[Bug c++/11506] New: SFINAE fails sometimes

[Bug c++/11506] New: SFINAE fails sometimes

[dave at boost-consulting dot com]

PLEASE REPLY TO gcc-bugzilla@gcc.gnu.org ONLY, *NOT* gcc-bugs@gcc.gnu.org.

http://gcc.gnu.org/bugzilla/show_bug.cgi?id=11506

           Summary: SFINAE fails sometimes
           Product: gcc
           Version: 3.2
            Status: UNCONFIRMED
          Severity: normal
          Priority: P2
         Component: c++
        AssignedTo: unassigned at gcc dot gnu dot org
        ReportedBy: dave at boost-consulting dot com
                CC: gcc-bugs at gcc dot gnu dot org
 GCC build triplet: i686-pc-linux
  GCC host triplet: i686-pc-cygwin
GCC target triplet: i686-pc-cygwin

Looking for a place in Bugzilla to upload my preprocessed file...

[Bug c++/11506] SFINAE fails sometimes

[dave at boost-consulting dot com]

PLEASE REPLY TO gcc-bugzilla@gcc.gnu.org ONLY, *NOT* gcc-bugs@gcc.gnu.org.

http://gcc.gnu.org/bugzilla/show_bug.cgi?id=11506



------- Additional Comments From dave at boost-consulting dot com  2003-07-12 03:25 -------
Created an attachment (id=4392)
 --> (http://gcc.gnu.org/bugzilla/attachment.cgi?id=4392&action=view)
preprocessed source

The enclosed ought to compile, but instead I get:

c:/boost/boost/iterator/transform_iterator.hpp: In instantiation of 
   `boost::transform_iterator<int (*)(int), int *, boost::use_default,
boost::use_default>':
c:/boost/boost/iterator/transform_iterator.hpp:127: no type named `type' in 
   `boost::detail::enable_if<boost::mpl::not_<boost::is_pointer<int (*)(int)>
>, char>'

Note that there's no instantiation backtrace!  Very strange.

Compile command:
    g++  -fmessage-length=0  -c -Wall -ftemplate-depth-100   -g -O0 -fno-inline
   -I "c:\boost"  -o
"bin\transform_iterator_test.test\gcc\debug\runtime-link-dynamic\transform_iterator_test.obj"
 "transform_iterator_test.cpp"

[Bug c++/11506] SFINAE fails sometimes

[dhazeghi at yahoo dot com]

PLEASE REPLY TO gcc-bugzilla@gcc.gnu.org ONLY, *NOT* gcc-bugs@gcc.gnu.org.

http://gcc.gnu.org/bugzilla/show_bug.cgi?id=11506


dhazeghi at yahoo dot com changed:

           What    |Removed                     |Added
----------------------------------------------------------------------------
   Attachment #4392|text/plain                  |application/octet-stream
          mime type|                            |

[Bug c++/11506] SFINAE fails sometimes

[dave at boost-consulting dot com]

PLEASE REPLY TO gcc-bugzilla@gcc.gnu.org ONLY, *NOT* gcc-bugs@gcc.gnu.org.

http://gcc.gnu.org/bugzilla/show_bug.cgi?id=11506


dave at boost-consulting dot com changed:

           What    |Removed                     |Added
----------------------------------------------------------------------------
   Attachment #4392|application/octet-stream    |application/zip
          mime type|                            |

[Bug c++/11506] SFINAE fails sometimes

[bangerth at dealii dot org]

PLEASE REPLY TO gcc-bugzilla@gcc.gnu.org ONLY, *NOT* gcc-bugs@gcc.gnu.org.

http://gcc.gnu.org/bugzilla/show_bug.cgi?id=11506



------- Additional Comments From bangerth at dealii dot org  2003-07-13 00:06 -------
I don't have the time to track it down right now, but just for information
for whoever wants to do it: there are multiple definitions of classes
enable_if in the code. Some take a bool as first parameter, but the one
we are talking here of takes a class. That's somewhat confusing, so just
as a warning ;-)
W.

[Bug c++/11506] SFINAE fails sometimes

[dave at boost-consulting dot com]

PLEASE REPLY TO gcc-bugzilla@gcc.gnu.org ONLY, *NOT* gcc-bugs@gcc.gnu.org.

http://gcc.gnu.org/bugzilla/show_bug.cgi?id=11506



------- Additional Comments From dave at boost-consulting dot com  2003-07-13 01:25 -------
Subject: Re:  SFINAE fails sometimes

"bangerth at dealii dot org" <gcc-bugzilla@gcc.gnu.org> writes:

> http://gcc.gnu.org/bugzilla/show_bug.cgi?id=11506

Hmm, that's really interesting.  I fixed that problem and the SFINAE
issue still remains.  So then I guess the *other* question is why GCC
wasn't reporting that error.

I tried to post the fix to bugzilla but the site seems to be
unresponsive at the moment.

[Bug c++/11506] SFINAE fails sometimes

[dave at boost-consulting dot com]

PLEASE REPLY TO gcc-bugzilla@gcc.gnu.org ONLY, *NOT* gcc-bugs@gcc.gnu.org.

http://gcc.gnu.org/bugzilla/show_bug.cgi?id=11506



------- Additional Comments From dave at boost-consulting dot com  2003-07-13 13:01 -------
Created an attachment (id=4400)
 --> (http://gcc.gnu.org/bugzilla/attachment.cgi?id=4400&action=view)
enable_if redefinition fixed in this version

the enable_if<class T, ... > version was moved from namespace boost::detail::
into namespace boost::iterator::

[Bug c++/11506] SFINAE fails sometimes

[pinskia at physics dot uc dot edu]

PLEASE REPLY TO gcc-bugzilla@gcc.gnu.org ONLY, *NOT* gcc-bugs@gcc.gnu.org.

http://gcc.gnu.org/bugzilla/show_bug.cgi?id=11506


pinskia at physics dot uc dot edu changed:

           What    |Removed                     |Added
----------------------------------------------------------------------------
             Status|UNCONFIRMED                 |WAITING


------- Additional Comments From pinskia at physics dot uc dot edu  2003-08-07 19:09 -------
I can confirm this in 3.3.1 but there is no way for me to test the mainline as the libstdc++ 
headers needs changes to support the mainline.
Can you test the mainline?
I think this is related to bug 7019.

[Bug c++/11506] SFINAE fails sometimes

[dhazeghi at yahoo dot com]

PLEASE REPLY TO gcc-bugzilla@gcc.gnu.org ONLY, *NOT* gcc-bugs@gcc.gnu.org.

http://gcc.gnu.org/bugzilla/show_bug.cgi?id=11506


dhazeghi at yahoo dot com changed:

           What    |Removed                     |Added
----------------------------------------------------------------------------
   Target Milestone|3.4                         |---


------- Additional Comments From dhazeghi at yahoo dot com  2003-08-24 19:16 -------
Dave, have you managed to check whether this works on current mainline yet? Thanks.

[Bug c++/11506] SFINAE fails sometimes

[dave at boost-consulting dot com]

PLEASE REPLY TO gcc-bugzilla@gcc.gnu.org ONLY, *NOT* gcc-bugs@gcc.gnu.org.

http://gcc.gnu.org/bugzilla/show_bug.cgi?id=11506



------- Additional Comments From dave at boost-consulting dot com  2003-08-24 20:20 -------
Subject: Re:  SFINAE fails sometimes

"dhazeghi at yahoo dot com" <gcc-bugzilla@gcc.gnu.org> writes:

> PLEASE REPLY TO gcc-bugzilla@gcc.gnu.org ONLY, *NOT* gcc-bugs@gcc.gnu.org.
>
> http://gcc.gnu.org/bugzilla/show_bug.cgi?id=11506
>
>
> dhazeghi at yahoo dot com changed:
>
>            What    |Removed                     |Added
> ----------------------------------------------------------------------------
>    Target Milestone|3.4                         |---
>
>
> ------- Additional Comments From dhazeghi at yahoo dot com  2003-08-24 19:16 -------
> Dave, have you managed to check whether this works on current mainline yet? Thanks.

No, and I didn't have any plans to.  I don't have the CVS mainline
anywhere.  I submitted a reproducible test case, didn't I?  Can't you
just try it yourself if you want to know whether it's fixed?  If I
*were* going to check it, that's all *I'd* do...

[Bug c++/11506] SFINAE fails sometimes

[pinskia at gcc dot gnu dot org]

PLEASE REPLY TO gcc-bugzilla@gcc.gnu.org ONLY, *NOT* gcc-bugs@gcc.gnu.org.

http://gcc.gnu.org/bugzilla/show_bug.cgi?id=11506



------- Additional Comments From pinskia at gcc dot gnu dot org  2003-08-24 20:24 -------
The reason why I asked is that we cannot check with your testcase as the compiler has become 
more standard's compliant so libstdc++ has to be changed for those changes and since your 
preproccessed source file contains libstdc++ headers included we cannot test your testcase.

[Bug c++/11506] SFINAE fails sometimes

[dhazeghi at yahoo dot com]

PLEASE REPLY TO gcc-bugzilla@gcc.gnu.org ONLY, *NOT* gcc-bugs@gcc.gnu.org.

http://gcc.gnu.org/bugzilla/show_bug.cgi?id=11506



------- Additional Comments From dhazeghi at yahoo dot com  2003-08-24 21:50 -------
What version of boost is your testcase from? Using current boost cvs, test seems to build fine with 
gcc 3.0.X-mainline. Thanks.

[Bug c++/11506] SFINAE fails sometimes

[dave at boost-consulting dot com]

PLEASE REPLY TO gcc-bugzilla@gcc.gnu.org ONLY, *NOT* gcc-bugs@gcc.gnu.org.

http://gcc.gnu.org/bugzilla/show_bug.cgi?id=11506



------- Additional Comments From dave at boost-consulting dot com  2003-08-24 23:23 -------
Subject: Re:  SFINAE fails sometimes

"pinskia at gcc dot gnu dot org" <gcc-bugzilla@gcc.gnu.org> writes:

> PLEASE REPLY TO gcc-bugzilla@gcc.gnu.org ONLY, *NOT* gcc-bugs@gcc.gnu.org.
>
> http://gcc.gnu.org/bugzilla/show_bug.cgi?id=11506
>
>
>
> ------- Additional Comments From pinskia at gcc dot gnu dot org  2003-08-24 20:24 -------
> The reason why I asked is that we cannot check with your testcase as the compiler has become 
> more standard's compliant so libstdc++ has to be changed for those changes and since your 
> preproccessed source file contains libstdc++ headers included we cannot test your testcase.

I'm not sure exactly what we did to produce the original bug report.
You *may* be able to reproduce this bug by:

1. Getting the Boost CVS state (see
   http://www.boost.org/more/download.html#CVS)

2. going to the libs/iterator/test subdirectory

3. Issuing the compile command shown in the bug report.  Make sure
   the Boost root directory is in the #include path.

Note that
http://boost.sourceforge.net/regression-logs/cs-Linux/developer_result_page.html#iterator
still shows one compiler bug for GCC-3.4/CVS on this library (a test
which is supposed to fail to compile does not).  You can find many
others by scanning the gcc-3.4 column at
http://boost.sourceforge.net/regression-logs/cs-Linux/developer_summary_page.html.

HTH,

[Bug c++/11506] SFINAE fails sometimes

[dave at boost-consulting dot com]

PLEASE REPLY TO gcc-bugzilla@gcc.gnu.org ONLY, *NOT* gcc-bugs@gcc.gnu.org.

http://gcc.gnu.org/bugzilla/show_bug.cgi?id=11506



------- Additional Comments From dave at boost-consulting dot com  2003-08-25 00:29 -------
Subject: Re:  SFINAE fails sometimes

"dhazeghi at yahoo dot com" <gcc-bugzilla@gcc.gnu.org> writes:

> PLEASE REPLY TO gcc-bugzilla@gcc.gnu.org ONLY, *NOT* gcc-bugs@gcc.gnu.org.
>
> http://gcc.gnu.org/bugzilla/show_bug.cgi?id=11506
>
>
>
> ------- Additional Comments From dhazeghi at yahoo dot com  2003-08-24 21:50 -------
> What version of boost is your testcase from? 

Possibly the CVS state from the day I reported it, though I may have
checked in a workaround so I can't guarantee that the non-working code
was ever in the repository.  One other thing you could try is to prune
out all the preprocessed content from the old std library and insert
the requisite #include directives.  That seems like the most reliable
approach, since the bug report doesn't really have anything to do
with the std library.

> Using current boost cvs, test seems to build fine with 
> gcc 3.0.X-mainline. Thanks.

[Bug c++/11506] SFINAE fails sometimes

[dhazeghi at yahoo dot com]

PLEASE REPLY TO gcc-bugzilla@gcc.gnu.org ONLY, *NOT* gcc-bugs@gcc.gnu.org.

http://gcc.gnu.org/bugzilla/show_bug.cgi?id=11506



------- Additional Comments From dhazeghi at yahoo dot com  2003-08-25 00:34 -------
Here's what I've done:

grabbed boost cvs.
cd boost
g++ -fmessage-length=0  -c -Wall -ftemplate-depth-100   -g -O0 -fno-inline -I. libs/iterator/
test/transform_iterator_test.cpp

Test compiles with both g++ 3.3 branch and mainline. OTOH, the preprocessed source you sent 
does exhibits the error you report (on 3.3, doesn't compile on 3.4). I'll look at the other boost links 
you sent in a minute, but I must admit I'm puzzled by these results.

[Bug c++/11506] SFINAE fails sometimes

[dave at boost-consulting dot com]

PLEASE REPLY TO gcc-bugzilla@gcc.gnu.org ONLY, *NOT* gcc-bugs@gcc.gnu.org.

http://gcc.gnu.org/bugzilla/show_bug.cgi?id=11506



------- Additional Comments From dave at boost-consulting dot com  2003-08-25 00:45 -------
Subject: Re:  SFINAE fails sometimes

"dhazeghi at yahoo dot com" <gcc-bugzilla@gcc.gnu.org> writes:

> PLEASE REPLY TO gcc-bugzilla@gcc.gnu.org ONLY, *NOT* gcc-bugs@gcc.gnu.org.
>
> http://gcc.gnu.org/bugzilla/show_bug.cgi?id=11506
>
>
>
> ------- Additional Comments From dhazeghi at yahoo dot com  2003-08-25 00:34 -------
> Here's what I've done:
>
> grabbed boost cvs.
> cd boost
> g++ -fmessage-length=0  -c -Wall -ftemplate-depth-100   -g -O0 -fno-inline -I. libs/iterator/
> test/transform_iterator_test.cpp
>
> Test compiles with both g++ 3.3 branch and mainline. OTOH, the preprocessed source you sent 
> does exhibits the error you report (on 3.3, doesn't compile on 3.4). I'll look at the other boost links 
> you sent in a minute, but I must admit I'm puzzled by these results.

Another way to get a clearer picture is to preprocess the Boost CVS
with g++ 3.3 and diff it with the preprocessed source I sent to see
what changed.

[Bug c++/11506] SFINAE fails sometimes

[dhazeghi at yahoo dot com]

PLEASE REPLY TO gcc-bugzilla@gcc.gnu.org ONLY, *NOT* gcc-bugs@gcc.gnu.org.

http://gcc.gnu.org/bugzilla/show_bug.cgi?id=11506



------- Additional Comments From dhazeghi at yahoo dot com  2003-08-25 15:51 -------
Plodding away. Cannot reproduce with boost cvs from 2003-07-07. I like your diff suggestion, 
though unfortunately it comes to about 3 megs in this case :-) Will try to strip out the includes 
from the preprocessed source...

[Bug c++/11506] SFINAE fails sometimes

[pinskia at gcc dot gnu dot org]

PLEASE REPLY TO gcc-bugzilla@gcc.gnu.org ONLY, *NOT* gcc-bugs@gcc.gnu.org.

http://gcc.gnu.org/bugzilla/show_bug.cgi?id=11506



------- Additional Comments From pinskia at gcc dot gnu dot org  2003-10-23 07:06 -------
Most likely the same bug as PR 12672  which has a small testcase (and looks like icc 
gets one wrong also).

[Bug c++/11506] SFINAE fails sometimes

[pinskia at gcc dot gnu dot org]

------- Additional Comments From pinskia at gcc dot gnu dot org  2003-11-26 16:19 -------
Closing as there really is no way to reduce this at all, and there is another PR which is 
closly related (might even be dup of that one) and there has been no feedback in 3 
months.

-- 
           What    |Removed                     |Added
----------------------------------------------------------------------------
             Status|WAITING                     |RESOLVED
         Resolution|                            |INVALID


http://gcc.gnu.org/bugzilla/show_bug.cgi?id=11506