QL 0.3.6 with gcc 2.95?

[Michael Dirkmann]

Ok, just a followup to my first mail from Firday.
QL 0.3.7, when trying to compile with gcc 2.95, does not even pass
configure.

QL 0.3.6 in fact does configure properly. It seems to work fine until it
gets to linearinterpolation.hpp (or cubicspline.hpp) where gcc leaves me
with an

../../ql/Math/linearinterpolation.hpp:36: Internal compiler error
../../ql/Math/cubicspline.hpp:90: Internal compiler error

I found a refence to similar(?) problem with VS 6.0 on the mailing list.
  But not real solution applicable to my problem.

I know either 0.3.6 and gcc 2.95 are more or less obsolete but if
somebody had a solution available I'd be greatful.

TIA

   Michael

[Jens Thiel-2]

Michael,

have you considered compiling your 0.3.7 QL implementations to a shared
library (.DLL/.so) that exports its functionality to the main application?

Gruss,

Jens.

--
Jens Thiel
http://ManagedXLL.net/
[phone]

> -----Original Message-----
> From: [hidden email]
> [mailto:[hidden email]] On Behalf
> Of [hidden email]
> Sent: Sunday, November 14, 2004 1:09 PM
> To: [hidden email]
> Subject: [Quantlib-users] QL 0.3.6 with gcc 2.95?
>
>
> Ok, just a followup to my first mail from Firday.
> QL 0.3.7, when trying to compile with gcc 2.95, does not even pass
> configure.
>
> QL 0.3.6 in fact does configure properly. It seems to work
> fine until it
> gets to linearinterpolation.hpp (or cubicspline.hpp) where
> gcc leaves me
> with an
>
> ../../ql/Math/linearinterpolation.hpp:36: Internal compiler error
> ../../ql/Math/cubicspline.hpp:90: Internal compiler error
>
> I found a refence to similar(?) problem with VS 6.0 on the
> mailing list.
>   But not real solution applicable to my problem.
>
> I know either 0.3.6 and gcc 2.95 are more or less obsolete but if
> somebody had a solution available I'd be greatful.
>
> TIA
>
>    Michael
>
>
>
>
> -------------------------------------------------------
> This SF.Net email is sponsored by: InterSystems CACHE
> FREE OODBMS DOWNLOAD - A multidimensional database that
> combines robust object and relational technologies, making it
> a perfect match for Java, C++,COM, XML, ODBC and JDBC.

www.intersystems.com/match8 _______________________________________________
Quantlib-users mailing list [hidden email]
https://lists.sourceforge.net/lists/listinfo/quantlib-users

[Michael Dirkmann]

Sure, that is exactly what I am trying to do. But object formats between
2.95 and 3.x have change and hence are incompatible. That is the reason
for my struggling.

Regards

   Michael

Jens Thiel wrote:

> Michael,
>
> have you considered compiling your 0.3.7 QL implementations to a shared
> library (.DLL/.so) that exports its functionality to the main application?
>
> Gruss,
>
> Jens.
>
> --
> Jens Thiel
> http://ManagedXLL.net/
> [phone]
>
>
>
>
>>-----Original Message-----
>>From: [hidden email]
>>[mailto:[hidden email]] On Behalf
>>Of [hidden email]
>>Sent: Sunday, November 14, 2004 1:09 PM
>>To: [hidden email]
>>Subject: [Quantlib-users] QL 0.3.6 with gcc 2.95?
>>
>>
>>Ok, just a followup to my first mail from Firday.
>>QL 0.3.7, when trying to compile with gcc 2.95, does not even pass
>>configure.
>>
>>QL 0.3.6 in fact does configure properly. It seems to work
>>fine until it
>>gets to linearinterpolation.hpp (or cubicspline.hpp) where
>>gcc leaves me
>>with an
>>
>>../../ql/Math/linearinterpolation.hpp:36: Internal compiler error
>>../../ql/Math/cubicspline.hpp:90: Internal compiler error
>>
>>I found a refence to similar(?) problem with VS 6.0 on the
>>mailing list.
>>  But not real solution applicable to my problem.
>>
>>I know either 0.3.6 and gcc 2.95 are more or less obsolete but if
>>somebody had a solution available I'd be greatful.
>>
>>TIA
>>
>>   Michael
>>
>>
>>
>>
>>-------------------------------------------------------
>>This SF.Net email is sponsored by: InterSystems CACHE
>>FREE OODBMS DOWNLOAD - A multidimensional database that
>>combines robust object and relational technologies, making it
>>a perfect match for Java, C++,COM, XML, ODBC and JDBC.
>
> www.intersystems.com/match8 _______________________________________________
> Quantlib-users mailing list [hidden email]
> https://lists.sourceforge.net/lists/listinfo/quantlib-users
>
>
>

[Luigi Ballabio-2]

Michael,

On 2004.11.14 13:09, [hidden email] wrote:

> Ok, just a followup to my first mail from Firday.
> QL 0.3.7, when trying to compile with gcc 2.95, does not even pass  
> configure.
>
> QL 0.3.6 in fact does configure properly. It seems to work fine until it  
> gets to linearinterpolation.hpp (or cubicspline.hpp) where gcc leaves me  
> with an
>
> ../../ql/Math/linearinterpolation.hpp:36: Internal compiler error
> ../../ql/Math/cubicspline.hpp:90: Internal compiler error
>
> I know either 0.3.6 and gcc 2.95 are more or less obsolete but if  
> somebody had a solution available I'd be greatful.

I'll try (emphasis on "try") and fix this for next release.

Nando: it will involve copying the data inside the interpolation object  
instead of just storing iterators. I'll do this for all compilers as it  
would be nightmarish to keep two widely different implementations. On the  
other hand, it might simplify the code :)
Any objection?

Later,
        Luigi

[Michael Dirkmann]

Thanks Luigi.

meanwhile I found a workaround:
gcc-2.95 compiles the code when taking the nested (Impl) classes of the
derived XyzInterpolation. It is a fudge but works for me.

Regards

   Michael


Luigi Ballabio wrote:

>
> Michael,
>
> On 2004.11.14 13:09, [hidden email] wrote:
>
>> Ok, just a followup to my first mail from Firday.
>> QL 0.3.7, when trying to compile with gcc 2.95, does not even pass  
>> configure.
>>
>> QL 0.3.6 in fact does configure properly. It seems to work fine until
>> it  gets to linearinterpolation.hpp (or cubicspline.hpp) where gcc
>> leaves me  with an
>>
>> ../../ql/Math/linearinterpolation.hpp:36: Internal compiler error
>> ../../ql/Math/cubicspline.hpp:90: Internal compiler error
>>
>> I know either 0.3.6 and gcc 2.95 are more or less obsolete but if  
>> somebody had a solution available I'd be greatful.
>
>
> I'll try (emphasis on "try") and fix this for next release.
>
> Nando: it will involve copying the data inside the interpolation object  
> instead of just storing iterators. I'll do this for all compilers as it  
> would be nightmarish to keep two widely different implementations. On
> the  other hand, it might simplify the code :)
> Any objection?
>
> Later,
>     Luigi

[Luigi Ballabio-2]

On 2004.11.15 21:19, [hidden email] wrote:
>
> meanwhile I found a workaround:
> gcc-2.95 compiles the code when taking the nested (Impl) classes of the  
> derived XyzInterpolation. It is a fudge but works for me.

Michael,
        I don't quite get what 'taking the nested classes' means. You mean  
it compiles if you declare them at namespace scope, i.e., outside the  
derived interpolation classes? Does this also apply to  
Interpolation::templateImpl? Can you send me a patch, or the modified  
files?

Also, did you have issues with 2-D interpolation as well, or did that work?

Later,
        Luigi