[Bug c++/35146] New: weird error in template function specialization

[Bug c++/35146] New: weird error in template function specialization

[grey_earl at web dot de]

[-- Warning: decoded text below may be mangled, UTF-8 assumed --]
[-- Attachment #1: Type: text/plain, Size: 2590 bytes --]

gcc -v:
Using built-in specs.
Target: i486-linux-gnu
Configured with: ../src/configure -v
--enable-languages=c,c++,fortran,objc,obj-c++,treelang --prefix=/usr
--enable-shared --with-system-zlib --libexecdir=/usr/lib
--without-included-gettext --enable-threads=posix --enable-nls
--with-gxx-include-dir=/usr/include/c++/4.2 --program-suffix=-4.2
--enable-clocale=gnu --enable-libstdcxx-debug --enable-mpfr
--disable-libmudflap --enable-targets=all --enable-checking=release
--build=i486-linux-gnu --host=i486-linux-gnu --target=i486-linux-gnu
Thread model: posix
gcc version 4.2.3 20080114 (prerelease) (Debian 4.2.2-7)


The error:
schroed2.cpp: In function ‘void blas_scal(typename ref<R>::result,
complex<T>*) [with T = double]’:
schroed2.cpp:16: error: cannot convert ‘complex<double>’ to ‘double’
for argument ‘1’ to ‘void cblas_zdscal(double, void*)’


The code:
template <typename T> struct complex { T re; T im; };

template <typename R> struct ref { typedef const R& result; };
template <> struct ref<double> { typedef double result; };
template <> struct ref<complex<double> > { typedef complex<double> result; };

void cblas_dscal(double alpha, double *x) { }
void cblas_zscal(void *alpha, void *x) { }
void cblas_zdscal(double alpha, void *x) { }

template <typename T> void blas_scal(typename ref<T>::result alpha, T* x);
template <typename T> void blas_scal(typename ref<T>::result alpha, complex<T>*
x);

template <> void blas_scal(double alpha, double* x) { cblas_dscal(alpha, x); }
template <> void blas_scal(complex<double> alpha, complex<double>* x) {
complex<double> cblas_alpha = alpha; cblas_zscal(&cblas_alpha, x); }
template <> void blas_scal(double alpha, complex<double>* x) {
cblas_zdscal(alpha, x); } // line 16 is here

int main() {
  return 0;
}


The ref-struct allows to use pass-by-reference for big datatypes, pass-by-value
for small ones. If I replace this struct in the template declaration for
blas_scal by T, everything works as expected. For T = double, the last
specialization is imho valid for the second template declaration - I don't even
know what gcc wants from me, since the parameters are exactly what is needed.


-- 
           Summary: weird error in template function specialization
           Product: gcc
           Version: 4.2.3
            Status: UNCONFIRMED
          Severity: normal
          Priority: P3
         Component: c++
        AssignedTo: unassigned at gcc dot gnu dot org
        ReportedBy: grey_earl at web dot de


http://gcc.gnu.org/bugzilla/show_bug.cgi?id=35146

[Bug c++/35146] weird error in template function specialization

[pinskia at gcc dot gnu dot org]

------- Comment #1 from pinskia at gcc dot gnu dot org  2008-02-09 01:45 -------
This works fine on the trunk.


-- 

pinskia at gcc dot gnu dot org changed:

           What    |Removed                     |Added
----------------------------------------------------------------------------
      Known to work|                            |4.3.0


http://gcc.gnu.org/bugzilla/show_bug.cgi?id=35146

[Bug c++/35146] [4.1/4.2 regression] weird error in template function specialization

[bangerth at dealii dot org]

------- Comment #2 from bangerth at dealii dot org  2008-02-10 01:13 -------
Confirmed. This was introduced in 3.4.x, and worked before. As Andrew
already noted, this has been fixed in mainline.

The problem can be more succinctly shown by the following testcase:
--------------------
template <typename T> struct S {};

template <typename R> struct ref;
template <>           struct ref<double> { typedef double result; };

template <typename T> void foo(typename ref<T>::result, S<T>*);
template <>           void foo(S<double>,               S<double>*);
template <>           void foo(double alpha,            S<double>* x)
{
  alpha; x;
}
-----------------------
g/x> c++-4.2.1 -S x.cc
x.cc: In function 'void foo(typename ref<R>::result, S<T>*) [with T = double]':
x.cc:10: error: 'alpha' was not declared in this scope
x.cc:10: error: 'x' was not declared in this scope

What is happening is that we are looking up the parameters from the wrong
declaration. The function in question on line 10 of course has them
declared and named, but we are looking at the specialization in line 7:
if the parameters are named there, then the code compiles. However, as
shown in the original testcase, the parameters then have the wrong
type.

It shouldn't be too hard to construct a case where this produces wrong
code.

Can someone determine whether this is still wrong with the released
4.2.3? I consider this a particularly heinous bug since it's almost
impossible to see in a real code where the error is coming from (if
we get one) or why the produced code is wrong.

W.


-- 

bangerth at dealii dot org changed:

           What    |Removed                     |Added
----------------------------------------------------------------------------
                 CC|                            |bangerth at dealii dot org
             Status|UNCONFIRMED                 |NEW
     Ever Confirmed|0                           |1
      Known to fail|                            |4.2.1 4.1.2 4.0.4 3.4.6
      Known to work|4.3.0                       |4.3.0 3.3.6
   Last reconfirmed|0000-00-00 00:00:00         |2008-02-10 01:13:11
               date|                            |
            Summary|weird error in template     |[4.1/4.2 regression] weird
                   |function specialization     |error in template function
                   |                            |specialization


http://gcc.gnu.org/bugzilla/show_bug.cgi?id=35146

[Bug c++/35146] [4.1/4.2 regression] weird error in template function specialization

[rguenth at gcc dot gnu dot org]

------- Comment #3 from rguenth at gcc dot gnu dot org  2008-02-10 23:06 -------
While the original testcase works for me on trunk, the testcase in comment #2
does 
not.  Or am I missing sth?


-- 


http://gcc.gnu.org/bugzilla/show_bug.cgi?id=35146

[Bug c++/35146] [4.1/4.2 regression] weird error in template function specialization

[bangerth at dealii dot org]

------- Comment #4 from bangerth at dealii dot org  2008-02-11 03:38 -------
(In reply to comment #3)
> While the original testcase works for me on trunk, the testcase in comment #2
> does 
> not.  Or am I missing sth?

I see the same behavior with yesterday's svn version. The testcase in 
comment #2 is definitely valid, so we still have a regression.

W.


-- 


http://gcc.gnu.org/bugzilla/show_bug.cgi?id=35146

[Bug c++/35146] [4.1/4.2 regression] weird error in template function specialization

[grey_earl at web dot de]

------- Comment #5 from grey_earl at web dot de  2008-02-15 17:12 -------
Both testcases fail with 4.2.3 release version, with exactly the same errors.

Using built-in specs.
Target: i486-linux-gnu
Configured with: ../src/configure -v
--enable-languages=c,c++,fortran,objc,obj-c++,treelang --prefix=/usr
--enable-shared --with-system-zlib --libexecdir=/usr/lib
--without-included-gettext --enable-threads=posix --enable-nls
--with-gxx-include-dir=/usr/include/c++/4.2 --program-suffix=-4.2
--enable-clocale=gnu --enable-libstdcxx-debug --enable-objc-gc --enable-mpfr
--disable-libmudflap --enable-targets=all --enable-checking=release
--build=i486-linux-gnu --host=i486-linux-gnu --target=i486-linux-gnu
Thread model: posix
gcc version 4.2.3 (Debian 4.2.3-1)


-- 


http://gcc.gnu.org/bugzilla/show_bug.cgi?id=35146

[Bug c++/35146] [4.1/4.2 regression] weird error in template function specialization

[rguenth at gcc dot gnu dot org]

-- 

rguenth at gcc dot gnu dot org changed:

           What    |Removed                     |Added
----------------------------------------------------------------------------
   Target Milestone|---                         |4.1.3


http://gcc.gnu.org/bugzilla/show_bug.cgi?id=35146

[Bug c++/35146] [4.2 regression] weird error in template function specialization

[jsm28 at gcc dot gnu dot org]

------- Comment #6 from jsm28 at gcc dot gnu dot org  2008-07-04 22:33 -------
Closing 4.1 branch.


-- 

jsm28 at gcc dot gnu dot org changed:

           What    |Removed                     |Added
----------------------------------------------------------------------------
            Summary|[4.1/4.2 regression] weird  |[4.2 regression] weird error
                   |error in template function  |in template function
                   |specialization              |specialization
   Target Milestone|4.1.3                       |4.2.5


http://gcc.gnu.org/bugzilla/show_bug.cgi?id=35146

[Bug c++/35146] [4.3/4.4/4.5 regression] weird error in template function specialization

[jsm28 at gcc dot gnu dot org]

------- Comment #7 from jsm28 at gcc dot gnu dot org  2009-03-31 15:10 -------
Closing 4.2 branch, original test fixed in 4.3.  However, there's a note here
about the test in comment#2 being valid and failing on trunk as of a year
ago, and that test does still fail, so marking as a 4.3/4.4/4.5 regression
instead of closing.

(I have not tried to check if the test in comment#2 is indeed valid, or if
it is indeed a regression from some previous version.)


-- 

jsm28 at gcc dot gnu dot org changed:

           What    |Removed                     |Added
----------------------------------------------------------------------------
      Known to fail|4.2.1 4.1.2 4.0.4 3.4.6     |4.2.1 4.1.2 4.0.4 3.4.6
                   |                            |4.2.5
      Known to work|4.3.0 3.3.6                 |3.3.6
   Last reconfirmed|2008-02-10 01:13:11         |2009-03-31 15:10:20
               date|                            |
            Summary|[4.2 regression] weird error|[4.3/4.4/4.5 regression]
                   |in template function        |weird error in template
                   |specialization              |function specialization
   Target Milestone|4.2.5                       |4.3.4


http://gcc.gnu.org/bugzilla/show_bug.cgi?id=35146

[Bug c++/35146] [4.3/4.4/4.5 regression] weird error in template function specialization

[jason at gcc dot gnu dot org]

------- Comment #8 from jason at gcc dot gnu dot org  2009-04-06 20:14 -------
The testcase in comment #2 is ill-formed;

template <>           void foo(S<double>,               S<double>*);

is not a specialization of any function template in scope.  But we're giving
very much the wrong error for that.  Testing a patch now.


-- 

jason at gcc dot gnu dot org changed:

           What    |Removed                     |Added
----------------------------------------------------------------------------
         AssignedTo|unassigned at gcc dot gnu   |jason at gcc dot gnu dot org
                   |dot org                     |
             Status|NEW                         |ASSIGNED
   Last reconfirmed|2009-03-31 15:10:20         |2009-04-06 20:14:50
               date|                            |


http://gcc.gnu.org/bugzilla/show_bug.cgi?id=35146

[Bug c++/35146] [4.3/4.4/4.5 regression] weird error in template function specialization

[jason at gcc dot gnu dot org]

------- Comment #9 from jason at gcc dot gnu dot org  2009-04-06 20:29 -------
...and GCC miscompiles the original testcase due to the same bug.


-- 

jason at gcc dot gnu dot org changed:

           What    |Removed                     |Added
----------------------------------------------------------------------------
           Keywords|                            |wrong-code


http://gcc.gnu.org/bugzilla/show_bug.cgi?id=35146

[Bug c++/35146] [4.3/4.4/4.5 regression] weird error in template function specialization

[jason at gcc dot gnu dot org]

------- Comment #10 from jason at gcc dot gnu dot org  2009-04-06 20:55 -------
Subject: Bug 35146

Author: jason
Date: Mon Apr  6 20:55:04 2009
New Revision: 145625

URL: http://gcc.gnu.org/viewcvs?root=gcc&view=rev&rev=145625
Log:
        PR c++/35146
        * pt.c (fn_type_unification): For DEDUCE_EXACT check that
        the deduced template arguments give us the parameter types
        we're looking for.

Added:
    trunk/gcc/testsuite/g++.dg/template/fnspec1.C
Modified:
    trunk/gcc/cp/ChangeLog
    trunk/gcc/cp/pt.c
    trunk/gcc/testsuite/ChangeLog


-- 


http://gcc.gnu.org/bugzilla/show_bug.cgi?id=35146

[Bug c++/35146] [4.3/4.4/4.5 regression] weird error in template function specialization

[jason at gcc dot gnu dot org]

------- Comment #11 from jason at gcc dot gnu dot org  2009-04-06 21:35 -------
Subject: Bug 35146

Author: jason
Date: Mon Apr  6 21:35:29 2009
New Revision: 145634

URL: http://gcc.gnu.org/viewcvs?root=gcc&view=rev&rev=145634
Log:
        PR c++/35146
        * pt.c (fn_type_unification): For DEDUCE_EXACT check that
        the deduced template arguments give us the parameter types
        we're looking for.

Added:
    branches/gcc-4_4-branch/gcc/testsuite/g++.dg/template/fnspec1.C
Modified:
    branches/gcc-4_4-branch/gcc/cp/ChangeLog
    branches/gcc-4_4-branch/gcc/cp/pt.c
    branches/gcc-4_4-branch/gcc/testsuite/ChangeLog


-- 


http://gcc.gnu.org/bugzilla/show_bug.cgi?id=35146

[Bug c++/35146] [4.3/4.4/4.5 regression] weird error in template function specialization

[jason at gcc dot gnu dot org]

------- Comment #12 from jason at gcc dot gnu dot org  2009-04-07 03:51 -------
Subject: Bug 35146

Author: jason
Date: Tue Apr  7 03:50:49 2009
New Revision: 145647

URL: http://gcc.gnu.org/viewcvs?root=gcc&view=rev&rev=145647
Log:
        PR c++/35146
        * pt.c (fn_type_unification): For DEDUCE_EXACT check that
        the deduced template arguments give us the parameter types
        we're looking for.

Added:
    branches/gcc-4_3-branch/gcc/testsuite/g++.dg/template/fnspec1.C
Modified:
    branches/gcc-4_3-branch/gcc/cp/ChangeLog
    branches/gcc-4_3-branch/gcc/cp/pt.c
    branches/gcc-4_3-branch/gcc/testsuite/ChangeLog


-- 


http://gcc.gnu.org/bugzilla/show_bug.cgi?id=35146

[Bug c++/35146] [4.3/4.4/4.5 regression] weird error in template function specialization

[jason at gcc dot gnu dot org]

------- Comment #13 from jason at gcc dot gnu dot org  2009-04-07 04:46 -------
Fixed.


-- 

jason at gcc dot gnu dot org changed:

           What    |Removed                     |Added
----------------------------------------------------------------------------
             Status|ASSIGNED                    |RESOLVED
         Resolution|                            |FIXED


http://gcc.gnu.org/bugzilla/show_bug.cgi?id=35146