fixes#15314, fixes#24002
The OpenSym behavior first added to generics in #23091 now also applies
to templates, since templates can also capture symbols that are meant to
be replaced by local symbols if the context imports symbols with the
same name, as in the issue #24002. The experimental switch
`templateOpenSym` is added to enable this behavior for templates only,
and the experimental switch `openSym` is added to enable it for both
templates and generics, and the documentation now mainly mentions this
switch.
Additionally the logic for `nkOpenSymChoice` nodes that were previously
wrapped in `nkOpenSym` now apply to all `nkOpenSymChoice` nodes, and so
these nodes aren't wrapped in `nkOpenSym` anymore. This means
`nkOpenSym` can only have children of kind `nkSym` again, so it is more
in line with the structure of symchoice nodes. As for why they aren't
merged with `nkOpenSymChoice` nodes yet, we need some way to signal that
the node shouldn't become ambiguous if other options exist at
instantiation time, we already captured a symbol at the beginning and
another symbol can only replace it if it's closer in scope and
unambiguous.
(cherry picked from commit 770f8d5513)
Close#22826
I am not sure why this code skips generic insts, so letting CI tell me.
Update: It has told me nothing. Maybe someone knows during review.
Issue itself seems to be that the generic instance is skipped thus it
ends up being just `float` which makes it use the wrong generic instance
of the proc because it matches the one in cache
---------
Co-authored-by: SirOlaf <>
(cherry picked from commit c13c48500b)
fixes#24104, refs #23955
The line `result.typ = dstTyp` added in #23955 changes the type of
`result`, which was the type of `n` due to the argument passed to
`newIntNodeT`, to the abstract type skipped `dstTyp`. The line is
removed to just keep the type as abstract.
(cherry picked from commit 6d362e0ffe)
It speeds up
```nim
proc foo =
let piece = cast[seq[char]](newSeqUninit[uint8](5220600386'i64))
foo()
```
Notes that `cast[ref](...)` is excluded because we need to keep the ref
alive if the parameter is something with pointer types (e.g.
`cast[ref](pointer)`or `cast[ref](makePointer(...))`)
---------
Co-authored-by: Andreas Rumpf <rumpf_a@web.de>
(cherry picked from commit 4ef06a5cc5)
fixes#8697, fixes#9620, fixes#23265
When matching a `template` with an `untyped` argument fails because of a
mismatching typed argument, `presentFailedCandidates` tries to sem every
single argument to show their types, but trying to type the `untyped`
argument can fail if it's supposed to use an injected symbol, so we get
an unrelated error message like "undeclared identifier".
Instead we use `tryExpr` as the comment suggests, setting the type to
`untyped` if it fails to compile. We could also maybe check if an
`untyped` argument is expected in its place and not try to compile the
expression if it is but this would require a bit of reorganizing the
code here and IMO it's better to have the information of what type it
would be if it can be typed.
(cherry picked from commit 34719cad9d)
fixes#23977
The problem is that for *any* body of a generic declaration,
[semstmts](2e4d344b43/compiler/semstmts.nim (L1610-L1611))
sets the sym of its value to the generic type name, and
[semtypes](2e4d344b43/compiler/semtypes.nim (L2143))
just directly gives the referenced type *specifically* when the
expression is a generic body. I'm blaming `semtypes` here because it's
responsible for the type given but the exact opposite behavior
specifically written in makes me think generating an alias type here
maybe breaks something.
(cherry picked from commit 58813a3b2e)
fixes#10753, fixes#22021, refs #19365 (was fixed by #22029, but more
faithful test added)
For whatever reason `compileTime` proc calls did not fold if the proc
was generic ([since this folding was
introduced](c25ffbf262 (diff-539da3a63df08fa987f1b0c67d26cdc690753843d110b6bf0805a685eeaffd40))).
I'm guessing the intention was for *unresolved* generic procs to not
fold, which is now the logic.
Non-magic `compileTime` procs also now don't fold at compile time in
`typeof` contexts to avoid possible runtime errors (only the important)
and prevent double/needless evaluation.
(cherry picked from commit f7c11a8978)
fixes#22850
The `is` operator checks the type of the left hand side, and if it's
generic or if it's a `typedesc` type with no base type, it leaves it to
be evaluated later. But `typedesc` types with no base type precisely
describe the default typeclass `type`/`typeclass`, so this condition is
removed. Maybe at some point this represented an unresolved generic
type?
(cherry picked from commit 995081b56a)
Provides a fix for #23817.
With target `arm-none-eabi`, GCC defines `int32_t` to `long int`. Nim
uses `__builtin_sadd_overflow` for 32-bit targets, but this emits
warnings on GCC releases 13 and under, while generating an error on GCC
14. More info regarding this
[here](https://gcc.gnu.org/gcc-14/porting_to.html#c) and
[here](https://gcc.gnu.org/pipermail/gcc-cvs/2023-December/394351.html).
The proposed PR attempts to address this issue for these targets by
defining the `nimAddInt`, `nimSubInt`, and `nimMulInt` macros to use the
appropriate compiler intrinsics for this platform.
As for as we know, the LLVM toolchain for bare metal Arm does not define
`int32_t` as `long int` and has no need for this patch. Thus, we only
define the above macros for GCC targeting `arm-non-eabi`.
(cherry picked from commit c5b206d4ac)
fixes#23853
Since #22610 generics turns the `Name` in the `GT.Name` expression in
the test code into a sym choice. The problem is when the compiler tries
to instantiate `GT.Name` it also instantiates the sym choice symbols.
`Name` has type `template (E: type ExtensionField)` which contains the
unresolved generic type `ExtensionField`, which the compiler mistakes as
an uninstantiated node, when it's just part of the type of the template.
The compilation of the node itself and hence overloading will handle the
instantiation of the proc, so we avoid instantiating it in `semtypinst`,
similar to how the first nodes of call nodes aren't instantiated.
(cherry picked from commit 97f5474545)
ref #20653
```nim
Error* = object
case kind*: ErrorType
of ErrorA:
discard
of ErrorB:
discard
```
For an object variants without fields, it shouldn't generate empty
brackets for default values since there are no fields at all in case
branches.
(cherry picked from commit 948bb38335)
Fix non-exported `setFileSize` to take optional `oldSize` to (on posix)
shrink differently than it grows (`ftruncate` not `posix_fallocate`)
since it makes sense to assume the higher address space has already been
allocated there and include the old file size in the `proc resize` call.
Also, do not even try `setFileSize` in the first place unless the `open`
itself works by moving the call into the `if newFileSize != -1` branch.
Just cosmetics, also improve some old 2011 comments, note a logic diff
for callers using both `mappedSize` & `newFileSize` from windows branch
in case someone wants to fix that & simplify code formatting a little.
(cherry picked from commit 8037bbe327)
fixes#10440, fixes#13871, fixes#14665, fixes#19672, fixes#23677
The false positive in #23677 was caused by behavior in
`implicitlyDiscardable` where only the last node of `if`/`case`/`try`
etc expressions were considered, as in the final node of the final
branch (in this case `else`). To fix this we use the same iteration in
`implicitlyDiscardable` that we use in `endsInNoReturn`, with the
difference that for an `if`/`case`/`try` statement to be implicitly
discardable, all of its branches must be implicitly discardable.
`noreturn` calls are also considered implicitly discardable for this
reason, otherwise stuff like `if true: discardableCall() else: error()`
doesn't compile.
However `endsInNoReturn` also had bugs, one where `finally` was
considered in noreturn checking when it shouldn't, another where only
`nkIfStmt` was checked and not `nkIfExpr`, and the node given for the
error message was bad. So `endsInNoReturn` now skips over
`skipForDiscardable` which no longer contains
`nkIfStmt`/`nkCaseStmt`/`nkTryStmt`, stores the first encountered
returning node in a var parameter for the error message, and handles
`finally` and `nkIfExpr`.
Fixing #23677 already broke a line in `syncio` so some package code
might be affected.
(cherry picked from commit 42e8472ca6)
refs
https://github.com/nim-lang/Nim/issues/23586#issuecomment-2102113750
In #20091 a bad kind of type inference was mistakenly left in where if
an identifier `abc` had an expected type of an enum type `Enum`, and
`Enum` had a member called `abc`, the identifier would change to be that
enum member. This causes bugs where a local symbol can have the same
name as an enum member but have a different value. I had assumed this
behavior was removed since but it wasn't, and CI seems to pass having it
removed.
A separate PR needs to be made for the 2.0 branch because these lines
were moved around during a refactoring in #23123 which is not in 2.0.
(cherry picked from commit c101490a0c)
fixes#18125
Previously a tuple type like `(T, int)` would match an expected tuple
type `(U, int)` if `T` is a subtype of `U`. This is wrong since the
codegen does not handle type conversions of individual tuple elements in
a type conversion of an entire tuple. For this reason the compiler
already does not accept `(float, int)` for a matched type `(int, int)`,
however the code that checked for which relations are unacceptable
checked for `< isSubtype` rather than `<= isSubtype`, so subtypes were
not included in the unacceptable relations.
Update: Now only considered unacceptable when inheritance is used, as in
[`paramTypesMatch`](3379d26629/compiler/sigmatch.nim (L2252-L2254)).
Ideally subtype relations that don't need conversions, like `nil`,
`seq[empty]`, `range[0..5]` etc would be their own relation
`isConcreteSubtype` (which would also allow us to differentiate with
`openArray[T]`), but this is too big of a refactor for now.
To compensate for this making things like `let x: (Parent, int) =
(Child(), 0)` not compile (they would crash codegen before anyway but
should still work in principle), type inference for tuple constructors
is updated such that they call `fitNode` on the fields and their
expected types, so a type conversion is generated for the individual
subtype element.
(cherry picked from commit cfd69bad1a)
Filling in some more logic in `typeRel` that I came across when poking
the compiler in another PR. Some of the cases where `typeRel` returns an
"incorrect" result are actually common, but `sumGeneric` ends up
breaking the tie correctly. There isn't anything wrong with that
necessarily, but I assume that it's preferred these functions behave
just as well in isolation as they do when integrated.
I will be following up this description with specific examples.
(cherry picked from commit ccc7c45d71)
fixes#23002, fixes#22841, refs comments in #23097
When an identifier is ambiguous in scope (i.e. multiple imports contain
symbols with the same name), attempt resolving it through type inference
(by creating a symchoice). To do this efficiently, `qualifiedLookUp` had
to be broken up so that `semExpr` can access the ambiguous candidates
directly (now obtained directly via `lookUpCandidates`).
This fixes the linked issues, but an example like:
```nim
let on = 123
{.warning[ProveInit]: on.}
```
will still fail, since `on` is unambiguously the local `let` symbol here
(this is also true for `proc on` but `proc` symbols generate symchoices
anyway).
Type symbols are not considered to not confuse the type inference. This
includes the change in sigmatch, up to this point symchoices with
nonoverloadable symbols could be created, they just wouldn't be
considered during disambiguation. Now every proper symbol except types
are considered in disambiguation, so the correct symbols must be picked
during the creation of the symchoice node. I remember there being a
violating case of this in the compiler, but this was very likely fixed
by excluding type symbols as CI seems to have found no issues.
The pure enum ambiguity test was disabled because ambiguous pure enums
now behave like overloadable enums with this behavior, so we get a
longer error message for `echo amb` like `type mismatch: got <MyEnum |
OtherEnum> but expected T`
(cherry picked from commit b280100499)
refs https://github.com/nim-lang/Nim/pull/23873#discussion_r1687995060,
fixes#23386, fixes#23385, supersedes #23572
Turns the `nfOpenSym` node flag implemented in #23091 and extended in
containing either `nkSym` or `nkOpenSymChoice`. Since this affects
macros working on generic proc AST, the node kind is now only generated
when the experimental switch `genericsOpenSym` is enabled, and a new
node flag `nfDisabledOpenSym` is set to the `nkSym` or `nkOpenSymChoice`
when the switch is not enabled so that we can give a warning.
Now that the experimental switch has more reasonable semantics, we
define `nimHasGenericsOpenSym2`.
(cherry picked from commit 0c890ff9a7)
actually fixes#23865 following up #23873
In the handling of `nkIdent` in `semExpr`, the compiler looks for the
closest symbol with the name and [checks the symbol
kind](6126a0bf46/compiler/semexprs.nim (L3171))
to also consider the overloads if the symbol kind is overloadable. But
it treats the normally overloadable template/macro/module sym kinds the
same as non-overloadable symbols, just calling `semSym` on it. We need
to mirror this behavior in `semOpenSym`; we treat the captured symchoice
as a fresh identifier, so if the symbol we find is a
template/macro/module, we use that symbol immediately as opposed to
waiting for overloads.
(cherry picked from commit a64aa51fe9)
fixes#23865
The node flag `nfOpenSym` implemented in #23091 for sym nodes is now
also implemented for open symchoices. This means the intended behavior
is still achieved when multiple overloads are in scope to be captured,
so the issue is fixed. The code for the flag is documented and moved
into a helper proc and the experimental switch is now enabled for the
compiler test suite.
(cherry picked from commit 469a6044c0)
