- Dec 12, 2023
-
-
Resolves #24190 Apparently the check was essentially always (as far as I can trace back: d0d47ba7), and while it does catch some mismatches, the type-checker will catch them too. OTOH, it prevents writing completely reasonable programs.
-
-
...and use it to generate slightly better code when dataToTag# is used at a "small data type" where there is no need to mess with "is_too_big_tag" or potentially look at an info table. Metric Decrease: T18304
-
- Dec 11, 2023
-
-
Arnaud Spiwack authored
For expediency, the initial implementation of linear types in GHC made it so that let and where binders would always be considered unrestricted. This was rather unpleasant, and probably a big obstacle to adoption. At any rate, this was not how the proposal was designed. This patch fixes this infelicity. It was surprisingly difficult to build, which explains, in part, why it took so long to materialise. As of this patch, let or where bindings marked with %1 will be linear (respectively %p for an arbitrary multiplicity p). Unmarked let will infer their multiplicity. Here is a prototypical example of program that used to be rejected and is accepted with this patch: ```haskell f :: A %1 -> B g :: B %1 -> C h :: A %1 -> C h x = g y where y = f x ``` Exceptions: - Recursive let are unrestricted, as there isn't a clear semantics of what a linear recursive binding would be. - Destructive lets with lazy bindings are unrestricted, as their desugaring isn't linear (see also #23461). - (Strict) destructive lets with inferred polymorphic type are unrestricted. Because the desugaring isn't linear (See #18461 down-thread). Closes #18461 and #18739 Co-authored-by: @jackohughes -
Arnaud Spiwack authored
-
-
-
-
-
Before this change, GHC used to accept `forall` as a term-level identifier: -- from constraints-0.13 forall :: forall p. (forall a. Dict (p a)) -> Dict (Forall p) forall d = ... Now it is a parse error. The -Wforall-identifier warning has served its purpose and is now a deprecated no-op.
-
- Dec 08, 2023
-
-
-
Relevant to #3458
-
Simon Peyton Jones authored
This MR fixes a very subtle bug exposed by #24242. See Note [Environment for simplLetUnfolding]. I also updated a bunch of Notes on shadowing
-
Add test case for bug #24224
-
-
This patch does two things, to fix #23209: * It improves SpecConstr so that it no longer quantifies over coercion variables. See Note [SpecConstr and casts] * It improves the rule matcher to deal nicely with the case where the rule does not quantify over coercion variables, but the the template has a cast in it. See Note [Casts in the template]
-
This partially fixes #24229. See the new Note [Pattern duplicate elimination] in SpecConstr
-
Otherwise we run into a panic in hscMaybeWriteIface: "Unexpected DT_Dyn state when writing simple interface" when dynamic-too is enabled We could remove the panic and just write the interface even if the state is `DT_Dyn`, but it seems pointless to run the pipeline twice when `hscMaybeWriteIface` is already designed to write both `hi` and `dyn_hi` files if dynamic-too is enabled. Fixes #23944.
-
- Dec 06, 2023
-
-
See the new `Note [Dead code may contain type confusions]`. Fixes #23862.
-
StgCse can revive dead binders: case foo of dead { Foo x y -> Foo x y; ... } ===> case foo of dead { Foo x y -> dead; ... } -- dead is no longer dead So we must zap occurrence information on case binders. Fix #14895 and #24233 -
Before, the source location would point at the surrounding function definition, causing the confusion in #24234. I also took the opportunity to introduce a new `LazyPatCtx :: HsMatchContext _` to make the warning message say "irrefutable pattern" instead of "pattern binding".
-
This patch implements the T2T (term-to-type) transformation in patterns. Patterns that are checked against a visible forall can now be written without the `type` keyword: \(type t) (x :: t) -> ... -- old \t (x :: t) -> ... -- new The `t` binder is parsed and renamed as a term pattern (Pat), but then undergoes a conversion to a type pattern (HsTyPat). See the new function pat_to_type_pat in compiler/GHC/Tc/Gen/Pat.hs
-
- Nov 30, 2023
-
-
We now require that in 'forall a -> ty', ty has kind TYPE r for some r. Fixes #24176
-
EpaLocation is used to position things. It has two constructors, EpaSpan holding a SrcSpan, and EpaDelta with a delta position and a possible list of comments. The comment list is needed because the location in EpaDelta has no absolute information to decide which comments should be emitted before them when printing. But it is also used for specifying the position of a comment. To prevent the absurdity of a comment position having a list of comments in it, we make EpaLocation parameterisable, using comments for the normal case and a constant for within comments. Updates haddock submodule. aarch64-darwin Metric Decrease: MultiLayerModulesTH_OneShot
-
- Nov 29, 2023
-
-
We no longer need the alternative variant of addLocM (addLocMA) nor wrapLocAM, wrapLocSndMA. aarch64-darwin Metric Increase: MultiLayerModulesTH_OneShot deb10-numa-slow Metric Decrease: libdir -
Objects in the data section may be referenced via tagged pointers. Thus, align those objects to a 4- or 8-byte boundary for 32- or 64-bit platforms, respectively. Note, this may need to be reconsidered if objects with a greater natural alignment requirement are emitted as e.g. 128-bit atomics. Fixes #24163.
-
- Nov 28, 2023
-
-
Now that we have HasLoc we can get rid of all the custom variants of mkScope For deb10-numa Metric Increase: libdir
-
- Nov 27, 2023
-
-
Now that we only have a single constructor for EpAnn, And it uses a SrcSpan for its location, we can do away with SrcSpanAnn completely. It only existed to wrap the original SrcSpan in a location, and provide a place for the exact print annotation. For darwin only: Metric Increase: MultiLayerModulesTH_OneShot Updates haddock submodule -
Fixes #24222
-
- Nov 26, 2023
-
-
The Note was written before GHC gained support for visible forall in types of terms. Rewrite a few sentences and use a better example.
-
This patch changes type variable lookup rules (lookupTypeOccRn) and implicit quantification rules (filterInScope) so that variables bound in the term namespace can be captured at the type level {-# LANGUAGE RequiredTypeArguments #-} f1 x = g1 @x -- `x` used in a type application f2 x = g2 (undefined :: x) -- `x` used in a type annotation f3 x = g3 (type x) -- `x` used in an embedded type f4 x = ... where g4 :: x -> x -- `x` used in a type signature g4 = ... This change alone does not allow us to accept examples shown above, but at least it gets them past the renamer. -
We no longer need the EpAnnNotUsed constructor for EpAnn, as we can represent an unused annotation with an anchor having a EpaDelta of zero, and empty comments and annotations. This simplifies code handling annotations considerably. Updates haddock submodule Metric Increase: parsing001 -
This warning had no name or flag and was triggered unconditionally. Now it is part of -Wcompat.
-
- Nov 24, 2023
-
-
This is called from PostProcess.hs, and adds spurious parens. With the looser version of exact printing we had before we could tolerate this, as they would be swallowed by the original at the same place. But with the next change (remove EpAnnNotUsed) they result in duplicates in the output. For Darwin build: Metric Increase: MultiLayerModulesTH_OneShot -
`genFMA3Code` assumed that we had to take extra precations to avoid overwriting the result of `getNonClobberedReg`. One of these special cases caused a bug resulting in broken assembly. I believe we don't need to hadle these cases specially at all, which means this MR simply deletes the special cases to fix the bug. Fixes #24160
-
- Nov 23, 2023
-
-
In Parser.y, the tup_tail rule had the following option | {- empty -} %shift { return [Left noAnn] } Once this works through PostProcess.hs, it means we add an extra Missing constructor if the last item was a comma. Change the annotation type to a Bool to indicate this, and use the EpAnn Anchor for the print location for the others. -
The hpc testsuite was split between testsuite/tests/hpc and the submodule libraries/hpc/test. This commit unifies the two testsuites in the GHC repository in the directory testsuite/tests/hpc.
-
- Nov 22, 2023
-
-
The Present constructor for a Tuple argument will never have an exact print annotation. So make this impossible.
-
- Nov 20, 2023
-
-
This is more natural, since we already need to deal with invalid RealSrcSpans, and that is exactly what SrcSpan.UnhelpfulSpan is for. Updates haddock submodule.
-
- Nov 18, 2023
-
-
Remove the final Monoid instances in the exact print infrastructure. For Windows CI Metric Decrease: T5205
-
