Can't build with intel 9.0

Mike Gibson

12 Aug 2005 12 Aug '05

10:01 p.m.

I was having problems getting boost 1.32 to build with the intel 9.0 compiler and saw that 1.33 was out and listed support for intel 9.0. I can't get it to build though. I get an error compiling greg_month.cpp in lexical_cast.hpp(156). I'm building using the following command: C:\src\boost\boost_1_33_0>bjam -sTOOLS=intel-win32 -sINTEL_PATH="C:\Program Files\Intel\Compiler\C++\9.0\IA32" install Here is the error: C:\src\boost\boost_1_33_0/boost/lexical_cast.hpp(156): error: no operator ">>" matches these operands operand types are: std::basic_stringstream >> boost::date_time::date_names_put>>::string_type stream >> output && ^

Show replies by date

Jeff Garland

13 Aug 13 Aug

3:21 a.m.

On Fri, 12 Aug 2005 16:01:33 -0600, Mike Gibson wrote

...

I was having problems getting boost 1.32 to build with the intel 9.0 compiler and saw that 1.33 was out and listed support for intel 9.0. I can't get it to build though. I get an error compiling greg_month.cpp in lexical_cast.hpp(156).

I'm building using the following command:

C:\src\boost\boost_1_33_0>bjam -sTOOLS=intel-win32 -sINTEL_PATH="C:\Program Files\Intel\Compiler\C++\9.0\IA32" install

Here is the error:

C:\src\boost\boost_1_33_0/boost/lexical_cast.hpp(156): error: no operator ">>" matches these operands operand types are: std::basic_stringstream >> boost::date_time::date_names_put>>::string_type stream >> output && ^

This is somewhat suprising for 1_33. Seems like you are getting the old I/O code. Are you defining USE_DATE_TIME_PRE_1_33_FACET_IO ? You can see how it gets set in boost/date_time/compiler_config.hpp -- it really shouldn't be enabled for Intel. BTW, the regression tests for date-time on Intel 9 are passing 100% so it would seem this is something environmental.... Jeff

John Maddock

10:11 a.m.

...

This is somewhat suprising for 1_33. Seems like you are getting the old I/O code. Are you defining USE_DATE_TIME_PRE_1_33_FACET_IO ? You can see how it gets set in boost/date_time/compiler_config.hpp -- it really shouldn't be enabled for Intel. BTW, the regression tests for date-time on Intel 9 are passing 100% so it would seem this is something environmental....

Another possible "gotcha", is if native wchar_t support is not enabled: we don't currently test that, but if I remember correctly it breaks all kinds of things in lexical_cast when you do that. John.

Jeff Garland

2:49 p.m.

On Sat, 13 Aug 2005 11:11:39 +0100, John Maddock wrote

...

...
This is somewhat suprising for 1_33. Seems like you are getting the old I/O code. Are you defining USE_DATE_TIME_PRE_1_33_FACET_IO ? You can see how it gets set in boost/date_time/compiler_config.hpp -- it really shouldn't be enabled for Intel. BTW, the regression tests for date-time on Intel 9 are passing 100% so it would seem this is something environmental....

Another possible "gotcha", is if native wchar_t support is not enabled: we don't currently test that, but if I remember correctly it breaks all kinds of things in lexical_cast when you do that.

Good point John -- wchar_t support may indeed have an impact. Based on the regression results this would need to be a user specified option -- but the problem report didn't indicate this :-( Jeff

Mike Gibson

15 Aug 15 Aug

4:47 p.m.

Seems that if I build with INTEL_BASE_MSVC_TOOLSET=vc-7_1 things seem to work. Now the only thing that doesn't build is the debug version of date-time. I'm working on that one. Mike Gibson megibson@gmail.com On 8/13/05, Jeff Garland wrote:

...

On Sat, 13 Aug 2005 11:11:39 +0100, John Maddock wrote

...
...
This is somewhat suprising for 1_33. Seems like you are getting the old I/O code. Are you defining USE_DATE_TIME_PRE_1_33_FACET_IO ? You can see how it gets set in boost/date_time/compiler_config.hpp -- it really shouldn't be enabled for Intel. BTW, the regression tests for date-time on Intel 9 are passing 100% so it would seem this is something environmental....

Another possible "gotcha", is if native wchar_t support is not enabled: we don't currently test that, but if I remember correctly it breaks all kinds of things in lexical_cast when you do that.

Good point John -- wchar_t support may indeed have an impact. Based on the regression results this would need to be a user specified option -- but the problem report didn't indicate this :-(

Jeff

_______________________________________________ Boost-users mailing list Boost-users@lists.boost.org http://lists.boost.org/mailman/listinfo.cgi/boost-users

Mike Gibson

5:47 p.m.

On 8/15/05, Mike Gibson wrote:

...

Seems that if I build with INTEL_BASE_MSVC_TOOLSET=vc-7_1 things seem to work. Now the only thing that doesn't build is the debug version of date-time. I'm working on that one.

Mike Gibson megibson@gmail.com

This problem is fixed by rebuilding. The take home message is that if you're using Intel, then you should make sure that INTEL_BASE_MSVC_TOOLSET is set correctly. I also had to make sure that INTEL_PATH was set correctly. The current code for figureing it out doesn't seem to work too well. I would like to add this information to the documentation for the next release, but haven't been able to find the process for contributions to documentation. What is it? Mike Gibson megibson@gmail.com

Jeff Garland

20 Aug 20 Aug

4:26 p.m.

On Mon, 15 Aug 2005 11:47:32 -0600, Mike Gibson wrote

...

On 8/15/05, Mike Gibson wrote:

...
Seems that if I build with INTEL_BASE_MSVC_TOOLSET=vc-7_1 things seem to work. Now the only thing that doesn't build is the debug version of date-time. I'm working on that one.

Mike Gibson megibson@gmail.com

This problem is fixed by rebuilding.

The take home message is that if you're using Intel, then you should make sure that INTEL_BASE_MSVC_TOOLSET is set correctly. I also had to make sure that INTEL_PATH was set correctly. The current code for figureing it out doesn't seem to work too well.

I would like to add this information to the documentation for the next release, but haven't been able to find the process for contributions to documentation. What is it?

Probably the best way is to create a patch for the file/files and post it on the developer list. I notice we don't have specific guidance on how to do this on the web-page. If this is boost.build thing you might want to post is on the boost.build list. See http://www.boost.org/more/mailing_lists.htm#projects HTH, Jeff

7027

Age (days ago)

7035

Last active (days ago)

List overview

Download

6 comments

3 participants

participants (3)

Jeff Garland
John Maddock
Mike Gibson