Todo

Ideas for Singular on top of libpolys

Fix the compilation of libkernel.a (BASE/kernel):

1. svn rebase!
2. make script (just make in BASE/kernel)
3. eliminate mod2.h (replace by aux...h)
4. reintroduce currRing/rSetCurrRing()
5. try compiling a test.cc with all the headerh #included inside (make test.oinBASE/kernel`):

• works except for kutil.h
• what is wrong with kutil.h? try to fix...

1. try compiling a test.cc with all the sources #included inside (make test.oinBASE/kernel`):

• no missing headers now!
• complains about a lot of stuff... :(

1. deal with structs.h!
2. install coeffs/mpr_complex.h and factory/templates/ftmpl_list.h
3. what about longrat.h modulop.h (i placed dummy including coeffs/number for now)
4. deal with other dummy's (re-introduced stubs instead of correcting lots of code)
5. Coordinate work with other active! people (please assign yourself) for working through files:

• structs.h

• dbm_sl.h

• eigenval.h

• fast_maps.h

• fegetopt.h

• fglm.h

• fglmgauss.h

• fglmvec.h

• gfan.h

• GMPrat.h

• hutil.h

• ideals.h

• idrec.h

• int64vec.h

• khstd.h

• kmatrix.h

• kstd1.h

• kstdfac.h

• mmalloc.h

• multicnt.h

• npolygon.h

• ratgring.h

• semic.h

• shiftgb.h

• spectrum.h

• splist.h

• stairc.h

• syz.h

• units.h

• walkMain.h

• walkProc.h

• walkSupport.h

• digitech.h // the rest (not in makefile)

• F5cData.h

• f5c.h

• F5cLists.h

• f5data.h

• f5gb.h

• f5lists.h

• fast_mult.h

• hutil.h

• linearAlgebra.h

• page.h

• ringgb.h

• run.h

• tgbgauss.h

• tgb.h

• timer.h

• kutil.h // problematic...!?

• tgb_internal.h // uses kutil :(

• F4.h // uses tgb_internal

• si_gmp.h // remove this!

• digitech.cc

• eigenval.cc

• f5c.cc

• F5cLists.cc

• f5data.cc

• f5gb.cc

• f5lists.cc

• fast_maps.cc

• fast_mult.cc

• febase.cc

• feread.cc

• fglmcomb.cc

• fglmgauss.cc

• fglmhom.cc

• fglmvec.cc

• fglmzero.cc

• gfan.cc

• GMPrat.cc

• gr_kstd2.cc

• hdegree.cc

• hilb.cc

• hutil.cc

• ideals.cc

• int64vec.cc

• khstd.cc

• kspoly.cc

• kstd1.cc

• kstd2.cc

• kstdfac.cc

• kutil.cc

• linearAlgebra.cc

• misc.cc

• multicnt.cc

• npolygon.cc

• ratgring.cc

• ringgb.cc

• semic.cc

• shiftgb.cc

• spectrum.cc

• splist.cc

• syz.cc

• syz0.cc

• syz1.cc

• syz2.cc

• syz3.cc

• tgb.cc

• tgbgauss.cc

• timer.cc

• units.cc

• walkMain.cc

• walkProc.cc

• walkSupport.cc

Ideas for improving libpolys:

• we have a lot of easy to-get-get-rid-of-warnings (in compile time): eliminate them.
• create a single libpolys.a (and libpolys.so!?) out of all our libraries (libpolys/misc.a .... etc)
• try to make use of clang and Sun-Studio-compiler for libpolys: don't work as of now :(
• no n_InpAdd/n_InpDiv: only n_InpMult!
• Moreover we have 3 (!) variants of division WITHOUT ANY DESCRIPTIONS!!!:
  1. number n_Div(number a, number b, const coeffs r)
  2. number n_IntDiv(number a, number b, const coeffs r)
  3. number n_ExactDiv(number a, number b, const coeffs r)
• Writing/reading/conversion-to-string OF polynomials/numbers is very inconsistent and misleading: n_Write/p_Write should take reporter as an argument (can be optionally = standard reporter)
• number output (n_Write) MUST NOT change a number!
• same goes for the conversion to integer by n_Int!
• Only n_Test(NUMBER, COEFFS): no way to test a domain consistency... same for polys/rings?

Errors in libpolys:

Note: Frank will take a look!

• In the face of brand new extension fields by Frank: the following seem to be obsolete together with the corresponding n_Par/n_ParDeg...right???
  1. number ndPar(int, const coeffs r) { return n_Init(0,r); }
  2. int ndParDeg(number, const coeffs) { return 0; }
• search for above and switch to the new interface (Done by Frank?)
• Ring-interface is scarcely documented: add more doxygen there (Done by Frank?)