Re: [flint-devel] FLINT 1.4 Released!

[Bill H. <goo...@go...>]

Hi Burcin,

Thanks for the patch. I'll still review all the code going into FLINT.
But it is then easier for you to commit directly.

FLINT is currently like a heart patient, laid out on the table in
pieces in the middle of surgery. But in a week or so I hope to have it
back together and sewn up for the release. Most of the changes only
affect FLINT 2.0 code, which is not being released yet. But it affects
the F_mpz layer, which I have released.

The way you've done the test function is fine by the way. It is what I
would do.

I guess your patch is against FLINT 1.4 rather than trunk, is that correct?

Bill.

2009/8/8 Burcin Erocal <bu...@er...>:
> Hi Bill,
>
> On Mon, 27 Jul 2009 21:32:05 +0100
> Bill Hart <goo...@go...> wrote:
>
>> Great that everything works. The easiest way to write a test function
>> is to cut and paste an existing one. It shouldn't be too hard to
>> convert random poly to zmod_polys then use the evaluate or composition
>> functions there, then compare the results. In some cases it is hard to
>> do randomised testing because it is difficult to find a simpler way of
>> writing the function so that you have a trusted implementation to
>> compare against. In such cases I tend to look for some algebraic way
>> of testing the result, e.g. Test addition by subtracting the addend
>> back off and see if you get the same thing back. I rarely check
>> individual cases but try to implement the randomised testing in such a
>> way as to pick up corner cases.
>
> After another long delay, I managed to write the test functions and the
> documentation. See the attached patch.
>
> For testing, I just used the already existing functions in the library.
> For example, in order to test fmpz_poly_evaluate_mod I call
> fmpz_poly_evaluate then reduce the result.
>
> I would be happy to revise the patch if you have any suggestions.
>
>> I don't actually have HG, but I can accept patches in git, svn or
>> standard patch format.
>>
>> If it would help, I can easily give you svn access, if you have a
>> sourceforge account.
>
> I like to know that my patches are being reviewed. Committing directly
> to SVN would eliminate that step. Though if you think this process is
> taking up too much time, and the possibility that I mess up the
> repository is low enough, my sf user name is burcin.
>
> Thanks.
>
> Burcin
> ------------------------------------------------------------------------------
> Let Crystal Reports handle the reporting - Free Crystal Reports 2008 30-Day
> trial. Simplify your report design, integration and deployment - and focus on
> what you do best, core application coding. Discover what's new with
> Crystal Reports now.  http://p.sf.net/sfu/bobj-july
> _______________________________________________
> flint-devel mailing list
> fli...@li...
> https://lists.sourceforge.net/lists/listinfo/flint-devel
>
>