld d = i f _ = 3 We now correctly recognise that the field selector 'fld' cannot fail, due to the outer pattern match which guarantees that the value 'd' has the field 'fld'. Fixes #25749 - - - - - 64b0d4d0 by Fangyi Zhou at 2025-02-25T19:24:07-05:00 wasm: use primitive opcodes for fabs and sqrt - Add new `WasmInstr` constructor `WasmSqrt` for sqrt, corresponding to primitivie operations in wasm. - When lowering CallishMachOp, use `WasmAbs` and `WasmSqrt` for F32 and F64 fabs and sqrt. - - - - - 272eaef0 by Cheng Shao at 2025-02-25T19:24:43-05:00 hadrian: enable building stage1 haddock for cross ghc This commit enables building stage1 haddock for cross ghc. Without this change, hadrian would panic with "Unknown program" error when building the _build/stage1/bin/cross-prefix-haddock program needed by the docs-haddock target, even if it only needs to copy from _build/stage0/bin/cross-prefix-haddock. - - - - - a794e733 by Cheng Shao at 2025-02-25T19:24:43-05:00 hadrian: enable building docs for cross targets Hadrian used to omit the docs target as a part of binary-dist-dir for cross targets. This commit enables docs to be built as a part of cross bindists and it works just fine in CI. - - - - - 6dba56e1 by Cheng Shao at 2025-02-25T19:24:43-05:00 ci: build haddock/sphinx-html for wasm jobs This commit enables building haddock & sphinx-html documentation for wasm targets. The docs are useful for end users and should be tested in CI. I've omitted pdf & manpage generation for the wasm target; I've never found the pdf version of docs to be useful, and the manpage only contains `ghc.1` where we really want `wasm32-wasi-ghc.1` but that should be a separate issue to fix. - - - - - 2d6a63ab by Cheng Shao at 2025-02-25T19:25:20-05:00 ghci: remove unused showBreakArray function GHCi.BreakArray.showBreakArray is not used anywhere, hence the housecleaning. - - - - - b228fcb5 by Cheng Shao at 2025-02-25T19:25:57-05:00 ghc-heap: fix HalfWord incompatible Binary instances for cross GHC ghc-heap defines HalfWord as Word32/Word16 depending on host word size. For cross GHC with different host/target word sizes, the Binary instances are incompatible and breaks iserv serialization of any message type that involves HalfWord, breaking the ghci debugger. This patch fixes the issue and has been tested to fix ghci debugger functionality of the wasm backend. Fixes #25420 #25781. - - - - - ec02f8c2 by Rodrigo Mesquita at 2025-02-26T11:03:38-05:00 ghci-debugger: display thunks provenance if avail Improves reporting on ghci breakpoints when IPE information is available by printing, next to the thunk, the source file and src span where the thunk originated. Closes #25746 - - - - - 6acaff2b by Vladislav Zavialov at 2025-02-26T11:04:15-05:00 Tidy up error messages for TypeAbstractions 1. Print the '@' symbol before invisible patterns and improve phrasing: T24557c.hs:8:4: error: [GHC-11983] - Invisible type pattern pat is not allowed here + Illegal invisible type pattern: @pat + An invisible type pattern must occur in an argument position. 2. Use a single error code for all type abstractions deemed illegal due to the TypeAbstractions extension being disabled. Before this change: * [GHC-78249] was used in function equations, lambdas * [GHC-17916] was used in constructor patterns After this change: * [GHC-78249] is used to report all illegal type abstractions * [GHC-17916] is no longer used There was no reason for both error codes to exist and this distinction was a source of complexity in GHC/Tc/Errors/* and GHC/Rename/Pat.hs 3. Group the different "invisible type pattern" error constructors under a single parent constructor, TcRnIllegalInvisibleTypePattern containing BadInvisPatReason - - - - - 1ce9502e by Ben Gamari at 2025-02-27T04:48:29-05:00 haddock/doc: Drop version and release We currently have no way of keeping this up-to-date with the project version - - - - - 7f358f25 by Matthew Pickering at 2025-02-27T04:49:06-05:00 testsuite: Add test for :steplocal performance This adds a simple test which exercises #25779 - - - - - a6a3ffa6 by Sven Tennie at 2025-02-27T23:34:47-05:00 Do not deallocate stack for jump/switch table jumps Though the name is misleading, we consider them to be branching. For branch instructions we do not deallocate (parts of) the stack, but keep the stack pointer (sp) intact. - - - - - 39e51ddb by Sven Tennie at 2025-02-27T23:34:47-05:00 Add reproducer for dealloc instructions in switch table jump expressions (#25733) Measures taken to make the test stable: - Use 'a' as variable prefix, because X86 32bit stumbled over the variable name 'i386' - Flush stdout to make test output deterministic - Use type annotations to support 32bit archs - - - - - d427df93 by Sylvain Henry at 2025-02-27T23:35:30-05:00 Remove redundant location strings in expectJust and friends (#25743) Now we can use HasDebugCallStack instead to avoid cluttering the code with strings and to avoid maintaining those strings (e.g. renaming them when functions are renamed...). - - - - - 90f1190e by Ben Gamari at 2025-02-28T20:53:53-05:00 compiler: Add export list to GHC.SysTools.Tasks - - - - - ec826009 by Ben Gamari at 2025-02-28T20:53:53-05:00 compiler: Pass --target to llvm-as As noted in #25793, this is necessary due to potential ambiguity on Apple machines with Rosetta. - - - - - 9c1647d1 by Andreas Klebinger at 2025-02-28T20:54:30-05:00 cmmMachOpFoldM: Add missing pattern matches for bitcasts. Fixes #25771 - - - - - 3b78e139 by John Ericson at 2025-03-03T15:27:39-05:00 Remove most of `GHC.Internal.Pack` Since bd82ac9f4716e28b185758ae514691d5a50c003f when `GHC.Pack` was deleted, it is no longer used except for one function by the RTS. - - - - - b4fe0850 by Rodrigo Mesquita at 2025-03-03T15:28:16-05:00 ghci: Don't set virtualCWD on every iteration The calls to withVirtualCWD were introduced to fix #2973, but this bug is no longer reproducible, even when `withVirtualCWD` is dropped. This cleanup was originally motivated by the performance of :steplocal, but the performance problem has now been fixed at its root in the next commit. Even then, `withVirtualCWD` seems to now be an unnecessary artifact, and removing it simplifies the interpreter with no apparent drawbacks (testsuite is also happy with this change) - - - - - 73ba1e6e by Rodrigo Mesquita at 2025-03-03T15:28:16-05:00 ghci debugger: improve break/resume control flow After interpreting bytecode (`evalStmt`), we may want to hand off control to "GHCi.UI" in order to display an interactive break prompt: 1. When an /active/ breakpoint (one set with :break ...) is hit 2. At any breakpoint, when using :step from a breakpoint 3. At any breakpoint in the same function f, when :steplocal is called from a breakpoint in f 4. At any breakpoint in the same module, when :stepmodule is used Whether to pass control to the UI is now fully determined by `handleRunStatus` which transforms an `EvalStatus_` into an `ExecResult`. When `ExecBreak` is returned from `handleRunStatus` to GHCi, it always means GHCi breaks. `handleRunStatus` determines whether to loop and resume evaluation right away, or when to return to GHCi (by returning `ExecBreak` or `ExecComplete`). - (1) is queried using the `BreakpointStatus` message (the `breakpointStatus` call) - (2,3,4) are determined by the predicate `breakHere step span`, which inspects the improved `SingleStep` type to determine whether we care about this breakpoint even if it is not active. This refactor solves two big performance problems with the previous control flow: - We no longer call `withArgs/withProgram` repeatedly in the break/resume loop, but rather just once "at the top". - We now avoid computing the expensive `bindLocalsAtBreakpoint` for breakpoints we'd never inspect. In the interpreter_steplocal test added, calling `:steplocal` after breaking on `main = fib 25` now takes 12 seconds rather than 49 seconds on my machine. ``` interpreter_steplocal(ghci) ghc/alloc 6,124,821,176 540,181,392 -91.2% GOOD ``` Fixes #25779 ------------------------- Metric Decrease: interpreter_steplocal ------------------------- - - - - - c78d8f55 by Cheng Shao at 2025-03-03T20:54:41+00:00 rts: fix top handler closure type signatures This commit fixes the runIO/runNonIO closure type signatures in the RTS which should be extern StgClosure. This allows us to remove an unnecessary type cast in the C foreign desugaring logic, as well as unneeded complications of JSFFI desugaring logic that also needs to generate C stubs that may refer to those top handler closures. Otherwise, we'll have to take special care to avoid generating "extern StgClosure" declarations for them as we would for other closures, just to avoid conflicting type signature error at stub compile time. - - - - - a204df3a by Cheng Shao at 2025-03-03T20:54:41+00:00 compiler: allow arbitrary label string for JSFFI exports This commit allows arbitrary label string to appear in a foreign export declaration, as long as the calling convention is javascript. Well, doesn't make sense to enforce it's a C function symbol for a JSFFI declaration anyway, and it gets in the way of implementing the "sync" flavour of exports. - - - - - 03ebab52 by Cheng Shao at 2025-03-03T20:54:41+00:00 compiler: wasm backend JSFFI sync exports This commit implements the synchronous flavour of the wasm backend JSFFI exports: - `foreign export javascript "foo sync"` exports a top-level Haskell binding as a synchronous JS function - `foreign import javascript "wrapper sync"` dynamically exports a Haskell function closure as a synchronous JS function - `foreign import javascript unsafe` is now re-entrant by lowering to a safe ccall - Also fix the issue that JSFFI dynamic exports didn't really work in TH & ghci (#25473) - - - - - b6ae908b by Cheng Shao at 2025-03-03T20:54:41+00:00 testsuite: test wasm backend JSFFI sync exports This commit repurposes some existing JSFFI test cases to make them cover JSFFI sync exports as well. - - - - - edae2874 by Cheng Shao at 2025-03-03T20:54:41+00:00 docs: document wasm backend JSFFI sync exports This commit updates wasm backend documentation to reflect the new JSFFI sync exports feature. - - - - - 9b54eecb by Cheng Shao at 2025-03-03T20:56:21+00:00 wasm: add error message to WouldBlockException This commit attaches an error message to WouldBlockException, for now the error message consists of the JS async import code snippet that thunk is trying to block for. This is useful for debugging synchronous callbacks that accidentally call an async JS function. - - - - - c331eebf by Cheng Shao at 2025-03-04T09:11:45-05:00 compiler: avoid overwriting existing writers in putWithTables This patch makes `putWithTables` avoid overwriting all existing UserData writers in the handle. This is crucial for GHC API users that use putWithUserData/getWithUserData for serialization logic that involve Names. - - - - - e9b7802b by Matthew Pickering at 2025-03-04T09:12:21-05:00 ghci: Serialise mi_top_env When loading core from interface files (or from a bytecode object in future) it's important to store what the top-level context of a module is. Otherwise, when you load the module into GHCi from the interface files, only exported identifiers from the top-level module are in scope on the repl. See the added test which demonstrates what this enables. The context at the GHCi prompt is everything that's in-scope in the TopEnvIface module. Since TopEnvIface imports identifier "a", we can evaluate "a" in the repl. In addition to all this, we can use this information in order to implement reifyModule in a more principled manner. This becomes even more important when you're debugging and what to set break-points on functions which are not imported. - - - - - 73e02068 by Matthew Pickering at 2025-03-04T09:12:21-05:00 Implement reifyModule in terms of mi_top_env mi_top_env provides precisely the information that reifyModule needs, the user written imports. This is important as it unblocks !9604 and #22188 Fixes #8489 - - - - - 0a99825d by Ben Gamari at 2025-03-04T09:12:57-05:00 hadrian: Refactor handling of test suite environment Previously we would set the environment variables used to run the testsuite driver using `setEnv` to set them in the Hadrian process. While looking into failures of a fix to #25752 I noticed this and took the opportunity to refactor. - - - - - 7ca72844 by Alan Zimmerman at 2025-03-04T09:13:34-05:00 [EPA] Sync with the ghc-exactprint repo This brings it into line with the changes in https://hackage.haskell.org/package/ghc-exactprint-1.12.0.0 But also keeps the latest changes from master. - - - - - 8f6cc90c by Matthew Pickering at 2025-03-05T04:48:02-05:00 perf: Speed up the bytecode assembler This commit contains a number of optimisations to the bytecode assembler. In programs which generate a large amount of bytecode, the assembler is called a lot of times on many instructions. 1. Specialise the assembleI function for the two intepreters to avoid having to materialise the intermediate free-monad like structure. 2. Directly compute the UArray and SmallArray needed rather than going via the intermediate SizedSeq 3. Use optimised monads 4. Define unrolled "any" and "mapM6" functions which can be inlined and avoid calling recursive functions. The resulting generated code is much more direct. Before: ./ByteCodeAsm /home/matt/ghc-profiling-light/_build/stage1/lib/ +RTS -s 48,923,125,664 bytes allocated in the heap 678,221,152 bytes copied during GC 395,648 bytes maximum residency (2 sample(s)) 50,040 bytes maximum slop 6 MiB total memory in use (0 MiB lost due to fragmentation) Tot time (elapsed) Avg pause Max pause Gen 0 11731 colls, 0 par 0.419s 0.425s 0.0000s 0.0004s Gen 1 2 colls, 0 par 0.001s 0.001s 0.0007s 0.0012s INIT time 0.000s ( 0.000s elapsed) MUT time 6.466s ( 6.484s elapsed) GC time 0.421s ( 0.426s elapsed) EXIT time 0.000s ( 0.000s elapsed) Total time 6.887s ( 6.910s elapsed) After: 1,518,321,200 bytes allocated in the heap 4,299,552 bytes copied during GC 322,288 bytes maximum residency (2 sample(s)) 50,280 bytes maximum slop 6 MiB total memory in use (0 MiB lost due to fragmentation) Tot time (elapsed) Avg pause Max pause Gen 0 369 colls, 0 par 0.003s 0.003s 0.0000s 0.0002s Gen 1 2 colls, 0 par 0.001s 0.001s 0.0007s 0.0012s INIT time 0.001s ( 0.001s elapsed) MUT time 0.465s ( 0.466s elapsed) GC time 0.004s ( 0.004s elapsed) EXIT time 0.000s ( 0.000s elapsed) Total time 0.470s ( 0.471s elapsed) - - - - - f2d43e11 by Teo Camarasu at 2025-03-05T04:48:40-05:00 ghc-boot-th: expose all TH packages from proper GHC.Boot.* modules Previously we defined some modules here in the GHC.Internal namespace. Others were merely re-exposed from GHC.Internal. Re-exposed modules weren't handled correctly by Haddock, so the Haddocks for the `template-haskell` library couldn't see them. This change also makes the home package of these modules a bit clearer. Work towards #25705 - - - - - 91ef82df by Teo Camarasu at 2025-03-05T04:48:40-05:00 ghc-boot-th: fix synopsis formatting `@...@` syntax doesn't seem to work in synposes and is just kept by Haddock verbatim. - - - - - eb9fe1ec by Brandon Chinn at 2025-03-05T04:49:17-05:00 Collapse string gaps as \& (#25784) In 9.10, "\65\ \0" would result in "A0", but in 9.12, it results in "\650", due to the string refactoring I did in !13128. Previously, we were resolving escape codes and collapsing string gaps as we come across them, but after the refactor, string processing is broken out into phases, which is both more readable and useful for multiline strings. - - - - - 8037f487 by Cheng Shao at 2025-03-05T04:49:54-05:00 ghc-experimental: make JSVal abstract in GHC.Wasm.Prim This commit makes JSVal an abstract type in the export list of GHC.Wasm.Prim. JSVal's internal representation is supposed to be a non user facing implementation detail subject to change at any time. We should only expose things that are newtypes of JSVal, not JSVal itself. - - - - - 4f342431 by Cheng Shao at 2025-03-05T04:49:54-05:00 wasm: make JSVal internal Weak# point to lifted JSVal JSVal has an internal Weak# with the unlifted JSVal# object as key to arrange its builtin finalization logic. The Weak# used to designate Unit_closure as a dummy value; now this commit designates the lifted JSVal closure as the Weak# value. This allows the implementation of mkWeakJSVal which can be used to observe the liveliness of a JSVal and attach a user-specified finalizer. - - - - - 55af20e6 by Cheng Shao at 2025-03-05T04:49:54-05:00 ghc-experimental: add mkWeakJSVal This commit adds a mkWeakJSVal function that can be used to set up a Weak pointer with a JSVal key to observe the key's lifetime and optionally attach a finalizer. - - - - - 8273d7d1 by Matthew Pickering at 2025-03-05T04:50:30-05:00 simplifier: Zap Id unfoldings before constructing InScopeSet in simpleOptExpr Care must be taken to remove unfoldings from `Var`s collected by exprFreeVars before using them to construct an in-scope set hence `zapIdUnfolding` in `init_subst`. Consider calling `simpleOptExpr` on an expression like ``` case x of (a,b) -> (x,a) ``` * One of those two occurrences of x has an unfolding (the one in (x,a), with unfolding x = (a,b)) and the other does not. (Inside a case GHC adds unfolding-info to the scrutinee's Id.) * But exprFreeVars just builds a set, so it's a bit random which occurrence is collected. * Then simpleOptExpr replaces each occurrence of x with the one in the in-scope set. * Bad bad bad: then the x in case x of ... may be replaced with a version that has an unfolding. Fixes #25790 - - - - - 07fe6d1d by Rodrigo Mesquita at 2025-03-05T04:51:07-05:00 docs: Fix ghci :doc documentation Fixes #25799 - - - - - a510b861 by Matthew Pickering at 2025-03-06T11:43:23+00:00 Add flag to control whether self-recompilation information is written to interface This patch adds the flag -fwrite-if-self-recomp which controls whether interface files contain the information necessary to answer the question: Do I need to recompile myself or is this current interface file suitable? Why? Most packages are only built once either by a distribution or cabal and then placed into an immutable store, after which we will never ask this question. Therefore we can derive two benefits from omitting this information. * Primary motivation: It vastly reduces the surface area for creating non-deterministic interface files. See issue #10424 which motivated a proper fix to that issue. Distributions have long contained versions of GHC which just have broken self-recompilation checking (in order to get deterministic interface files). * Secondary motivation: This reduces the size of interface files slightly.. the `mi_usages` field can be quite big but probably this isn't such a great benefit. * Third motivation: Conceptually clarity about which parts of an interface file are used in order to **communicate** with subsequent packages about the **interface** for a module. And which parts are used to self-communicate during recompilation checking. The main tracking issue is #22188 but fixes issues such as #10424 in a proper way. - - - - - 5b05c27b by Matthew Pickering at 2025-03-06T11:43:23+00:00 Disable self recomp in release flavour The interface files that we distribute should not contain any information which is used by the recompilation checking logic since source file will never be compiled again. I am not 100% sure this won't cause unexpected issues, there many be downstream consumers which are incorrectly using the information from interfaces, but this commit can be reverted if we detect issues. - - - - - 1d4c9824 by Matthew Craven at 2025-03-06T18:11:59-05:00 Cmm: Add surface syntax for Word/Float bitcast ops - - - - - 25c4a2a2 by Matthew Craven at 2025-03-06T18:11:59-05:00 Cmm: Add constant-folding for Word->Float bitcasts - - - - - 30bdea67 by Matthew Craven at 2025-03-06T18:11:59-05:00 Add tests for #25771 - - - - - 44bf5fa1 by Matthew Pickering at 2025-03-07T13:48:18+00:00 iface: Store flags in interface files When reporting the reason why a module is recompiled (using `-dump-hi-diffs`), it is much more informative to inform the user about which flag exactly has changed, rather than just an opaque reference to a hash. Now, when the user enables `-fwrite-if-self-recomp-flags` there is a difference the precise part of the flags is reported: ``` codegen flags changed: before: [Opt_NoTypeableBinds, Opt_OmitYields] after: [Opt_NoTypeableBinds, Opt_OmitYields, Opt_DictsStrict] ``` Fixes #25571 - - - - - 324222bd by Oleg Grenrus at 2025-03-08T08:50:18-05:00 Run fix-whitespace on compiler/ https://hackage.haskell.org/package/fix-whitespace IMO this should be included into lint suite - - - - - 1e53277a by sheaf at 2025-03-08T16:32:25-05:00 Allow defaulting of representational equalities This commit generalises the defaulting of equality constraints that was introduced in 663daf8d (with follow-up in 6863503c) to allow the defaulting of *representational* equality constraints. Now we default a representational equality ty1 ~R# ty2 by unifying ty1 ~# ty2. This allows the following defaulting to take place: - Coercible alpha[tau] Int ==> alpha := Int - Coercible (IO beta[tau]) (IO Char) ==> beta := Char See Note [Defaulting representational equalities] in GHC.Tc.Solver.Default for more details. Fixes #21003 - - - - - d6c40afc by Andreas Klebinger at 2025-03-08T16:33:02-05:00 Revert "Use `Infinite` in unique generation, and clean up some other partial uni patterns as well." This reverts commit 643dd3d86968c527ba07ece9cc337728dbdfe2a0. As described in #25817 this commit introduced a subtle bug in AArch64 code generation. So for the time being I will simply revert it wholesale. - - - - - 68310e11 by Andreas Klebinger at 2025-03-08T16:33:39-05:00 Properly describe acceptance window for stat tests. The relative metric is already in %, so no need to multiply by 100. - - - - - cca68421 by Cheng Shao at 2025-03-08T22:04:42-05:00 wasm: do not use wasm type reflection in dyld The wasm dynamic linker used to depend on v8's experimental wasm type reflection support to generate stub functions when treating GOT.func items that aren't exported by any loaded library yet. However, as we work towards wasm ghci browser mode (#25399), we need to ensure the wasm dyld logic is portable across browsers. So this commit removes the usage of wasm type reflection in wasm dyld, and it shall only be added many months later when this feature is widely available in browsers. - - - - - 75fcc5c9 by Cheng Shao at 2025-03-08T22:05:19-05:00 wasm: don't create a wasm global for dyld poison There's a much more efficient way to convert an unsigned i32 to a signed one. Thanks, o3-mini-high. - - - - - fd40eaa1 by Cheng Shao at 2025-03-08T22:05:19-05:00 wasm: revamp JSFFI internal implementation and documentation This patch revamps the wasm backend's JSFFI internal implementation and documentation: - `JSValManager` logic to allocate a key is simplified to simple bumping. According to experiments with all major browsers, the internal `Map` would overflow the heap much earlier before we really exhaust the 32-bit key space, so there's no point in the extra complexity. - `freeJSVal` is now idempotent and safe to call more than once. This is achieved by attaching the `StablePtr#` to the `JSVal#` closure and nullifying it when calling `freeJSVal`, so the same stable pointer cannot be double freed. - `mkWeakJSVal` no longer exposes the internal `Weak#` pointer and always creates a new `Weak#` on the fly. Otherwise by finalizing that `Weak#`, user could accidentally drop the `JSVal`, but `mkWeakJSVal` is only supposed to create a `Weak` that observes the `JSVal`'s liveliness without actually interfering it. - `PromisePendingException` is no longer exported since it's never meant to be caught by user code; it's a severe bug if it's actually raised at runtime. - Everything exported by user-facing `GHC.Wasm.Prim` now has proper haddock documentation. - Note [JSVal representation for wasm] has been updated to reflect the new JSVal# memory layout. - - - - - cbae3708 by Ben Gamari at 2025-03-11T06:09:58-04:00 users guide: Fix typo - - - - - 1951eb7a by Ben Gamari at 2025-03-11T06:10:35-04:00 testsuite: Don't count fragile passes as failures in JUnit output As noted in #25806, the testsuite driver's JUnit output previously considered passes of fragile tests to be failures. Fix this. Closes #25806. - - - - - 589f40b9 by Matthew Pickering at 2025-03-11T06:11:11-04:00 Use panic rather than error in expectJust Otherwise, we would not get a callstack printed out when the exception occurs. Fixes #25829 - - - - - d450e88e by sheaf at 2025-03-11T06:42:59-04:00 Solve Wanted quantified constraints from Givens This commit ensures we directly solve Wanted quantified constraints from matching inert Given quantified constraints,instead of going through the trouble of emitting an implication constraint and processing that. This is not just an optimisation; it makes our lives easier when generating RULEs for specialisation. See Note [Solving Wanted QCs from Given QCs] for details Fixes #25758 - - - - - 48daaf53 by Ben Gamari at 2025-03-11T06:42:59-04:00 testsuite: Add testcase for #25577 - - - - - d2ffb0ce by Ben Gamari at 2025-03-11T06:42:59-04:00 testsuite/ghc-api: Eliminate Makefile usage from various GHC API tests These tests can be expressed perfectly well using the testsuite driver itself. - - - - - 2275b642 by Ben Gamari at 2025-03-11T06:42:59-04:00 rts/linker/MachO: Assert that GOT relocations have GOT entries In #25577 we found that some GOT relocation types were not being given relocation entries. Add assertions to catch this sort of failure in the future. - - - - - 8c96bcb4 by Ben Gamari at 2025-03-11T06:42:59-04:00 rts/linker/MachO: Account for internal GOT references in GOT construction Previously we failed to give GOT slots to symbols which were referred to by GOT relocations in the same module. This lead to #25577. Fix this by explicitly traversing relocation lists and maintaining a `needs_got` flag for each symbol. Fixes #25577. - - - - - 7b84c588 by Vladislav Zavialov at 2025-03-11T06:43:02-04:00 One list in ConPat (part of #25127) This patch changes PrefixCon to use one list instead of two: -data HsConDetails tyarg arg rec - = PrefixCon [tyarg] [arg] +data HsConDetails arg rec + = PrefixCon [arg] | RecCon rec | InfixCon arg arg The [tyarg] list is now gone. To understand the effect of this change, recall that there are three instantiations of HsConDetails: 1. type HsConPatDetails p = HsConDetails (HsConPatTyArg (NoGhcTc p)) -- tyarg (LPat p) -- arg (HsRecFields p (LPat p)) -- rec 2. type HsConDeclH98Details pass = HsConDetails Void -- tyarg (HsScaled pass (LBangType pass)) -- arg (XRec pass [LConDeclField pass]) -- rec 3. type HsPatSynDetails pass = HsConDetails Void -- tyarg (LIdP pass) -- arg [RecordPatSynField pass] -- rec In cases (2) and (3), tyarg was instantiated to Void, so the [tyarg] list was always empty. Its removal is basically a no-op. The interesting case is (1), which is used in ConPat to represent pattern matching of the form (MkE @tp1 @tp2 p1 p2). With this patch, its representation is changed as follows: ConPat "MkE" [tp1, tp2] [p1, p2] -- old ConPat "MkE" [InvisP tp1, InvisP tp2, p1, p2] -- new The new mixed-list representation is consintent with lambdas, where InvisP is already used to deal with \ @tp1 @tp2 p1 p2 -> body. The immediate effect of the new representation is an improvement to error messages. Consider the pattern (Con x @t y). Previously it resulted in a parse error because @t could not occur after x. Now it is reported as [GHC-14964]. Test case: TyAppPat_MisplacedApplication. In the long term, this is intended as preparation for #18389 and #25127, which would make (Con x @t y) potentially valid, e.g. if its type is Con :: forall a -> forall b. (a, b) -> T The TH AST is left unchanged for the moment to avoid breakage. - - - - - cce869ea by Vladislav Zavialov at 2025-03-11T06:43:02-04:00 Error message with EmptyCase and RequiredTypeArguments (#25004) Fix a panic triggered by a combination of \case{} and forall t -> ghci> let f :: forall (xs :: Type) -> (); f = \case {} panic! (the 'impossible' happened) GHC version 9.10.1: Util: only The new error message looks like this: ghci> let f :: forall (xs :: Type) -> (); f = \case {} <interactive>:5:41: error: [GHC-48010] • Empty list of alternatives in \case expression checked against a forall-type: forall xs -> ... This is achieved as follows: * A new data type, BadEmptyCaseReason, is now used to describe why an empty case has been rejected. Used in TcRnEmptyCase. * HsMatchContextRn is passed to tcMatches, so that the type checker can attach the syntactic context to the error message. * tcMatches now rejects type arguments if the list of alternatives is empty. This is what fixes the bug. - - - - - 37d8b50b by sheaf at 2025-03-11T06:43:06-04:00 user's guide: consolidate defaulting documentation This commit adds a new section on defaulting, which consolidates various parts of documentation surrounding defaulting into one central place. It explains type class defaulting in detail, extensions to it with OverloadedStrings, NamedDefaults and ExtendedDefaultRules, as well as other defaulting mechanisms (e.g. kind-based defaulting such as RuntimeRep defaulting, and defaulting of equalities). - - - - - 0c9fd8d4 by sheaf at 2025-03-11T06:43:06-04:00 user's guide: flesh out XOverloadedStrings docs This commit extends the documentation of the OverloadedStrings extension with some usage information, in particular suggestions to: - use default declarations, such as `default (Text)` or `default IsString(Text)` (with the NamedDefaults extension), - enable the ExtendedDefaultRules extension to relax the requirement that a defaultable type variable must only appear in unary standard classes Fixes #23388 - - - - - 2df171d4 by sheaf at 2025-03-11T06:43:06-04:00 user's guide: NamedDefaults vs ExtendedDefaultRules This commit clarifies the defaulting rules with NamedDefaults, in particular in situations where a type variable appears in other constraints than standard/unary constraints. - - - - - 77df05d0 by Matthew Pickering at 2025-03-11T06:43:07-04:00 Take into account all flags when computing iface_hash The "interface hash" should contain a hash of everything in the interface file. We are not doing that yet but at least a start is to include a hash of everything in `mi_self_recomp`, rather than just `mi_src_hash` and `mi_usages`. In particular, this fixes #25837, a bug where we should recompile a `dyn_hi` file but fail to do so. - - - - - 48b8f110 by Matthew Pickering at 2025-03-11T06:43:07-04:00 Pass -fPIC to dynamicToo001 test to avoid platform dependence issues On darwin platforms, `-fPIC` is always enabled but on linux it is only enabled in the dynamic flavour. This can cause a difference in interface files (see #25836). The purpose of this test isn't to test module A recompilation, so we avoid this platform dependency by always passing `-fPIC`. - - - - - 03c72f01 by Matthew Pickering at 2025-03-11T06:43:07-04:00 Remove mi_used_th field from interface files In the distant past, recompilation checking was disabled for interfaces which used TemplateHaskell, but for several years now recompilation checking has been more fine-grained. This has rendered this field unused and lingering in an interface file. - - - - - 6bb0e261 by Matthew Pickering at 2025-03-11T06:43:07-04:00 Remove mi_hpc field from interface files The `mi_hpc` field is not used for anything as far as I can discern so there is no reason to record in the private interface of a module that there are modules in the transitive closure which use `hpc`. You can freely mix modules which use `-fhpc` and ones which don't. Whether to recompile a module due to `-fhpc` being passed to the module itself is determined in `fingerprintDynFlags`. - - - - - b6d5b091 by Simon Peyton Jones at 2025-03-11T22:39:23-04:00 We can't UNPACK multi-constructor GADTs This MR fixes #25672 See Note [Unpacking GADTs and existentials] in GHC.Types.Id.Make - - - - - 8eae151d by Teo Camarasu at 2025-03-11T22:40:00-04:00 template-haskell: Add explicit exports lists to all remaining modules - - - - - db621b58 by Teo Camarasu at 2025-03-11T22:40:00-04:00 template-haskell: fix haddocks It seems that we need a direct dependency on ghc-internal, otherwise Haddock cannot find our haddocks The bug seems to be caused by Hadrian because if I rebuild with cabal-install (without this extra dependency) then I get accurate Haddocks. Resolves #25705 - - - - - 64ea68d9 by Ben Gamari at 2025-03-12T07:11:51-04:00 mk-ghcup-metadata: Clean up and add type annotations Getting this file right has historically been quite painful as it is a dynamically-typed script running only late in the release pipeline. - - - - - b3f80b07 by Ben Gamari at 2025-03-12T07:12:27-04:00 rts: Drop imports of pthreads functions in cmm sources These are no longer used. I noticed these while looking for uses of __PIC__ in Cmm sources. - - - - - 915a6781 by Matthew Pickering at 2025-03-13T01:46:41-04:00 interfaces: Ensure that forceModIface deeply forces a ModIface A ModIface is the result of compilation that we keep for a long time in memory. Therefore, it's very important to manage what we are going to retain and remove any external references to things which we might have captured compilation. If storing your ModIface in memory uses too much space, then store less things or make it use a more efficient representation. In the past there have been many space leak bugs by not sufficiently forcing a ModIface (#15111) This patch adds all the missing NFData instances for all the places I could find where we weren't deeply forcing the structure. - - - - - 24d373a6 by Matthew Craven at 2025-03-13T01:47:18-04:00 Add interface-stability test for ghc-prim - - - - - 0cb1db92 by sheaf at 2025-03-14T13:11:44-04:00 Don't report used duplicate record fields as unused This commit fixes the bug reported in #24035 in which the import of a duplicate record field could be erroneously reported as unused. The issue is that an import of the form "import M (fld)" can import several different 'Name's, and we should only report an error if ALL of those 'Name's are unused, not if ANY are. Note [Reporting unused imported duplicate record fields] in GHC.Rename.Names explains the solution to this problem. Fixes #24035 - - - - - f1830d74 by Matthew Pickering at 2025-03-14T13:12:21-04:00 binary: Directly copy ShortByteString to buffer rather than go via ByteString This avoids allocating an intermediate bytestring. I just noticed on a profile that `putFS` was allocating, and it seemed strange to me why since it should just copy the contents of the FastString into the already allocated buffer. It turned out we were going indirectly via a ByteString. Fixes #25861 - - - - - b15fca2b by Matthew Pickering at 2025-03-15T05:36:40-04:00 iface: Store logical parts of ModIface together The ModIface structure is divided into several logical parts: 1. mi_mod_info: Basic module metadata (name, version, etc.) 2. mi_public: The public interface of the module (the ABI), which includes: - Exports, declarations, fixities, warnings, annotations - Class and type family instances - Rewrite rules and COMPLETE pragmas - Safe Haskell and package trust information - ABI hashes for recompilation checking 4. mi_self_recomp: Information needed for self-recompilation checking (see Note [Self recompilation information in interface files]) 5. mi_simplified_core: Optional simplified Core for bytecode generation (only present when -fwrite-if-simplified-core is enabled) 6. mi_docs: Optional documentation (only present when -haddock is enabled) 7. mi_top_env: Information about the top-level environment of the original source 8. mi_ext_fields: Additional fields for extensibility This structure helps organize the interface data according to its purpose and usage patterns. Different parts of the compiler use different fields. By separating them logically in the interface we can arrange to only deserialize the fields that are needed. This patch also enforces the invariant that the fields of ModIface are lazy. If you are keeping a ModIface on disk, then force it using `forceModIface`. Otherwise, when the `ModIface` is read from disk, only the parts which are needed from the interface will be deserialised. In a follow-up patch I will tackle follow-up issues: * Recompilation checking doesn't take into account exported named defaults (#25855) * Recompilation checking does not take into account COMPLETE pragmas (#25854) * mi_deps_ field in an interface is confused about whether the information is for self-recompilation checking or part of the ABI (#25844) Fixes #25845 ------------------------- Metric Decrease: MultiLayerModulesDefsGhciWithCore ------------------------- - - - - - c758cb71 by Ben Gamari at 2025-03-15T05:37:17-04:00 configure: Fix incorrect SettingsLlvmAsFlags value Previously this was set to `LlvmAsCmd` rather than `LlvmAsFlags`, resulting in #25856. - - - - - cfaaca14 by sheaf at 2025-03-18T20:05:03-04:00 Fix buglet in isEmptyWorkList There was a missing case in GHC.Tc.Solver.InertSet.isEmptyWorkList; it mistakenly ignored the 'wl_rw_eqs' field. This commit simply fixes that. No test case. - - - - - 9f9fe0b3 by sheaf at 2025-03-18T20:05:03-04:00 Add mapMaybeTM method to TrieMap class This commit adds a new method to the TrieMap class, mapMaybeTM, and adds implementations to all the instances. mapMaybeTM is useful when filtering containers that contain other containers. - - - - - 393531ff by Simon Peyton Jones at 2025-03-18T20:05:03-04:00 Specialising expressions -- at last This MR addresses #24359, which implements the GHC proposal 493 on SPECIALISE pragmas. * The old code path (using SpecSig and SpecPrag) still exists. * The new code path (using SpecSigE and SpecPragE) runs alongside it. * All SPECIALISE pragmas are routed through the new code path, except if you give multiple type sigs, when the old code path is still used. * Main documentation: Note [Handling new-form SPECIALISE pragmas] in GHC.Tc.Gen.Sig` Thanks to @sheaf for helping with this MR. The Big Thing is to introduce {-# SPECIALISE forall x. f @Int x True #-} where you can give type arguments and value argument to specialise; and you can quantify them with forall, just as in Rules. I thought it was going to be pretty simple, but it was a Long, Long Saga. Highlights * Overview Note [Handling new-form SPECIALISE pragmas] in GHC.Tc.Gen.Sig - New data constructor `SpecSigE` in data type `L.H.S.Binds.Sig` - New data construtor `SpecPragE` in data type `GHC.Hs.Binds.TcSpecPrag` - Renamer: uses `checkSpecESigShape` to decide which function to assocate the SPECIALISE pragma with - Some of the action is in `GHC.Tc.Gen.Sig.tcSpecPrag` - The rest is in `GHC.HsToCore.Binds.dsSpec` * We use a new TcS mode, TcSFullySolve, when simplifying the Wanteds that arise from the specialise expression. The mechanism is explained in Note [TcSFullySolve] in GHC.Tc.Solver.Monad. The reason why we need to do this is explained in Note [Fully solving constraints for specialisation] in GHC.Tc.Gen.Sig. * All of GHC.Tc.Gen.Rule is moved into GHC.Tc.Gen.Sig, because the code is very closely related. * The forall'd binders for SPECIALISE are the same as those for a RULE, so I refactored, introducing data type `L.H.S.Binds.RuleBndrs`, with functions to rename, zonk, typecheck it. I refactored this data type a bit; nicer now. * On the LHS of RULES, or SPECIALISE, we want to disable the tricky mechanims described in Note [Desugaring non-canonical evidence] in GHC.HsToCore.Expr. Previously it wasn't fully disabled (just set to the empty set), and that didn't quite work in the new regime. * There are knock-on changes to Template Haskell. * For the LHS of a RULE and a SPECIALISE expression, I wanted to simplify it /without/ inlining the let-bindings for evidence variables. I added a flag `so_inline` to the SimpleOpt optimiser to support this. The entry point is `GHC.Core.SimpleOpt.simpleOptExprNoInline` * Since forever we have had a hack for type variables on the LHS of RULES. I took the opportunity to tidy this up. The main action is in the zonker. See GHC.Tc.Zonk.Type Note [Free tyvars on rule LHS], and especially data construtor `SkolemiseFlexi` in data type `GHC.Tc.Zonk.Env.ZonkFlexi` * Move `scopedSort` from GHC.Core.TyCo.FVs to GHC.Core.Predicate Reason: it now works for Ids as well, and I wanted to use isEvVar, which is defined in GHC.Core.Predicate Avoiding module loops meant that instead of exporting GHC.Core.TyCo.Tidy from GHC.Core.Type, modules now import the former directly. I also took the opportunity to remove unused exports from GHC.Core.Type.hs-boot * Flag stuff: - Add flag `-Wdeprecated-pragmas` and use it to control the warning when using old-style SPECIALISE pragmas with multiple type ascriptions, - Add flag `-Wuseless-specialisations` and use it to control the warning emitted when GHC determines that a SPECIALISE pragma would have no effect. Don't want if the SPECIALISE is SPECIALISE INLINE (#4444) In response to #25389, we continue to generate these seemingly code for these seemingly useless SPECIALISE pragmas - Adds deprecations to Template Haskell `pragSpecD` and `pracSpecInlD`, * Split up old-style SPECIALISE pragmas in GHC.Internal.Float, GHC.Internal.Numeric, GHC.Internal.Real * Remove useless SPECIALISE pragmas in Data.Array (updating the array submodule) Smaller things: - Update the Users Guide - Add mention of the changes to the 9.14 release notes as well as the Template Haskell changelog, - - - - - 1884dd1a by Simon Peyton Jones at 2025-03-18T20:05:03-04:00 Add -Wrule-lhs-equalities warning This commit adds a new warning, controlled by the warning flag, -Wrule-lhs-equalities, which is emitted when the LHS of a RULE gives rise to equality constraints that previous GHC versions would have quantified over. GHC instead discards such RULES, as GHC was never able to generate a rule template that would ever fire; it's better to be explicit about the fact that the RULE doesn't work. - - - - - b00b3ef0 by Ben Gamari at 2025-03-18T20:05:41-04:00 compiler: Add export list to GHC.SysTools.Process This also revealed that `readProcessEnvWithExitCode` and its local helpers were dead code. - - - - - 25850b22 by ARATA Mizuki at 2025-03-18T20:06:25-04:00 Fix code generation for SSE vector operations The new implementation generates correct code even if the registers overlap. Closes #25859 - - - - - e576468c by Andreas Klebinger at 2025-03-18T20:07:02-04:00 Bump nofib submodule. Fixes #25867. (Ben-raytrace being broken by library changes) - - - - - 443fc8b1 by Sjoerd Visscher at 2025-03-19T12:01:04-04:00 Multiplicity annotation on records Needing to store multiplicity annotations on records triggered a refactoring of AST of data declarations: Moved HsBangTy and HsRecTy from HsType to HsTypeGhcPsExt, the extension of HsType during parsing, since they are only needed during parsing. New HsConDeclField that stores all source data shared by all constructor declaration fields: unpackedness, strictness, multiplicity, documentation and the type of the field. Merged HsMultAnn and HsArrowOf, so all multiplicity annotations share the same data type. HsBang was no longer needed as a separate type, and was inlined into HsSrcBang. - - - - - 313cf271 by Ben Gamari at 2025-03-19T12:01:43-04:00 gitlab-ci: Drop CentOS 7 binary distributions CentOS 7 is EoL and moreover we cannot even build images for it. See #25061. - - - - - 5b94f99f by Ben Gamari at 2025-03-19T12:02:21-04:00 rts: Ensure that WinIO flag is set when --io-manager=auto As noted in #25838, previously `selectIOManager` failed to set `rts_IOManagerIsWin32Native` in its `IO_MNGR_FLAG_AUTO`. This meant that the MIO path was taken when WinIO was supposedly selected, resulting in chaos. Fixes #25838. - - - - - 1a3f1131 by Peng Fan at 2025-03-19T12:03:10-04:00 Pass the mcmodel=medium parameter to CC via GHC Ensure that GHC-driver builds default to mcmodel=medium, so that GHC passes this default parameter to CC without having to add it to the compiled project. Commit e70d41406b5d5638b42c4d8222cd03e76bbfeb86 does not ensure that all GHC-built object files have a default model of medium, and will raise an R_LARCH_B26 overflow error. - - - - - 27cf7361 by Matthew Craven at 2025-03-19T12:03:48-04:00 Add interface-stability test for ghc-bignum As with ghc-prim, it makes sense to have some protection against accidental interface changes to this package caused by changes in ghc-internal. - - - - - 25d46547 by Matthew Craven at 2025-03-19T12:03:48-04:00 Add README reference for the interface-stability tests - - - - - 5d65393e by Simon Peyton Jones at 2025-03-20T05:41:24-04:00 Remove the Core flattener This big MR entirely removes the "flattener" that took a type and replaced each type-family application with a fresh type variable. The flattener had its origin in the paper Injective type families for Haskell But (a) #25657 showed that flattening doesn't really work. (b) since we wrote the paper we have introduced the so-called "fine-grained" unifier GHC.Core.Unify, which can return * SurelyApart * Unifiable subst * MaybeApart subst where the MaybeApart says that the two types are not unifiable by a substitution, but could (perhaps) be unified "later" after some type family reductions. This turns out to subsume flattening. This MR does a major refactor of GHC.Core.Unify to make it capable of subsuming flattening. The main payload is described in Note [Apartness and type families] and its many wrinkles. The key (non-refactoring) implementation change is to add `um_fam_env` to the `UMState` in the unification monad. Careful review with Richard revealed various bugs in the treament of `kco`, the kind coercion carried around by the unifier, so that is substantially fixed too: see Note [Kind coercions in Unify]. Compile-time performance is improved by 0.1% with a few improvements over 1% and one worsening by 1.3% namely T9872a. (I have not investigated the latter.) Metric Decrease: T9872b T9872c TcPlugin_RewritePerf Metric Increase: T9872a - - - - - 9003ef0a by sheaf at 2025-03-20T05:42:08-04:00 Remove SDoc from UnknownSubordinate/MissingBinding This commit replaces unstructured SDoc arguments in error message constructors with uses of the following two datatypes: - SigLike: for different kinds of signatures (e.g. standalone kind signatures, fixity signatures, COMPLETE pragmas, etc) - Subordinate: for class methods, associated types, and record fields The following error message constructors now no longer have any SDocs in them: - TcRnIllegalBuiltinSyntax: SDoc -> SigLike - MissingBinding: SDoc -> SigLike - UnknownSubordinate: SDoc -> (Name, Subordinate) - SuggestMoveToDeclarationSite: SDoc -> SigLike - - - - - 4329f3b6 by sheaf at 2025-03-20T05:42:08-04:00 Remove SDocs from HsDocContext This commit removes the remaining SDocs from the HsDocContext data type. It adds the following constructors: ClassInstanceCtx -- Class instances ClassMethodSigCtx -- Class method signatures SpecialiseSigCtx -- SPECIALISE pragmas PatSynSigCtx -- Pattern synonym signatures We now report a bit more information for errors while renaming class instances, which slightly improves renamer-emitted error messages. - - - - - 75c29aa1 by sheaf at 2025-03-20T05:42:08-04:00 Reject instance with non-class head in renamer This commit modifies rnClsInstDecl so that, when renaming, we reject a class instance declaration in which the head is not a class. Before this change, it would get rejected in the type-checker, but that meant that the renamer could emit unhelpful error messages, e.g.: data Foo m a instance Foo m where fmap _ x = case x of would rather unhelpfully say: ‘fmap’ is not a (visible) method of class ‘Foo’ when of course 'Foo' is not even a class. We now reject the above program with the following error message: Illegal instance for data type ‘Foo’. Instance heads must be of the form C ty_1 ... ty_n where ‘C’ is a class. Fixes #22688 - - - - - a8f543a9 by Cheng Shao at 2025-03-20T18:47:19-04:00 testsuite: mark T7919 as fragile on i386 as well T7919 may also fail i386 CI jobs with test timeout. - - - - - 256ac29c by sheaf at 2025-03-20T18:48:07-04:00 Don't cache solved [W] HasCallStack constraints This commit ensures we do not add solved Wanted constraints that mention HasCallStack or HasExceptionContext constraints to the set of solved Wanted dictionary constraints: caching them is invalid, because re-using such cached dictionaries means using an old call-stack instead of constructing a new one, as was reported in #25529. Fixes #25529. - - - - - 47646ce2 by Andrew Lelechenko at 2025-03-20T18:48:43-04:00 Improve haddock-visible documentation for GHC.Driver.Flags - - - - - 2e4e15ed by Sylvain Henry at 2025-03-21T17:49:36-04:00 Document -fnum-constant-folding (#25862) - - - - - 044a6e08 by sheaf at 2025-03-21T17:50:24-04:00 LLVM: fix typo in padLiveArgs This commit fixes a serious bug in the padLiveArgs function, which was incorrectly computing too many padding registers. This caused segfaults, e.g. in the UnboxedTuples test. Fixes #25770 Fixes #25773 - - - - - 1745c749 by Teo Camarasu at 2025-03-22T18:42:37-04:00 template-haskell: remove Language.Haskell.TH.Lib.Internal This module is purely used for the implementation of TH quote desugarring. Historically this needed to be exposed from `template-haskell`, since that's the package that the desugarred expressions referenced but since https://gitlab.haskell.org/ghc/ghc/-/merge_requests/12479, this is no longer the case. Now these identifiers are in `ghc-internal`. Note that this module has carried the following warning for a long time:
This is not a part of the public API, and as such, there are no API guarantees for this module from version to version.
Resolves #24766
- - - - -
3bc507db by Alan Zimmerman at 2025-03-22T18:43:13-04:00
EPA: Fix exact printing of SPECIALISE pragma
This commit fixes two minor issues with exactprinting of the
SPECIALISE pragma after !12319 landed
- The span for the RHS did not include the optional signature type
- The `::` was printed twice when the legacy path was used
Closes #25885
- - - - -
bdf93da8 by sheaf at 2025-03-24T11:25:12+01:00
Renamer: improve handling of export children
This commit refactors the 'childGREPriority' function which is used when
renaming subordinate items in export lists and class declarations.
Instead of having a complicated LookupChild parameter, we now simply pass the
GREInfo of the parent, which allows us to decide what is a valid child:
- classes can have children that are in the type constructor namespace,
- promoted data constructors should be treated the same as normal data
constructors.
Fixes #24027
- - - - -
1dbc7846 by Matthew Pickering at 2025-03-24T15:35:17-04:00
Revert "mk-ghcup-metadata: Clean up and add type annotations"
This reverts commit 64ea68d9a206ec4db5020b0a3fc563199ab18be3.
See #25889
- - - - -
6941c825 by doyougnu at 2025-03-25T10:05:58-04:00
base: construct compat RTSFlags
-- see CLC #289
- - - - -
730e6f77 by doyougnu at 2025-03-25T10:06:02-04:00
base: construct compat GHC.Stats
-- see CLC #289
- - - - -
cd9e6605 by Ben Gamari at 2025-03-25T17:59:52-04:00
rel_eng/upload: Clarify usage directions
Previously it was not made clear that the directory name is significant.
- - - - -
7d18c19b by Cheng Shao at 2025-03-25T22:00:56+00:00
ghci: make the Pipe type opaque
This commit makes the Pipe type in ghci opaque, and introduce the
mkPipeFromHandles constructor for creating a Pipe from a pair of
Handles. Pipe is now just a pair of receiver/sender continuations
under the hood. This allows a Pipe to be potentially backed by other
IPC mechanisms (e.g. WebSockets) which is essential for wasm ghci
browser mode.
- - - - -
a2103fd2 by Cheng Shao at 2025-03-25T22:00:56+00:00
ghci: use improved Pipe logic for wasm iserv
This commit makes wasm iserv take advantage of the Pipe refactoring by
passing binary receiver/sender js callbacks from the dyld script. This
paves the way for piping the binary messages through WebSockets in
order to run wasm iserv in the browser, but more importantly, it
allows us to get rid of a horrible hack in the dyld script: we no
longer have to fake magical wasi file descriptors that are backed by
nodejs blocking I/O! The legacy hack was due to these facts:
- iserv only supported exchanging binary messages via handles backed
by file descriptors
- In wasi you can't access host file descriptors passed by host ghc
- The nodejs wasi implementation only allows mapping host directories
into the wasi vfs, not host file descriptors
- Named pipes with file paths (mkfifo) doesn't work well with nodejs
wasi implementation, causes spurious testsuite failures on macos
But starting from this point, we can fully take advantage of
non-blocking I/O on the js side.
- - - - -
fa2fbd2b by Cheng Shao at 2025-03-25T22:00:56+00:00
ghci: fix ^C handling for wasm iserv
This commit fixes ^C handling for wasm iserv. Previously we didn't
handle it at all, so ^C would kill the node process and host ghc would
then crash as well. But native ghc with external interpreter can
handle ^C just fine and wasm should be no different. Hence the fix:
wasm iserv exports its signal handler as a js callback to be handled
by the dyld script. Also see added note for details.
- - - - -
efcebed6 by Cheng Shao at 2025-03-25T22:00:56+00:00
wasm: fix post-link.mjs for browser
The wasm ghci browser mode needs to run dyld.mjs in the browser which
imports post-link.mjs. This script makes post-link.mjs runnable in the
browser by deferring node-specific module imports to their actual use
sites.
- - - - -
27bb73c6 by Cheng Shao at 2025-03-25T22:00:56+00:00
wasm: use console.assert in dyld script
This commit uses console.assert() instead of node-specific strict
assert in the dyld script, in order to make it runnable in the
browser. console.assert() only warns and doesn't crash upon assertion
failure, but this is fine; we can always trivially define a strict
assert function shall it be necessary when debugging, and there hasn't
been such an assertion failure seen in the wild for long enough.
- - - - -
929df0ba by Cheng Shao at 2025-03-25T22:00:56+00:00
wasm: asyncify the dylink.0 custom section parser
This commit refactors the simple binary parser in the dyld script in
charge of parsing the dylink.0 custom section. Previously the parser
was synchronous and operated on the entire input buffer; this was
simple and easy and worked well enough when the input wasm modules are
instantly read from local filesystem.
However, when running dyld in the browser, the wasm modules are
transferred via fetch() requests. The host ghc and the browser might
not be on the same machine, so slow network uplink does need to be
considered. We only need to parse dylink.0 custom section to extract
dependency info, and dylink.0 is the very first custom section in the
wasm shared library binary payload, so the parsing process should not
require fetch() to complete and should return the parsing result asap.
Hence the refactorings in this commit: asyncify the parser, make it
only consume as many bytes as needed by invoking an async consumer
callback. The input is a readable stream from the fetch() response;
once the response is available, the async wasm compilation can start
in the background, and dylink.0 parsing shall end asap which results
in more wasm shared libraries to be loaded earlier. Profit.
- - - - -
9a697181 by Cheng Shao at 2025-03-25T22:00:56+00:00
wasm: fix dyld setImmediate usage in browser
The wasm dyld script used to only run in node and directly uses
setImmediate in globalThis. In case of browsers, it needs to import
setImmediate from the prelude, hence this commit.
- - - - -
d9b71e82 by Cheng Shao at 2025-03-25T22:00:57+00:00
wasm: fix dyld downsweep filepath handling in browser
The wasm dyld downsweep logic used to rely on nodejs path module to
handle filepaths. That's not available in browsers, so this commit
implements poor man's filepath handling in js, which is not elegant
for sure but works for both nodejs and the browser.
- - - - -
7003a399 by Cheng Shao at 2025-03-25T22:00:57+00:00
wasm: isolate nodejs-specific logic with the isNode flag in dyld
As we move towards supporting running the dyld script in the browser,
this commit implements the isNode module-level binding which is true
if dyld is running in nodejs. The nodejs-specific bits are gated under
isNode.
For the browser case, this commit introduces @bjorn3/browser_wasi_shim
as the wasi implementation; we already use it in quite a few projects
and it simply works.
- - - - -
22ba2a78 by Cheng Shao at 2025-03-25T22:00:57+00:00
wasm: isolate dyld side effects that might require IPC
This commit spins out a DyLDHost class from DyLD that handles side
effects that must be run in the same host environment that runs
wasm32-wasi-ghc. When the dyld script runs in the browser, it'll need
to do IPC to find libraries, fetch wasm library, etc, and the other
side of dyld that runs on nodejs would simply expose the DyLDHost
methods as endpoints for WebSockets/HTTP.
- - - - -
e93fc33d by Cheng Shao at 2025-03-25T22:00:57+00:00
wasm: implement wasm ghci browser mode
This commit implements the rest of dyld logic that delivers the ghci
browser mode:
- The dyld script can now fully run in the browser. It communicates
back with dyld-on-nodejs via WebSockets and also plain HTTP 1.1
requests.
- The host dyld starts a server and acts as a broker between the GHC
process and the browser side. GHC doesn't need to know anything
about the browser mode; no driver flags need to be added and no
recompilation needs to happen, the GHC driver continues to use the
original iserv binary messages protocol.
- The dyld broker doesn't need to parse any message between the
browser dyld and GHC; it merely sets up WebSockets connections to
redirect these messages as well as ^C signals.
- Plain HTTP 1.1 is used for IPC requests (e.g. downloading a wasm
module).
- The dyld broker serves a main.js script that bootstraps iserv in the
browser, and a main.html empty page playground for testing. CORS is
enabled so it could be possible to inject iserv into other websites
and use ghci to play with them.
- All the RPC logic is opaque to the DyLD class, the majority of the
wasm dynamic linker code is already portable and runs fine in
firefox/chrome/webkit.
Closes #25399.
- - - - -
fc576798 by Cheng Shao at 2025-03-25T22:00:57+00:00
wasm: add puppeteer/playwright support for ghci browser mode
This commit adds support for using puppeteer/playwright for
automatically launching a headless browser that backs the ghci browser
mode. This is useful for testing the ghci browser mode as a part of
GHC testsuite, and it's also convenient for local development since
the step to start iserv can be automated away.
- - - - -
ad7e271d by Cheng Shao at 2025-03-25T22:00:57+00:00
wasm: support wasi console redirect for the ghci browser mode
This commit adds optional support for redirecting wasi console
stdout/stderr back to the host when running wasm ghci browser mode. By
default, the wasi console outputs are only available under F12
devtools console, but in case of testing against a mobile browser, the
devtools console may not be readily available, and it would be more
convenient to at least get wasi console output on the host side.
The redirection logic is simple, just adding another two WebSockets
connections that pump the line-buffered textual messages back to
host.
- - - - -
731217ce by Cheng Shao at 2025-03-25T22:00:57+00:00
wasm: add brotli compression for ghci browser mode
This commit adds brotli compression for wasm shared libraries for ghci
browser mode. With BROTLI_MIN_QUALITY, the overhead is negligible, and
it helps reducing amount of transferred data when the browser connects
to the server over a slow connection.
- - - - -
ac70e643 by Cheng Shao at 2025-03-26T13:37:19+00:00
testsuite: add browser001 test for wasm ghci browser mode
This commit adds support for testing the wasm ghci browser mode in the
testsuite, as well as a simple first test case browser001 that makes
use of TH, JSFFI and browser-specific DOM API. See added note and
comments for details.
- - - - -
6ef5c0d2 by Cheng Shao at 2025-03-26T13:37:24+00:00
docs: add wasm ghci subsection in user manual
This commit updates the user manual to add wasm ghci subsection.
- - - - -
37381bcf by Cheng Shao at 2025-03-26T13:37:24+00:00
docs: update Note [The Wasm Dynamic Linker]
This commit updates Note [The Wasm Dynamic Linker] to reflect recent
developments, in particular the wasm ghci browser mode.
- - - - -
4b5a0f61 by Cheng Shao at 2025-03-26T13:37:24+00:00
ci: bump DOCKER_REV and test wasm ghci browser mode
This commit bumps the ci-images revision for updated wasm toolchain,
and adds the launch options required to test wasm ghci browser mode.
- - - - -
c6a3bc8f by Cheng Shao at 2025-03-26T13:37:24+00:00
driver: implement wasm ghci browser mode flags
This commit implements GHC driver flags that enable the wasm ghci
browser mode.
- - - - -
f75e823e by Cheng Shao at 2025-03-26T18:01:54-04:00
rts: add hs_try_putmvar_with_value to RTS API
This commit adds hs_try_putmvar_with_value to rts. It allows more
flexibility than hs_try_putmvar by taking an additional value argument
as a closure to be put into the MVar. This function is used & tested
by the wasm backend runtime, though it makes sense to expose it as a
public facing RTS API function as well.
- - - - -
9cd9f347 by Cheng Shao at 2025-03-26T18:01:54-04:00
wasm: use MVar as JSFFI import blocking mechanism
Previously, when blocking on a JSFFI import, we push a custom
stg_jsffi_block stack frame and arrange the `promise.then` callback to
write to that stack frame. It turns out we can simply use the good old
MVar to implement the blocking logic, with a few benefits:
- Less maintenance burden. We can drop the stg_jsffi_block related Cmm
code without loss of functionality.
- It interacts better with existing async exception mechanism. throwTo
would properly block the caller if the target thread is masking
async exceptions.
- - - - -
da34f0aa by Cheng Shao at 2025-03-26T18:01:54-04:00
wasm: properly pin the raiseJSException closure
We used to use keepAlive# to pin the raiseJSException closure when
blocking on a JSFFI import thunk, since it can potentially be used by
RTS. But raiseJSException may be used in other places as well (e.g.
the promise.throwTo logic), and it's better to simply unconditionally
pin it in the JSFFI initialization logic.
- - - - -
dc904bfd by Cheng Shao at 2025-03-26T18:01:54-04:00
wasm: implement promise.throwTo() for async JSFFI exports
This commit implements promise.throwTo() for wasm backend JSFFI
exports. This allows the JavaScript side to interrupt Haskell
computation by raising an async exception. See subsequent docs/test
commits for more details.
- - - - -
7f80455e by Cheng Shao at 2025-03-26T18:01:55-04:00
testsuite: add test for wasm promise.throwTo() logic
This commit adds a test case to test the wasm backend
promise.throwTo() logic.
- - - - -
afdd3fe7 by Cheng Shao at 2025-03-26T18:01:55-04:00
docs: document the wasm backend promise.throwTo() feature
- - - - -
65dc65dc by sheaf at 2025-03-26T18:02:43-04:00
Refactor CtEvidence into Given/Wanted
This commit is a simple refactoring which splits up the CtEvidence,
giving each constructor its own datatype:
data CtEvidence
= CtGiven GivenCtEvidence
| CtWanted WantedCtEvidence
data GivenCtEvidence =
GivenCt
{ ctev_pred :: TcPredType
, ctev_evar :: EvVar
, ctev_loc :: CtLoc }
data WantedCtEvidence =
WantedCt
{ ctev_pred :: TcPredType
, ctev_dest :: TcEvDest
, ctev_loc :: CtLoc
, ctev_rewriters :: RewriterSet }
This enables a few minor simplifications in the code, notably removing
a panic from GHC.Tc.Solver.Solve.solveWantedForAll_implic.
Fixes #25848
- - - - -
ada04031 by sheaf at 2025-03-26T18:03:27-04:00
Export lists: same prio for NoParent & RightParent
This commit ensures that, when we are renaming children in an export
list item such as
module M ( P(A,B,C) )
we consider children with NoParent to have the same priority as children
which have the correct parent (P in this case). This is because we
should **not** prioritise a data constructor (with the right parent)
over a pattern synonym we are bundling (which, before bundling, has
no parent).
Fixes #25892
- - - - -
721628a0 by Adriaan Leijnse at 2025-03-27T09:10:10-04:00
TTG: Replace HsUnboundVar with HsHole
Context:
The HsUnboundVar constructor in Language.Haskell.Syntax.Expr contained a
RdrName, which stood in the way of the work towards a dedicated
haskell-syntax library. The constructor was overloaded for unbound
variables, anonymous and named holes, and parse errors.
This commit:
Replaces HsUnboundVar with HsHole. In the surface syntax HsHole only
represents an anonymous expression hole ("_"). It is extended with the
XHole type family.
In the concrete GHC implementation of the language on the other hand,
HsHole is used for "any thing which is not necessarily a valid or fully
defined program fragment, but for which a type can be derived". This use
is similar to how HsUnboundVar was used, but the parse error case is now
made explicit with a ParseError case for XHole. This is in anticipation
of future work on a fault tolerant compilation pipeline.
- - - - -
dbd852f5 by Ben Gamari at 2025-03-27T09:10:48-04:00
rel_eng: Finish removal of CentOS jobs
Remove centos7 from release fetch and ghcup metadata generation scripts.
Closes #25893.
- - - - -
0e0231e7 by Matthew Pickering at 2025-03-28T18:36:33-04:00
hadrian: Make hash_unit_ids into a flavour transformer (and enable for release flavour)
The primary reason for this change is to make the `release` flavour
enable `--hash-unit-ids` by default without any further user intervention.
* Packagers don't have to be aware of this special flag they should be
using.
* release builds on CI are uniformly testing with hashes (see !13418)
Fixes #25379
- - - - -
9fc54c12 by Rodrigo Mesquita at 2025-03-28T18:37:10-04:00
driver: Move DynFlags consistency fixes off Main
These consistency fixes found in Main.hs are required for the proper
functioning of the compiler and should live together with all remaining
fixes in `makeDynFlagsConsistent`.
This is especially relevant to GHC applications that shouldn't have to
copy/fix themselves possibly inconsistent DynFlags.
Additionally, outputs information when verbosity is high about these
consistency fixes that were previously quiet, adds information to the
Note on consistency of DynFlags, and improves one of the fixes that
incorrectly used `dynNow`.
- - - - -
2fdd0be9 by sheaf at 2025-03-28T18:37:53-04:00
Remove GhcHint from TcRnNotInScope constructor
This is a tiny refactoring which:
- removes GhcHint/ImportError fields from some constructors of
TcRnMessage, using the TcRnMessageDetailed mechanism instead to
report this informaiton:
- removes the GhcHint and ImportErrors fields from TcRnNotInScope
- removes the GhcHint field from TcRnTermNameInType
- ensures that we only include these hints and import errors when the
-fhelpful-errors flag is turned on
Fixes #25874
- - - - -
9e5cd064 by ARATA Mizuki at 2025-03-31T14:53:34-04:00
Better support for SSE3 and SSE4.1
In particular:
* Pass appropriate attributes to LLVM
* Define preprocessor macros for them
- - - - -
c2c7dd51 by ARATA Mizuki at 2025-03-31T14:53:34-04:00
x86: Add support for SSSE3
This commit adds the `-mssse3` flag, which controls usage of SSSE3 instructions in x86 code generation.
- - - - -
d7c62580 by ARATA Mizuki at 2025-03-31T14:53:34-04:00
x86 NCG SIMD: Implement 128-bit integer vector arithmetics
This commit implements the following operations on integer vectors:
* negateIntNXM#
* plus{Int,Word}NXM#
* minus{Int,Word}NXM#
* times{Int,Word}NXM#
* quot{Int,Word}NXM#
* rem{Int,Word}NXM#
* min{Int,Word}NXM#
* max{Int,Word}NXM#
where (N,M) is one of (8,16), (16,8), (32,4), or (64,2).
Closes #25643
- - - - -
f5ea4e7e by ARATA Mizuki at 2025-03-31T14:53:34-04:00
x86 NCG SIMD: Implement 128-bit integer vector shuffle
This commit implements the following operations:
* shuffle{Int,Word}8X16#
* shuffle{Int,Word}16X8#
* shuffle{Int,Word}32X4#
* shuffle{Int,Word}64X2#
See #25643
- - - - -
5eeb6645 by Simon Peyton Jones at 2025-03-31T14:54:16-04:00
Re-jig the way that the Simplifier tries RULES
As #25170 showed, if a new RULE appears, it could change the simplifier's
behaviour a bit, even if it never fires; and that messes up deterministic
compilation. (This was particularly nasty if the rule wasn't even transitively
below the module being compiled.)
This MR rejigs the use of `tryRules` so that behaviour does not change
when a new, unrelated RULE is added. It's described in
* Note [When to apply rewrite rules]
* Note [tryRules: plan (BEFORE)]
* Note [tryRules: plan (AFTER)]
The main change is in the refactored version of
* simplOutId
* rebuildCall
The little state machine that was embedded in ArgInfo is gone.
As I wandered around the Simplifier I also found opportunities for
some loosely-related refactoring:
* In several places, the /substitution/ in the SimpleEnv is empty;
all we care about is the in-scope set and the flags. So
- I made a synonym `SimplEnvIS` that embodies that invariant,
- used it in a number of type signatures (notably `rebuild`)
- added some assertion checks (via `checkSimlEnvIS`)
* I moved the hanlding of `runRW` out of `rebuildCall` (where we
would have to test repeatedly) and into the new `simplOutId`, which
fires up `rebuildCall`. Now it is only tested once.
-------------------------
Metric Decrease:
T9020
T9961
-------------------------
- - - - -
f534474a by sheaf at 2025-03-31T14:54:59-04:00
Add comment: no qualified Names in the LocalRdrEnv
This commit adds a reference to section 5.5.1 of the Haskell 2010 report,
to explain that qualified names can't occur in the LocalRdrEnv.
Fixes #25875
- - - - -
d5ea80c6 by Patrick at 2025-03-31T14:55:42-04:00
Fix deadlock/loop in interface rehydration (#25858)
In #25858, GHC hangs when processing modules with class defaults due to a
circular dependency in the interface rehydration process. The deadlock/loop
occurred when eager class defaults rehydration accessed not-yet-complete module details.
To fix the immediate deadlock/loop.
`tcIfaceDefaults` is refactored, we use the class name directly from the iface
and use `forM` for lazy loading the class, which algins with the handling of other fields
of ModDetails. This laziness ensure rehydration waits for HomePackageTable (HPT) to
be updated and prevent premature evaluation of ModDetails inside `fixIO``.
As suggested by Matthew, class defaults importing is also refactored
to align with the compiler's established interface loading conventions.
- add class defaults field to ExternalPackageState (EPS).
- rehydrate and store class defaults in EPS at `loadInterface`.
- Instead of using `tcIfaceDefaults` in `tcRnImports`, we add and use
`tcGetClsDefaults` to read defaults directly from HPT or EPS
when importing modules.
Tests:
- T25858, T25858v1-2: Test class hydration in defaults
- T25858v3-4: Test type list hydration in defaults
New Note [Tricky rehydrating IfaceDefaults loop] is added.
Thanks to @sheaf (Sam), @mpickering (Matthew), and @simonpj (Simon) for their
valuable input and analysis.
Fixes #25858.
- - - - -
2d419d8d by Matthew Pickering at 2025-04-02T16:12:36-04:00
Use unsafePerformIO in definition of computeFingerprint
computeFingerprint is morally a pure function, which is implemented by
mutating a buffer. Using unsafePerformIO inside the definition allows it
to be used in pure contexts, fixing one place where an ad-hoc call to
unsafePerformIO is already needed.
- - - - -
ccdf979b by Matthew Pickering at 2025-04-02T16:12:37-04:00
driver: Fix recompilation checking for exported defaults
Since the exported defaults are not associated with any identifier from
the module, they are just added to the export hash rather than
the fine-grained recompilation logic.
Fixes #25855
- - - - -
c5bf9892 by Matthew Pickering at 2025-04-02T16:12:37-04:00
driver: Fix recompilation checking for COMPLETE pragmas
A {-# COMPLETE P, Q #-} pragma is associated with the pattern synonyms P
and Q during recompilation checking. Therefore, the existence of a
pattern synonym becomes part of the ABI hash for P and Q.
Then if a module uses these pattern synonyms and a complete pragma
changes, it will trigger recompilation in that module.
Fixes #25854
- - - - -
d0fd9370 by sheaf at 2025-04-02T16:14:05-04:00
Handle named default exports separately
This commit changes the way we check for duplicate exports of named
default declarations. They are now treated entirely separately from
other exports, because in an export list of the form
module M ( default Cls, Cls )
the default declaration does not export the class 'Cls', but only its
default declarations.
Also fixes a bug in Backpack where named default exports were getting
dropped entirely. No test for that.
Fixes #25857
- - - - -
62d04494 by Cheng Shao at 2025-04-03T05:56:17-04:00
ci: add x86_64-linux-ubuntu24_04 nightly/release jobs
- - - - -
327952e4 by Cheng Shao at 2025-04-03T05:56:17-04:00
rel-eng: add ubuntu24_04 bindists to ghcup metadata and fetch gitlab scripts
- - - - -
aa1e3b8b by sheaf at 2025-04-03T05:57:24-04:00
GHC settings: always unescape escaped spaces
In #25204, it was noted that GHC didn't properly deal with having
spaces in its executable path, as it would compute an invalid path
for the C compiler.
The original fix in 31bf85ee49fe2ca0b17eaee0774e395f017a9373 used a
trick: escape spaces before splitting up flags into a list. This fixed
the behaviour with extra flags (e.g. -I), but forgot to also unescape
for non-flags, e.g. for an executable path (such as the C compiler).
This commit rectifies this oversight by consistently unescaping the
spaces that were introduced in order to split up argument lists.
Fixes #25204
- - - - -
34a9b55d by lazyLambda at 2025-04-04T06:22:26-04:00
Driver: make MonadComprehensions imply ParallelListComp
This commit changes GHC.Driver.Flags.impliedXFlags to make the
MonadComprehensions extension enable the ParallelListComp extension.
Fixes #25645
- - - - -
d99eb7cd by sheaf at 2025-04-04T06:23:28-04:00
NamedDefaults: handle poly-kinded unary classes
With this commit, we accept named default declarations for poly-kinded
classes such as Typeable, e.g.
default Typeable (Char)
This used to fail because we assumed the kind of the class was monomorphic,
e.g.
Type -> Constraint
(Type -> Type) -> Constraint
Nat -> Constraint
Now, we can handle a simple polymorphic class such as
Typeable :: forall k. k -> Constraint
Note that we keep the restriction that the class must only have
one visible argument.
This is all explained in the new Note [Instance check for default declarations]
in GHC.Tc.Gen.Default.
Fixes #25882
- - - - -
4cbc90de by sheaf at 2025-04-04T11:39:05-04:00
LLVM: add type annotations to AtomicFetch_cmm.cmm
- - - - -
e2237305 by sheaf at 2025-04-04T11:39:05-04:00
Cmm lint: lint argument types of CallishMachOps
This commit adds a new check to Cmm lint to ensure that the argument
types to a CallishMachOp are correct. The lack of this check was
detected in the AtomicFetch test: the literals being passed as the
second arguments to operations such as 'fetch_add', 'fetch_and'... were
of the wrong width, which tripped up the LLVM backend.
- - - - -
9363e547 by Cheng Shao at 2025-04-04T11:39:50-04:00
ci: add ghc-wasm-meta integration testing jobs
This patch adds ghc-wasm-meta integration testing jobs to the CI
pipeline, which are only triggered via the `test-wasm` MR label or
manually when the `wasm` label is set.
These jobs will fetch the wasm bindists and test them against a
variety of downstream projects, similarly to head.hackage jobs for
native bindists, offering a convenient way to catch potential
downstream breakage while refactoring the wasm backend.
- - - - -
27029e60 by Adam Gundry at 2025-04-04T11:40:36-04:00
base: Minor fixes to GHC.Records haddocks
This corrects a stale reference to OverloadedRecordFields (which should
be OverloadedRecordDot), fixes the haddock link syntax and adds an
@since pragma.
- - - - -
f827c4c6 by Rodrigo Mesquita at 2025-04-07T11:22:10-04:00
Parametrize default logger action with Handles
Introduce `defaultLogActionWithHandles` to allow GHC applications to use
GHC's formatting but using custom handles.
`defaultLogAction` is then trivially reimplemented as
```
defaultLogActionWithHandles stdout stderr
```
- - - - -
5dade5fd by sheaf at 2025-04-07T11:23:02-04:00
Finer-grained recompilation checking for exports
This commit refines the recompilation checking logic, to avoid
recompiling modules with an explicit import list when the modules they
import start exporting new items.
More specifically, when:
1. module N imports module M,
2. M is changed, but in a way that:
a. preserves the exports that N imports
b. does not introduce anything that forces recompilation downstream,
such as orphan instances
then we no longer require recompilation of N.
Note that there is more to (2a) as initially meets the eye:
- if N includes a whole module or "import hiding" import of M,
then we require that the export list of M does not change,
- if N only includes explicit imports, we check that the imported
items don't change, e.g.
- if we have @import M(T(K, f), g)@, we must check that N
continues to export all these identifiers, with the same Avail
structure (i.e. we should error if N stops bundling K or f with
T)
- if we have @import M(T(..))@, we must check that the children
of T have not changed
See Note [When to recompile when export lists change?] in GHC.Iface.Recomp.
This is all tested in the new tests RecompExports{1,2,3,4,5}
Fixes #25881
- - - - -
f32d6c2b by Andreas Klebinger at 2025-04-07T22:01:25-04:00
NCG: AArch64 - Add -finter-module-far-jumps.
When enabled the arm backend will assume jumps to targets outside of the
current module are further than 128MB away.
This will allow for code to work if:
* The current module results in less than 128MB of code.
* The whole program is loaded within a 4GB memory region.
We have seen a few reports of broken linkers (#24648) where this flag might allow
a program to compile/run successfully at a very small performance cost.
-------------------------
Metric Increase:
T783
-------------------------
- - - - -
553c280b by Andreas Klebinger at 2025-04-07T22:02:11-04:00
Revert "rts: fix small argument passing on big-endian arch (fix #23387)"
Based on analysis documented in #25791 this doesn't fully fix the big
while introducing new bugs on little endian architectures.
A more complete fix will have to be implemented to fix #23387
This reverts commit 4f02d3c1a7b707e609bb3aea1dc6324fa19a5c39.
- - - - -
b0dc6599 by Andreas Klebinger at 2025-04-07T22:02:11-04:00
Interpreter: Fixes to handling of subword value reads/writes.
Load subword values as full words from the stack truncating/expanding as
neccesary when dealing with subwords. This way byte order is implicitly
correct.
This commit also fixes the order in which we are pushing literals onto
the stack on big endian archs.
Last but not least we enable a test for ghci which actually tests these
subword operations.
- - - - -
ed38c09b by Cheng Shao at 2025-04-07T22:02:53-04:00
testsuite: don't test WasmControlFlow stdout
This patch solves a potential test flakiness in `WasmControlFlow` by
removing `WasmControlFlow.stdout` which is not so portable/stable as
it seems. See added `Note [WasmControlFlow]` for more detailed
explanation.
- - - - -
f807c590 by Rodrigo Mesquita at 2025-04-08T17:41:51-04:00
debugger: Add docs to obtainTermFromId
- - - - -
5dba052d by Rodrigo Mesquita at 2025-04-08T17:41:51-04:00
Move logic to find and set Breakpoint to GHC
Breakpoints are uniquely identified by a module and an index unique
within that module. `ModBreaks` of a Module contains arrays mapping from
this unique breakpoint index to information about each breakpoint. For
instance, `modBreaks_locs` stores the `SrcSpan` for each breakpoint.
To find a breakpoint using the line number you need to go through all
breakpoints in the array for a given module and look at the line and
column stored in the `SrcSpan`s. Similarly for columns and finding
breakpoints by name.
This logic previously lived within the `GHCi` application sources,
however, it is common to any GHC applications wanting to set
breakpoints, like the upcoming `ghc-debugger`.
This commit moves this logic for finding and setting breakpoints to the
GHC library so it can be used by both `ghci` and `ghc-debugger`.
- - - - -
bc0b9f73 by Rodrigo Mesquita at 2025-04-08T17:41:51-04:00
Refactor and move logic for identifier breakpoints
Breakpoints can be set on functions using syntax of the form
`[Module.]function`. The parsing, resolution (e.g. inferring implicit
module), and validation of this syntax for referring to functions was
tightly coupled with its GHCi use.
This commit extracts the general purpose bits of resolving this syntax
into `GHC.Runtime.Debugger.Breakpoints` so it can be further used by
other GHC applications and to improve the code structure of GHCi.
Moreover, a few utilities that do splitting and joining of identifiers
as strings were moved to `GHC.Runtime.Eval.Utils`, which also can be
used in the future to clean up `GHC.Runtime.Eval` a bit.
- - - - -
4f728d21 by Rodrigo Mesquita at 2025-04-08T17:41:51-04:00
debugger: derive Ord for BreakpointIds
- - - - -
5528771c by Rodrigo Mesquita at 2025-04-08T17:41:51-04:00
debugger: Move context utils from GHCi to GHC
Moves `enclosingTickSpan`, `getCurrentBreakSpan`, and
`getCurrentBreakModule`, general utilities on the internal debugger
state, into the GHC library.
- - - - -
4871f543 by sheaf at 2025-04-08T17:42:43-04:00
Implicit quantification in type synonyms: add test
This adds a test for ticket #24090, which involves implicit
quantification in type synonyms.
The underlying issue was fixed in 0d4ee209dfe53e5074d786487f531dabc36d561c.
- - - - -
48917d3c by sheaf at 2025-04-08T17:42:44-04:00
Turn on implicit-rhs-quantification by default
This flag was added to GHC 9.8, and will be removed in a future GHC
release. In preparation, this commit adds it to the default warning
flags.
- - - - -
629be068 by Rodrigo Mesquita at 2025-04-08T17:43:26-04:00
debugger: Add breakpoints to every Stmt
While single-stepping through a Haskell program we stop at every
breakpoint. However, we don't introduce breakpoints at every single
expression (e.g. single variables) because they would be too many and
uninteresting.
That said, in a do-block, it is expected that stepping over would break
at every line, even if it isn't particularly interesting (e.g. a single
arg like getArgs). Moreover, let-statements in do-blocks, despite only
being evaluated once needed, lead to surprising jumps while stepping
through because some have outermost (outside the let) breakpoints
while others don't.
This commit makes every statement in a do-block have a breakpoint.
This leads to predictable stepping through in a do-block.
Duplicate breakpoints in the same location are avoided using the
existing blacklist mechanism, which was missing a check in one relevant place.
Fixes #25932
- - - - -
99a3affd by Matthew Pickering at 2025-04-08T17:44:08-04:00
driver: refactor: Split downsweep and MakeAction into separate modules.
This will facilitate using the downsweep functions in other parts of
the compiler than just --make mode.
Also, the GHC.Driver.Make module was huge. Now it's still huge but
slightly smaller!
- - - - -
ecfec4df by sheaf at 2025-04-09T14:13:12-04:00
Store user-written qualification in the GhcRn AST
This commit ensures we store the original user-written module
qualification in the renamed AST. This allows us to take into account
the user-written qualification in error messages.
Fixes #25877
- - - - -
97c884e2 by sheaf at 2025-04-09T14:13:12-04:00
TcRnIllegalTermLevelUse: simpler error when possible
This commit makes GHC emit a simple error message in the case of an
illegal term-level use of a data constructor: we will try to report an
out-of-scope error instead of a "Illegal term level use" error, as the
latter might be a bit overwhelming for newcomers.
We do this when we have a data constructor import suggestion to provide
to the user. For example:
module M where { data A = A }
module N where
import M(A)
x = Bool
-- Illegal term-level use of Bool
y = A
-- Data constructor not in scope: A.
-- Perhaps add 'A' to the import list of 'M'.
This commit also revamps the "similar names" suggestion mechanism,
and in particular its treatment of name spaces. Now, which name spaces
we suggest is based solely on what we are looking for, and no longer on
the NameSpace of the Name we have. This is because, for illegal term-level
use errors, it doesn't make much sense to change the suggestions based
on the fact that we resolved to e.g. a type constructor/class; what
matters is what we were expecting to see in this position.
See GHC.Rename.Unbound.{suggestionIsRelevant,relevantNameSpace} as well
as the new constructors to GHC.Tc.Errors.Types.WhatLooking.
Fixes #23982
- - - - -
bff645ab by Rodrigo Mesquita at 2025-04-09T14:13:57-04:00
driver: Split Session functions out of Main
This commit moves out functions that help in creating and validating a
GHC multi session from Main into the ghc library where they can be used by
other GHC applications.
Moreover, `Mode` processing and `checkOptions` linting were moved to
separate modules within the ghc-bin executable package.
In particular:
- Move `Mode` types and functions (referring to the mode GHC is running
on) to `ghc-bin:GHC.Driver.Session.Mode`
- Move `checkOptions` and aux functions, which validates GHC DynFlags
based on the mode, to `ghc-bin:GHC.Driver.Session.Lint`
- Moves `initMulti`, `initMake`, and aux functions, which initializes a make/multi-unit
session, into `ghc:GHC.Driver.Session.Units`.
- - - - -
501b015e by Rodrigo Mesquita at 2025-04-09T14:13:57-04:00
docs: Improve haddock of ExecComplete
- - - - -
dea98988 by Andreas Klebinger at 2025-04-09T19:23:57-04:00
Avoid oversaturing constructor workers.
Constructor applications always need to take the exact number of
arguments. If we can't ensure that instead apply the constructor worker
like a regular function.
Fixes #23865
- - - - -
f1acdd2c by sheaf at 2025-04-09T19:25:41-04:00
NamedDefaults: require the class to be standard
We now only default type variables if they only appear in constraints
of the form `C v`, where `C` is either a standard class or a class with
an in-scope default declaration.
This rectifies an oversight in the original implementation of the
NamedDefault extensions that was remarked in #25775; that implementation
allowed type variables to appear in unary constraints which had arbitrary
classes at the head.
See the rewritten Note [How type-class constraints are defaulted] for
details of the implementation.
Fixes #25775
Fixes #25778
- - - - -
5712e0d6 by Vladislav Zavialov at 2025-04-10T05:17:38+00:00
Retry type/class declarations and instances (#12088)
Retry type/class declarations and instances to account for non-lexical
dependencies arising from type/data family instances.
This patch improves the kind checker's ability to use type instances in kind
checking of other declarations in the same module.
* Key change: tcTyAndClassDecls now does multiple passes over the TyClGroups,
as long as it is able to make progress.
See the new Note [Retrying TyClGroups] in GHC.Tc.TyCl
* Supporting change: FVs of a TyClGroup are now recorded in its extension
field, namely XCTyClGroup.
See the new Note [Prepare TyClGroup FVs] in GHC.Rename.Module
* Instances are no longer inserted at the earliest positions where their FVs
are bound. This is a simplification.
See the new Note [Put instances at the end] in GHC.Rename.Module
* Automatic unpacking is now more predictable, but fewer fields get unpacked
by default. Use explicit {-# UNPACK #-} pragmas instead.
See the new Note [Flaky -funbox-strict-fields with type/data families]
For the wide range of newly accepted programs, consult the added test cases.
Fixed tickets:
#12088, #12239, #14668, #15561, #16410, #16448, #16693,
#19611, #20875, #21172, #22257, #25238, #25834
Metric Decrease:
T8095
- - - - -
bc73a78d by sheaf at 2025-04-10T15:07:24-04:00
checkFamApp: don't be so eager to cycle break
As remarked in #25933, a pure refactoring of checkTyEqRhs in
ab77fc8c7adebd610aa0bd99d653f9a6cc78a374 inadvertently changed behaviour,
as it caused GHC to introduce cycle-breaker variables in some
unnecessary circumstances.
This commit refactors 'GHC.Tc.Utils.Unify.checkFamApp' in a way that
should restore the old behaviour, so that, when possible, we first
recur into the arguments and only introduce a cycle breaker if this
recursion fails (e.g. due to an occurs check failure).
Fixes #25933
- - - - -
3acd8182 by Andreas Klebinger at 2025-04-10T22:32:12-04:00
Expand docs for RTS flag `-M`.
The behaviour of how/when exceptions are raised was not really covered
in the docs.
- - - - -
026c1a39 by Adam Sandberg Ericsson at 2025-04-10T22:32:56-04:00
add cases for more SchedulerStatus codes in rts_checkSchedStatus
- - - - -
5977c6a1 by sheaf at 2025-04-10T22:33:46-04:00
Squash warnings in GHC.Runtime.Heap.Inspect
There were incomplete record selector warnings in GHC.Runtime.Heap.Inspect
due to the use of the partial 'dataArgs' record selector. This is fixed
by passing the fields to the 'extractSubTerms' function directly,
rather than passing a value of the parent data type.
- - - - -
6a3e38f5 by Andreas Klebinger at 2025-04-11T15:13:53-04:00
hadrian: Make ghcWithInterpreter the universal source of truth about availability of the interpreter
We were doing some ad-hoc checks in different places in hadrian to
determine whether we supported the interprter or not. Now this check if
confined to one function, `ghcWithInterpreter`, and all the places which
use this information consult `ghcWithInterpreter` to determine what to
do.
Fixes #25533.
- - - - -
207de6f1 by Matthew Pickering at 2025-04-11T15:14:37-04:00
testsuite: Fix running TH tests with profiled dynamic compiler
Previously, I had failed to update the ghc_th_way_flags logic for the
profiled dynamic compiler.
In addition to this `ghc_dynamic` was incorrectly set for profiled
dynamic compiler.
I also updated MultiLayerModulesTH_OneShot test to work for any compiler
linkage rather than just dynamic.
Fixes #25947
-------------------------
Metric Decrease:
MultiLayerModulesTH_Make
-------------------------
- - - - -
5455f2b9 by Matthew Pickering at 2025-04-12T08:31:36-04:00
driver: Add support for "Fixed" nodes in the ModuleGraph
A fixed node in the module graph is one which we presume is already
built. It's therefore up to the user to make sure that the interface
file and any relevant artifacts are available for a fixed node.
Fixed/Compile nodes are represented by the ModuleNodeInfo type, which
abstracts the common parts of Fixed/Compile nodes with accessor
functions of type `ModuleNodeInfo -> ...`.
Fixed nodes can only depend on other fixed nodes. This invariant can be
checked by the function `checkModuleGraph` or `mkModuleGraphChecked`.
--make mode is modified to work with fixed mode. In order to "compile" a
fixed node, the artifacts are just loaded into the HomePackageTable.
Currently nothing in the compiler will produce Fixed nodes but this is
tested with the FixedNodes GHC API test.
In subsequent patches we are going to remove the ExternalModuleGraph and
use Fixed nodes for modules in the module graph in oneshot mode.
Fixes #25920
- - - - -
ad64d5c2 by Cheng Shao at 2025-04-12T08:32:19-04:00
ci: remove manual case of ghc-wasm-meta downstream testing jobs
This patch removes the manual case of ghc-wasm-meta downstream testing
jobs; now the only way of including them in the pipeline and running
them is via the test-wasm label.
The reason of the removal is it proves to be problematic for MRs with
only the wasm label; the wasm job would succeed, then the pipeline
status would be waiting for manual action instead of succeeding. There
needs to be separate jobs for the label-triggered and manual-triggered
cases, but I don't think it's worth that extra complexity, the
label-triggered case is already sufficient.
- - - - -
b34890c7 by Vladislav Zavialov at 2025-04-13T01:08:21+03:00
Fix EmptyCase panic in tcMatches (#25960)
Due to faulty reasoning in Note [Pattern types for EmptyCase],
tcMatches was too keen to panic.
* Old (incorrect) assumption: pat_tys is a singleton list.
This does not hold when \case{} is checked against a function type
preceded by invisible forall. See the new T25960 test case.
* New (hopefully correct) assumption: vis_pat_tys is a singleton list.
This should follow from:
checkArgCounts :: MatchGroup GhcRn ... -> TcM VisArity
checkArgCounts (MG { mg_alts = L _ [] })
= return 1
...
- - - - -
84806ebc by Vladislav Zavialov at 2025-04-13T11:40:08-04:00
Remove unused type: TokenLocation
- - - - -
05eb50df by Vladislav Zavialov at 2025-04-13T19:16:38-04:00
Register EpToken in Parser.PostProcess.Haddock (#22558)
This change allows us to reject more badly placed Haddock comments.
Examples:
module
-- | Bad comment for the module
T17544_kw where
data Foo -- | Bad comment for MkFoo
where MkFoo :: Foo
newtype Bar -- | Bad comment for MkBar
where MkBar :: () -> Bar
class Cls a
-- | Bad comment for clsmethod
where
clsmethod :: a
- - - - -
01944e5e by Vladislav Zavialov at 2025-04-13T19:17:21-04:00
Reject puns in T2T (#24153)
This patch implements pun detection in T2T. Consider:
x = 42
f, g :: forall a -> ...
f (type x) = g x
In accordance with the specification, the `g x` function call is renamed
as a term, so `x` refers to the top-level binding `x = 42`, not to the
type variable binding `type x` as one might expect.
This is somewhat counterintuitive because `g` expects a type argument.
Forbidding puns in T2T allows us to produce a helpful error message:
Test.hs:5:16: error: [GHC-09591]
Illegal punned variable occurrence in a required type argument.
The name ‘x’ could refer to:
‘x’ defined at Test.hs:3:1
‘x’ bound at Test.hs:5:9
This commit is a follow up to 0dfb1fa799af254c8a1e1045fc3996af2d57a613
where checking for puns was left as future work.
- - - - -
cc580552 by Vladislav Zavialov at 2025-04-13T19:18:02-04:00
Additional test cases for #12088, #13790
Extract more test cases from ticket discussions, including multi-module
examples. Follow up to 5712e0d646f611dfbfedfd7ef6dff3a18c016edb
- - - - -
d47bf776 by Matthew Pickering at 2025-04-14T16:44:41+01:00
driver: Use ModuleGraph for oneshot and --make mode
This patch uses the `hsc_mod_graph` field for both oneshot and --make
mode. Therefore, if part of the compiler requires usage of the module
graph, you do so in a uniform way for the two different modes.
The `ModuleGraph` describes the relationship between the modules in the
home package and units in external packages. The `ModuleGraph` can be
queried when information about the transitive closure of a package is
needed. For example, the primary use of the ModuleGraph from within the
compiler is in the loader, which needs to know the transitive closure of
a module so it can load all the relevant objects for evaluation.
In --make mode, downsweep computes the ModuleGraph before any
compilation starts.
In oneshot mode, a thunk is created at the start of compilation, which
when forced will compute the module graph beneath the current module.
The thunk is only forced at the moment when the user uses Template
Haskell.
Finally, there are some situations where we need to discover what
dependencies to load but haven't loaded a module graph at all. In this
case, there is a fallback which computes the transitive closure on the
fly and doesn't cache the result. Presumably if you are going to call
getLinkDeps a lot, you would compute the right ModuleGraph before you
started.
Importantly, this removes the ExternalModuleGraph abstraction. This was quite
awkward to work with since it stored information about the home package
inside the EPS.
This patch will also be very useful when implementing explicit level
imports, which requires more significant use of the module graph in
order to determine which level instances are available at.
Towards #25795
-------------------------
Metric Decrease:
MultiLayerModulesTH_Make
MultiLayerModulesTH_OneShot
-------------------------
- - - - -
395e0ad1 by sheaf at 2025-04-16T12:33:26-04:00
base: remove .Internal modules (e.g. GHC.TypeLits)
This commit removes the following internal modules from base,
as per CLC proposal 217:
- GHC.TypeNats.Internal
- GHC.TypeLits.Internal
- GHC.ExecutionStack.Internal
Fixes #25007
- - - - -
e0f3ff11 by Patrick at 2025-04-17T04:31:12-04:00
Refactor Handling of Multiple Default Declarations
Fixes: #25912, #25914, #25934
Previously, GHC discarded all loaded defaults (tcg_default) when local
defaults were encountered during typechecking. According to the
exportable-named-default proposal (sections 2.4.2 and 2.4.3), local
defaults should be merged into tcg_default, retaining any defaults
already present while overriding where necessary.
Key Changes:
* Introduce DefaultProvenance to track the origin of default declarations
(local, imported, or built-in), replacing the original cd_module
in ClassDefaults with cd_provenance :: DefaultProvenance.
* Rename tcDefaults to tcDefaultDecls, limiting its responsibility to only
converting renamed class defaults into ClassDefaults.
* Add extendDefaultEnvWithLocalDefaults to merge local defaults into the
environment, with proper duplication checks:
- Duplicate local defaults for a class trigger an error.
- Local defaults override imported or built-in defaults.
* Update and add related notes: Note [Builtin class defaults],
Note [DefaultProvenance].
* Add regression tests: T25912, T25914, T25934.
Thanks sam and simon for the help on this patch.
Co-authored-by: sheaf