LinBox build failure on OSX

[Mark Giesbrecht]

Hi,

I’m getting a build error for LinBox on OSX, pulling the latest src from GitHub (with very recently pulled and built Givaro and FFLAS-FFPack)

First error is: 

In file included from ../../linbox/algorithms/block-massey-domain.h:42:

../../linbox/algorithms/polynomial-matrix/order-basis.h:617:42: error: ambiguous conversion for functional-style cast from 'size_t' (aka 'unsigned long') to 'LinBox::integer' (aka 'Givaro::Integer')

                        size_t log_order=integer(order).bitsize();

Only happens on OSX; linux builds are fine.

Any thoughts?

Thanks.

Mark

__

Dr. Mark Giesbrecht, Professor 

David R. Cheriton School of Computer Science, University of Waterloo, Canada

Phone:+1.519.888.4080   Email:m...@uwaterloo.ca  URL:https://cs.uwaterloo.ca/~mwg

[Clement Pernet]

Hi,
This is a classic problem on OSX, where size_t are not silently casted to uint64_t.
I pushed an explicit cast. Let me know if it fixes it.
I'm suprised nobody caught this one earlier. Pascal has been fixing numerous such casts issues for
OSX lately.
Best.
Clément

Le 26/09/2018 à 18:22, Mark Giesbrecht a écrit :
> Hi,
>
> I’m getting a build error for LinBox on OSX, pulling the latest src from GitHub (with very recently
> pulled and built Givaro and FFLAS-FFPack)
>
> First error is:
>
> In file included from ../../linbox/algorithms/block-massey-domain.h:42:

> *../../linbox/algorithms/polynomial-matrix/order-basis.h:617:42: **error: **ambiguous conversion for
> functional-style cast from 'size_t' (aka 'unsigned long') to 'LinBox::integer' (aka 'Givaro::Integer')*

>                         size_t log_order=integer(order).bitsize();
>
> Only happens on OSX; linux builds are fine.
>
> Any thoughts?
>
> Thanks.
>
> Mark
>
> __
> Dr. Mark Giesbrecht, Professor
> David R. Cheriton School of Computer Science, University of Waterloo, Canada

> Phone:+1.519.888.4080 <tel:+15198884080>   Email:m...@uwaterloo.ca <mailto:m...@uwaterloo.ca>
> URL:https://cs.uwaterloo.ca/~mwg <https://cs.uwaterloo.ca/%7Emwg>
>
> --
> You received this message because you are subscribed to the Google Groups "linbox-use" group.
> To unsubscribe from this group and stop receiving emails from it, send an email to
> linbox-use+...@googlegroups.com <mailto:linbox-use+...@googlegroups.com>.
> To post to this group, send email to linbo...@googlegroups.com <mailto:linbo...@googlegroups.com>.
> Visit this group at https://groups.google.com/group/linbox-use.
> For more options, visit https://groups.google.com/d/optout.

[Clement Pernet]

Apologies, I forgot to push. It's done now.
Clément

Le 26/09/2018 à 23:52, Mark Giesbrecht a écrit :
> Hi Clément,
>
> Did you push the fixed code with the explicit cast to the master branch?  I’m not seeing it.

>
> Thanks.
>
> Mark
>
> __
> Dr. Mark Giesbrecht, Professor
> David R. Cheriton School of Computer Science, University of Waterloo, Canada
> Phone:+1.519.888.4080 <tel:+15198884080>   Email:m...@uwaterloo.ca <mailto:m...@uwaterloo.ca>
> URL:https://cs.uwaterloo.ca/~mwg <https://cs.uwaterloo.ca/%7Emwg>
>
>

>> On Sep 26, 2018, at 4:05 PM, Clement Pernet <clement...@gmail.com

>>> <mailto:linbo...@googlegroups.com> <mailto:linbo...@googlegroups.com>.

[Pascal Giorgi]

Hi,

ok I missed this one since it is in our online order basis algorithm and this one was not suitably tested.

Your fix is good and I modified the test suite for order basis to provide a larger class of check.

Unfortunately, our online order basis (OPMBasis) is not working properly.

I removed it from the test suite and I opened an issue for this.

Mark, if you want to use order basis code please prefer to use PMBasis rather than OPMbasis.

Best.

Pascal.