| |
| |
| |
|
Page: 1 2 3 4 5 6 7 8 9 10 11 12
Comments:
<0> void Foo();
<1> but what do you mean about "see from a function declaration"
<1> see what? the stars? your own arsehole without a mirror? what?!?! ;)
<0> I mean "given a function declaration"
<2> check the contents of the function, (header, body) to see if there's any c++-only syntax.
<1> ah i get it
<2> to determine if the function is either definitely c++ or either c++ or c.
<0> There is no body, just the declaration.
<2> for shame.
<1> well i mean you could write a C parser and c++ parser and see if either accept the string heh
<1> but within the language itself?!
<0> A template will be C++, a func decl with a (return) type that contains the scope operator (::) will be C++, right?
<2> i don't know of any other identifying characteristics.
<0> Can structs have member functions in C? :)
<1> no, structs cannot
<2> they can in c++...
<1> also if you have reference parameters it's c++, as well as default parameter values, that's not C
<2> i dunno about that..
<0> Lemme look for a declaration now that contains a quote... if there isn't one, then I don't have to worry about a "::" between quotes.
<1> hmm what else? oh and i think if you do: foo() a strict C compiler should complain it should be written as -- foo(void)
<3> if it's compiled in a C++ compiler and not in an extern "c" block, it's C++
<1> yes, that is true as well.
<3> Trying to determine it by source code analysis is an exercise in frustration
<4> lo, guys, what's up?
<1> hey speaking of that.. asriel.. what is the best policy on function pointers p***ed to C functions (like signal() or pthread_create()).. is it best to declare those as extern "C"?
<4> Nilac it shouldn't matter
<1> i never bother declaring them as extern C because i have never seen it fail..
<4> the pointer has resolved all the decorating issues
<1> vawjrwrk, yeah i figured it shouldn't since C and C++ both have pretty much the same calling convention
<1> and the extern thing was only for mangling issues, right?
<3> linkage issues, but basically eys
<4> well, it's for mangling/decorating issues for sure...there might be others
<1> can extern "C" functions still throw exceptions? heh..
<1> and if so.. is that bad?
<5> btw, if one declaration has c++ stuff, the rest of the declarations in the same file are c++ as well :P
<5> #includes of <cplusplus_header> are a good indication of c++ as well
<5> heh
<4> what sparked this discussion?
<5> Run
<5> lol
<0> I see no " in any function declaration of this project... but that doesn't say much.
<0> Can't some template parameter (literal) contain a quote?
<1> run what do you mean about quotes?
<1> run: yes, i mean if the parameter would then be of type const char * const...
<0> The type of declarations I have here like this:
<0> void std::vector<_Tp, _Alloc>::push_back(const _Tp&) [with _Tp = std::string, _Alloc = std::allocator<std::string>]
<0> There is no '"' in there.
<0> I wonder if it's possible to have something like .. with _Tp = "some \"::\" string", ...
<2> but there's scope resolution...
<0> In other words: do I need to parse quotes or not.
<4> I'm lost
<2> ah
<2> you might as well parse them, just in case..
<1> i think a declaration can contain quotes if it gives default parameters, etc..
<5> just get a normal lexer and voila
<0> I suppose can be fairly sure that any template will contain a '<' and no C function will contain a '<'.
<4> outside of
<4> ".." yes
<2> i'm hungry.
<1> for instance: void foo(const char * str = "blah");
<0> Nilac: Ok, but there are no default parameters at linker level, and these declarations only show what the linker/demangler sees.
<4> did they make that legal in C99?
<1> vawjrwrk, what's so illegal about it?
<1> Run, ah, i see. nevermind. :)
<4> Run working on the demangler again?
<0> no
<0> This is printed directly by the compiler (a patched compiler, to make it print this)
<0> I suppose the compiler knows what language it is, and I could have it print that too... but I think it's simpler to just parse it :p
<0> Can't be that hard to see if it's C++ or not.
<1> vawjrwrk, oh wait i just understood your C99 comment.. i didn't realize c99 also has default params now.. wow
<4> Nilac I don't know that it does, that was a question I asked, not a statement
<1> oh
<0> I threw away the most obvious C++ functions now: those containing "::" and those containing a "<".
<0> That still leaves me with C++ declarations:
<0> "int __cxa_atexit(void (*)(void*), void*, void*)"
<0> "void operator delete [](void*)"
<1> in some ways i find it a bit unsettling how C99 is more liberal about some things than C++
<4> I find it unsettling that it exists
<0> Hmm, that's it. So, if I scan for 'operator' and '(*)' I'd have caught everything for THIS project... but I'm not sure that will work in general.
<0> It's the void (*)(void*) that isn't C I thought... maybe I'm wrong.
<1> no run, that's C allright
<0> I haven't done C for too long :/
<0> There was some big different between C and C++ there though :/
<0> overloading?
<1> on what? function pointers?
<1> well C++ supports pointers to members
<1> which of course C doesn't since there are no members to speak of..
<0> Hmm... C functions don't have parameter types at ALL.
<1> at the link level? i wouldn't be surprised..
<0> But this prints them anyway, so the only way to detect C++ here then is to see if there are any functions overloaded.
<0> If I see "int f(int)" then that can be C, but when I also see "int f(float)" then I know 'f' is a C++ function... I suppose.
<0> Grrmbl.
<0> Ok, this isn't going to work thus.
<0> Once could have an extern "C" int f(int); and a normal int f(float); in a C++ file. Then I can't know which is C and which is C++.
<0> s/Once/One/
<0> Lemme try that...
<1> hey what's that about the unnamed namespace? stuff in it is always static?
<0> no
<0> I think it's exported, you just don't know the name of the namespace.
<1> so how do you refer to it?
<1> or you can't?
<1> ooutside of it..
<0> you can't
<4> operator would be a legal name in C wouldn't it?
<1> so you can have namespace { int foo; } namespace { int foo; } in the same file and it will be two different foo's?
<1> and whey will never clash?
<4> Nilac I don't think so
<1> i wonder how the compiler goes about doing that bit of mangling.. i guess it doesn't export the names
<0> "int test()"
<0> "int test(float)"
<0> Hey hey! It prints () for the extern "C"! Not (int).
<4> and of course it doesn't expport the names.... the whole point of unnamed namespace is that
<1> so it's equivalent to static linkage?
<0> It *does* export it.
<0> /usr/src/cppgraph/cppgraph-objdir/src>nm -C troep.o
<0> 00000000 T (anonymous namespace)::f(int)
<0> 'T' means it's a totally normal (exported) function.
<0> Nilac: yes
<0> But not because it isn't exported, but because you can't name it from outside the same compilation unit.
<1> just why does it export it then? does that mean two namespace { int foo; } in two different translation units end up referring to the same foo?
<0> With dlopen and some bit-level tricks you would still be able to link with it.
<0> Just need to find the name ;)
<0> (in my last test case, that would be: _ZN37_GLOBAL__N_troep.cc_00000000_795DA9251fEi)
<0> If you ignore the anon namespace mangling there (the _GLOBAL_N_), the name would be:
<0> int _GLOBAL__N_troep.cc_00000000_795DA9251::f(int);
<0> (apart from that '.' is illegal in function names)
<1> actually the compiler makes anonymous namespaces contain the filename from whence they came.. neat.. so they are effectively static.. just they are exported :)
<0> The name of the namespace is uniq for that compilation unit: it exists of the name of the compilation unit (troep.cc) plus something that makes it uniq inside that per anon namespace PLUS a random hash (just to be sure).
<6> mmm... hash...
<7> http://today.reuters.com/news/newsarticle.aspx?type=oddlyEnoughNews&storyid=2006-01-13T141713Z_01_EIC351411_RTRUKOC_0_US-VAMPIRE.xml
<0> Well, it's not random... and this is only for g++ of course.
<7> Sharkey also pledged to execute convicted murders and child molesters personally by impaling them on a wooden pole outside the state capitol.
<7> hahaha
<7> that's the best part
<0> sick
<0> Anyone if favour of torture should be kicked out of politics immediately.
<6> that's not torture.
<6> you'd die pretty quick.
<0> yeah right
<6> maybe cruel and unusual...
<0> I'd prefer a bullet through the head.
<8> depends on where they are impaled
<8> through the stomach wouldn't be that pleasant
<6> indeed.
<6> torture implies the possibility of survival.
<8> hilarious political stance, however
<0> This is torture till death, and you know it. Don't give me this **** NineVolt
<6> i guess.
<0> It's disgusting.
Return to
#c++
or
Go to some related
logs:
resimtakasi
What is a pipistrel or a flittermouse ?
Pi-ADL
#linux
deeza
#linux
#ubuntu
ubuntu valarray
#teens
yourporn
|
|