Add const version of Integer::operator=()

[arrowd]

When building sage with latest releases of linbox and givaro I get following error:

[sagelib-9.1] In file included from build/cythonized/sage/matrix/matrix_modn_sparse.cpp:4089:
[sagelib-9.1] In file included from /usr/local/include/linbox/solutions/charpoly.h:34:
[sagelib-9.1] In file included from /usr/local/include/linbox/algorithms/bbcharpoly.h:46:
[sagelib-9.1] In file included from /usr/local/include/linbox/solutions/det.h:608:
[sagelib-9.1] /usr/local/include/linbox/algorithms/det-rational.h:95:39: error: no viable overloaded '='
[sagelib-9.1]                 void setDiv (const Integer& d) {div = d;}
[sagelib-9.1]                                                 ~~~ ^ ~
[sagelib-9.1] /usr/local/include/gmp++/gmp++_int.h:236:34: note: candidate function not viable: 'this' argument has type 'const LinBox::Integer' (aka 'const Givaro::Integer'), but method is not marked const
[sagelib-9.1]         giv_all_inlined Integer& operator = (const Integer& n);
[sagelib-9.1]                                  ^

This PR fixes the error, but I'm not sure it is a correct solution.

[jgdumas]

Hum, I do not think it is a good idea to const cast in Givaro.
The problem comes from det-rational.h
There is an inconsistency between mul/div being "const Integer&", and having modifier method setDiv/setMul.

I would suggest to LinBox to instead make a copy and declare mul/div to be just "Integer".

[arrowd]

Sorry, I didn't get it. Should I provide a patch, or you'll fix this yourself?

[jgdumas]

Well I cannot reproduce the issue you mention. So can you give more details on compilers, flags, etc. ?
Otherwise can you try what I propose on your install, if it works I can test it further and propose a PR ...

[arrowd]

I'm compiling on FreeBSD 13-CURRENT, which uses Clang 10 as compiler. And that's the culprit I suspect, because the error doesn't appear on FreeBSD 12.1, which uses older clang.

Can you try building givaro with Clang 10?

[jgdumas]

OK I could reproduce the issue. Thank you.
This is now solved in LinBox PR256 : linbox-team/linbox#256