fixes#23499
In the
8990626ca9
the effect of `skipAddr` changed to skip `nkAddr` and `nkHiddenAddr`.
Some old code was not adapted. In the
https://github.com/nim-lang/Nim/pull/23477, the magic `addr` function
was handled in the semantic analysis phase, which causes it be skipped
incorrectly
fixes#23326
In a routine declaration node in a template, if the routine is marked as
`gensym`, the compiler adds it as a new symbol to a preliminary scope of
the template. If it's not marked as gensym, then it searches the
preliminary scope of the template for the name of the routine, then when
it matches a template parameter or a gensym identifier, the compiler
replaces the name node with a symbol node of the found symbol.
This makes sense for the template parameter since it has to be replaced
later, but not really for the gensym identifier, as it doesn't allow us
to inject a routine with the same name as an identifier previously
declared as gensym (the problem in #23326 is when this is in another
`when` branch).
However this is the only channel to reuse a gensym symbol in a
declaration, so maybe removing it has side effects. For example if we
have:
```nim
proc foo(x: int) {.gensym.} = discard
proc foo(x: float) {.gensym.} = discard
```
it will not behave the same as
```nim
proc foo(x: int) {.gensym.} = discard
proc foo(x: float) = discard
```
behaved previously, which maybe allowed overloading over the gensym'd
symbols.
A note to the "undeclared identifier" error message has also been added
for a potential error code that implicitly depended on the old behavior
might give, namely ``undeclared identifier: 'abc`gensym123'``, which
happens when in a template an identifier is first declared gensym in
code that doesn't compile, then as a routine which injects by default,
then the identifier is used.
fixes#16771
follow up https://github.com/nim-lang/Nim/pull/16536
Ideally it should be handled in the IR part in the future
I have also checked the double evaluation of `swap` in the JS runtime
https://github.com/nim-lang/Nim/issues/16779, that might be solved by a
copy flag or something. Well, it should be best solved in the IR so that
it doesn't bother backends anymore.
fixes#23399
The new case introduced in #21657 is triggered by `efWantStmt` but the
`when` statement doesn't normally propagate this flag, so propagate it
when the `semCheck` param in `semWhen` is true which happens when the
`when` statement is `efWhenStmt` anyway.
fixes#22284fixes#22282
```
Error: j(uRef, proc (config: F; sources: auto) {.raises: [].} = discard ) can raise an unlisted exception: Exception
```
The problem is that `n.typ.n` contains the effectList which shouldn't
appear in the parameter of a function defintion. We could not simply use
`n.typ.n` as `n[paramsPos]`. The effect lists should be stripped away
anyway.
ref #23354
The new move analyzer requires types that have the tfAsgn flag
(otherwise `lastRead` will return true); tfAsgn is included when the
destructor is not trival. But it should consider the assignement for
objects in this case because objects might have a trival destructors but
it's the assignement that matters when it is passed to sink parameters.
This also prevents unwanted `raises: [ValueError]` effects from bubbling
up from correct format strings which makes `fmt` broadly unusable with
`raises`.
The old runtime-based `formatValue` overloads are kept for
backwards-compatibility, should anyone be using runtime format strings.
---------
Co-authored-by: Andreas Rumpf <rumpf_a@web.de>
This fixes an oversight with a change that I made a while ago.
Basically, these two snippets should both compile. Currently the
`varargs` version will fail.
```nim
template s(d: typed)=discard
proc something()=discard
proc something(x:int)=discard
s(something)
```
```nim
template s(d: varargs[typed])=discard
proc something()=discard
proc something(x:int)=discard
s(something)
```
Potentially unrelated, but this works currently for some reason:
```nim
template s(a: varargs[typed])=discard
proc something()=discard
proc something(x:int)=discard
s:
something
```
also, this works:
```nim
template s(b:untyped, a: varargs[typed])=discard
proc something()=discard
proc something(x:int)=discard
s (g: int):
something
```
but this doesn't, and the error message is not what I would expect:
```nim
template s(b:untyped, a: varargs[typed])=discard
proc something()=discard
proc something(x:int)=discard
s (g: int), something
```
So far as I can tell, none of these issues persist for me after the code
changes in this PR.
fixes#22909
required by https://github.com/nim-lang/Nim/pull/23267
```nim
proc foo: string =
assert false
result = ""
```
In the function `foo`, `assert false` raises an exception, which can
cause `result` to be uninitialized if the default result initialization
is optimized out
fixes#23247closes#23251 (which accounts for why the openarray type is lifted
because ops are lifted for openarray conversions)
related: https://github.com/nim-lang/Nim/pull/18713
It seems to me that openarray doesn't own the data, so it cannot destroy
itself. The same case should be applied to
https://github.com/nim-lang/Nim/issues/19435. It shouldn't be destroyed
even openarray can have a destructor. A cleanup will be followed for
https://github.com/nim-lang/Nim/pull/19723 if it makes sense.
According to https://github.com/nim-lang/Nim/pull/12073, it lifts
destructor for openarray when openarray is sunk into the function, when
means `sink openarray` owns the data and needs to destroy it. In other
cases, destructor shouldn't be lifted for `openarray` in the first place
and it shouldn't destroy the data if it doesn't own it.
---------
Co-authored-by: Andreas Rumpf <rumpf_a@web.de>
fixes#22597
```nim
proc autoToOpenArray*[T](s: Slice[T]): openArray[T] =
echo "here twice"
result = toOpenArray(s.p, s.first, s.last)
```
For functions returning openarray types, `fixupCall` creates a temporary
variable to store the return value: `let tmp = autoToOpenArray()`. But
`genOpenArrayConv` cannot handle openarray assignements with side
effects. It should have stored the right part of the assignment first
instead of calling the right part twice.
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.
fixes#17163, refs #23204
Types that aren't `tyRange` and are bigger than 16 bits, so `int32`,
`uint64`, `int` etc, are disallowed as array index range types.
`tyRange` is excluded because the max array size is backend independent
(except for the specific size of `high(uint64)` which crashes the
compiler) and so there should still be an escape hatch for people who
want bigger arrays.
This PR modifies `specializeResetT` so that it generates the proper
reset code for range types. I've tested it in the examples for issues
#23214 and #22462 as well as our codebase, and it seems to fix the
issues I had been experiencing.
