Clean up partitions_c.cc

[jdemeyer]

1. Remove some unused stuff.

2. Compute the double and long double constants using MPFR. This makes a lot of sense since we already computed MPFR versions of these constants anyway. For extra safety, we also increase the precision used to compute these constants.

3. No longer use the header file partitions_c.h. Instead, include the .cc file in Cython.

Points 2. and 3. above allow to remove some platform-specific hacks.

4. Replace T(int(x)) by T(x) to convert x to type T.

5. Only one temporary MPFR variable is really needed, call it mptemp.

6. Merge initialize_mpz_and_mpq_variables, initialize_mpfr_variables and initialize_constants in one function initialize_globals.

7. Use some more specialized MPFR functions like mpfr_set_prec() where applicable.

8. Define all internal functions as static.

9. Use MPFR_RNDF rounding mode where exact rounding is not required.

10. Ensure that the file compiles without compiler warnings.

11. Various style improvements.

CC: @dimpase

Component: combinatorics

Author: Jeroen Demeyer

Branch/Commit: 2dba6c8

Reviewer: Dima Pasechnik

Issue created by migration from https://trac.sagemath.org/ticket/24667

[jdemeyer]

Branch: u/jdemeyer/remove_unused_stuff_from_partitions_c_cc

[jdemeyer]

New commits:

93de16e

Clean up partitions_c.cc

[jdemeyer]

Commit: 93de16e

[dimpase]

comment:16

in this chunk

@@ -83,16 +86,6 @@
  *      along with this program.  If not, see <http://www.gnu.org/licenses/>.
  */
 
-
-#if defined(__sun) || defined(__CYGWIN__)
-extern "C" long double fabsl (long double);
-extern "C" long double sinl (long double);
-extern "C" long double cosl (long double);
-extern "C" long double sqrtl (long double);
-extern "C" long double coshl (long double);
-extern "C" long double sinhl (long double);
-#endif

you're removing Gygwin stuff. Are you sure it's OK?

[jdemeyer]

comment:17

Replying to @dimpase:

Are you sure it's OK?

Yes because I removed the calls to those functions. That is what point 2 in the ticket description achieves.

[sagetrac-git]

Branch pushed to git repo; I updated commit sha1. New commits:

2dba6c8

Minor fixes

[sagetrac-git]

Changed commit from 93de16e to 2dba6c8

[mezzarobba]

comment:19

Perhaps a stupid question, sorry, but is there any reason to use this implementation instead of the one in arb?

[jdemeyer]

comment:20

Replying to @mezzarobba:

Perhaps a stupid question, sorry, but is there any reason to use this implementation instead of the one in arb?

I don't know. I'm just fixing the existing code.

[tscrim]

comment:21

Replying to @jdemeyer:

Replying to @mezzarobba:

Perhaps a stupid question, sorry, but is there any reason to use this implementation instead of the one in arb?

I don't know. I'm just fixing the existing code.

IMO, it is not a bad thing to have redundancy checking with independent implementations.

[dimpase]

comment:22

naturally, we can have several backends to computing a function. Patches are welcome.

[dimpase]

Reviewer: Dima Pasechnik

[dimpase]

comment:23

looks good to me. Send it to the bots!

[vbraun]

Changed branch from u/jdemeyer/remove_unused_stuff_from_partitions_c_cc to 2dba6c8