Re: [flint-devel] FLINT 1.1 beta

[Michael A. <mab...@go...>]

On Fri, Dec 19, 2008 at 7:53 AM, Bill Hart <goo...@go...> wrote:

[I am not sure if I should trim the CC list or not, so feel free to
complain if you want off]

> Hi All,

Hi All

> I just released FLINT 1.1.0 beta. As the FLINT website is currently
> down due to the upgrade of the machine (sage.math) which it is hosted
> on, I give the following link:
>
> http://sage.math.washington.edu/home/wbhart/webpage/flint-1.1.0.tar.gz
>
> The documentation is here:
>
> http://sage.math.washington.edu/home/wbhart/webpage/flint-1.1.0.pdf
>
> I would be very pleased if any of you could run the test code on your
> local machines and report any bugs. Running the tests is simply "make
> check" in this version of FLINT. Please note that each module reports
> separately - the final "all tests passed" only refers to the final
> module tested. You have to scroll back through to check that all
> modules passed all their tests.

Ok, will do.

> Note you must link against GMP/MPIR and zn_poly
> (http://www.math.harvard.edu/~dmharvey/zn_poly/) for this release. To
> edit the lib and include directories for these, edit the file
> "flint_env" and do "source flint_env". Full instructions are to be
> found in the first few pages of the documentation if you get stuck. If
> you cannot build zn_poly on your machine (considered very unlikely)
> please report to David Harvey (dmh...@ma...) and refer to
> the FLINT documentation for how to build without zn_poly. It is not
> possible to build FLINT without either GMP or MPIR.
>
> All the documentation is complete. Tests pass on both the old and new
> 64 bit sage.math, on 32 bit Cygwin and on a 32 bit linux. All
> functions that should alias, do, and there is full test code for this.
> Anyone wishing to use/wrap the new functionality in FLINT can use this
> release. I've spent the past month testing and documenting it. There
> are very few changes to the interface for FLINT 1.1.0.
>
> Once build tests have passed and any issues resolved, I will valgrind
> everything (I have already done a lot of this) and the beta status
> will be lifted.

Have you ever fixed the deallocation issue of your custom allocation
stack at exit? We have been seeing it hang around in Sage at exit.

> I will post an announcement on flint-devel and sage-devel when that
> time comes, including a very long list of all the new features of
> FLINT 1.1. This is a MAJOR new feature release, thus the beta release
> before final!!
>
> If any of you wish to be on the FLINT development list and are not
> already on it, you can register with sourceforge:
>
> http://sourceforge.net/account/registration/
>
>  and subscribe at the following link:
>
> http://sourceforge.net/mail/?group_id=182716
>
> The traffic is very low except during English summer when
> undergraduates work on the project when the traffic is medium. I would
> value your contributions to the list.

Ok, I really dislike the sf list management and especially their
management of the archives.

> If there is sufficient demand for a google groups for FLINT
> development, I will create one and switch.

Yes, I would be very happy to see you switch.

> Subsequent feature releases will now be issued every 3 months, not
> every year. FLINT 1.2, to be released in about 3 months, will contain
> numerous speed ups and a new functions that got cut from 1.1. FLINT
> 2.0 will be issued in about six to nine months and will include a new
> polynomial module, and Andy Novocin's new factoring
> technique/heuristics for ZZ[x]. About 1/3 of the code for FLINT 2.0 is
> already written, but not included in this release.
>
> Many thanks to Andy Novocin for help with final documentation and
> alias testing and thank you to the many contributors to this release
> which are listed on the webpage:
>
> http://sage.math.washington.edu/home/wbhart/webpage (normally
> http://www.flintlib.org/)
>
> The highlights of this release are fast polynomial GCD for ZZ[x],
> (using heuristic GCD and modular half-gcd), which is now almost
> uniformly faster than Magma:

Nice.

> http://sage.math.washington.edu/home/wbhart/flint-trunk/graphing/gcd19.png
> (ignore the heading which is incorrect),
>
> polynomial composition (due to Andy Novocin) and evaluation (which are
> asymptotically faster than Magma) and factoring of polynomials in
> Z/pZ[x] for word sized primes (much faster than Magma, due to Richard
> Howell-Peak), new primality testing and factoring functions for word
> sized integers (due to Peter Shrimpton) and a much faster quadratic
> sieve for factoring integers.

:) - we need to hook up the new quadratic sieve code into Sage since
we are currently still using ancient code there.

> Enjoy!
>
> Bill.
>

Cheers,

Michael