Make the Visitor class inherit the ParseTimeVisitor

[RazvanN7]

@ibuclaw Is there anything else that should be done in order to make the visitor compatible with gdc? Should the header files be updated so that the inheritance is visible?

[dlang-bot]

Thanks for your pull request, @RazvanN7! We are looking forward to reviewing it, and you should be hearing from a maintainer soon.

Some tips to help speed things up:

• smaller, focused PRs are easier to review than big ones

• try not to mix up refactoring or style changes with bug fixes or feature enhancements

• provide helpful commit messages explaining the rationale behind each change

Bear in mind that large or tricky changes may require multiple rounds of review and revision.

Please see CONTRIBUTING.md for more information.

Bugzilla references

Your PR doesn't reference any Bugzilla issue.

If your PR contains non-trivial changes, please reference a Bugzilla issue or create a manual changelog.

[andralex]

Eh, that's a fair amount of duplication, but I guess it'll do. @RazvanN7 what's with the unittest failures? @ibuclaw does this float your boat?

[RazvanN7]

@andralex I don't know what duplication you are talking about.
Those must have been spurious failures since everything seems to be green now.

Later edit: It seems that travis fails a build on a betterC test, altough I don't think it's related to this PR

[PetarKirov]

Also CC @kinke @JohanEngelen and @klickverbot.

[ibuclaw]

I think that we should hold this off until some tests are in place. Otherwise you're just shooting in the dark here.

[RazvanN7]

@ibuclaw Indeed, but the last PR made it so that the Visitor is just an alias to a templated class. This PR makes the visitor an extern C++ class again so it can't make things worse.

[andralex]

@ibuclaw I'll merge this under the assumption it'll make things better even before having added tests.

[ibuclaw]

@andralex @RazvanN7 - OK I can confirm 👍 on both accounts.

First: I was correct in assuming that this broke C++.
Second: This patch fixes it.

[ibuclaw]

Here's a small test:

#include "expression.h"
#include "globals.h"
#include "module.h"
#include "mtype.h"
#include "objc.h"
#include "target.h"
#include "visitor.h"

/**********************************/

extern "C" int rt_init();
extern "C" void gc_disable();

void initialize()
{
    rt_init();
    gc_disable();

    global._init();
    global.params.isLinux = true;

    Type::_init();
    Module::_init();
    Expression::_init();
    Objc::_init();
    Target::_init();
}

/**********************************/

class TestVisitor : public Visitor
{
  public:
    void visit(Expression *)
    {
        printf("Success!\n");
    }
};

void test_visitors()
{
    TestVisitor tv;
    IntegerExp *ie = IntegerExp::create(Loc(), 42, Type::tint32);
    ie->accept(&tv);
}

/**********************************/

int main(int argc, char **argv)
{
    initialize();

    test_visitors();
}

[ibuclaw]

Without this patch:

$ ../generated/linux/release/64/dmd-cxx
dmd-cxx: dmd/visitor.h:374: virtual void Visitor::visit(Dsymbol*): Assertion `0' failed.
Aborted

Backtrace:

#0  0x00007ffff6d75428 in __GI_raise (sig=sig@entry=6) at ../sysdeps/unix/sysv/linux/raise.c:54
#1  0x00007ffff6d7702a in __GI_abort () at abort.c:89
#2  0x00007ffff6d6dbd7 in __assert_fail_base (fmt=<optimized out>, assertion=assertion@entry=0x79410e "0", file=file@entry=0x794100 "dmd/
visitor.h", line=line@entry=374, function=function@entry=0x794220 <Visitor::visit(Dsymbol*)::__PRETTY_FUNCTION__> "virtual void Visitor::
visit(Dsymbol*)") at assert.c:92
#3  0x00007ffff6d6dc82 in __GI___assert_fail (assertion=0x79410e "0", file=0x794100 "dmd/visitor.h", line=374, function=0x794220 <Visitor
::visit(Dsymbol*)::__PRETTY_FUNCTION__> "virtual void Visitor::visit(Dsymbol*)") at assert.c:101
#4  0x00000000006a806b in Visitor::visit (Dsymbol *) at dmd/visitor.h:374
#5  0x00000000006a8820 in Visitor::visit (Declaration *) at dmd/visitor.h:417
#6  0x00000000006a88f0 in Visitor::visit (VarDeclaration *) at dmd/visitor.h:421
#7  0x00000000006a898c in Visitor::visit (TypeInfoDeclaration *) at dmd/visitor.h:425
#8  0x00000000006a8a90 in Visitor::visit (TypeInfoArrayDeclaration *) at dmd/visitor.h:430
#9  0x000000000059cf1a in IntegerExp::accept(GenericVisitor<ASTCodegen>*) (this=0x7ffff7ed51e0, v=0x7fffffffdbd0) at dmd/expression.d:254
#10 0x00000000006a7330 in test_visitors () at dmd/cxxtest.c:61

With this patch:

$ ../generated/linux/release/64/dmd-cxx
Success!

[ibuclaw]

@MartinNowak - We are under freeze for release now right? If applicable, this patch may need to be applied to stable.