added compatibility with Flint 3.1.0

[mahrud]

No description provided.

[d-torrance]

I'm still getting a build error with FLINT 3.1.0 after applying this patch:

CXX NCAlgebras/NCF4.cpp
In file included from /home/profzoom/src/macaulay2/M2/M2/BUILD/doug/usr-host/include/flint/flint.h:197,
                 from ../../../../Macaulay2/e/aring-qq-flint.hpp:13,
                 from ../../../../Macaulay2/e/aring-qq.hpp:6,
                 from ../../../../Macaulay2/e/aring-translate.hpp:18,
                 from ../../../../Macaulay2/e/aring-glue.hpp:7,
                 from ../../../../Macaulay2/e/VectorArithmetic.hpp:42,
                 from ../../../../Macaulay2/e/NCAlgebras/NCF4.hpp:14,
                 from ../../../../Macaulay2/e/NCAlgebras/NCF4.cpp:1:
/home/profzoom/src/macaulay2/M2/M2/BUILD/doug/usr-host/include/flint/longlong.h:74: warning: "__ll_B" redefined
   74 | #define __ll_B ((ulong) 1 << (FLINT_BITS / 2))
      | 
In file included from /usr/include/recint/recdefine.h:145,
                 from /usr/include/recint/ruruint.h:43,
                 from /usr/include/recint/rumanip.h:42,
                 from /usr/include/recint/ruconvert.h:21,
                 from /usr/include/gmp++/gmp++_int.h:26,
                 from /usr/include/gmp++/gmp++.h:57,
                 from /usr/include/givaro/givinteger.h:19,
                 from /usr/include/givaro/givpoly1dense.h:22,
                 from /usr/include/givaro/givpoly1.h:38,
                 from /usr/include/fflas-ffpack/ffpack/ffpack.h:38,
                 from ../../../../Macaulay2/e/aring-zzp-ffpack.hpp:17,
                 from ../../../../Macaulay2/e/aring-translate.hpp:17:
/usr/include/recint/reclonglong.h:46: note: this is the location of the previous definition
   46 | #define __ll_B ((UWtype) 1 << (W_TYPE_SIZE / 2))
      | 
/home/profzoom/src/macaulay2/M2/M2/BUILD/doug/usr-host/include/flint/longlong.h:75: warning: "__ll_lowpart" redefined
   75 | #define __ll_lowpart(t) ((ulong) (t) & (__ll_B - 1))
      | 
/usr/include/recint/reclonglong.h:47: note: this is the location of the previous definition
   47 | #define __ll_lowpart(t) ((UWtype) (t) & (__ll_B - 1))
      | 
/home/profzoom/src/macaulay2/M2/M2/BUILD/doug/usr-host/include/flint/longlong.h:76: warning: "__ll_highpart" redefined
   76 | #define __ll_highpart(t) ((ulong) (t) >> (FLINT_BITS / 2))
      | 
/usr/include/recint/reclonglong.h:48: note: this is the location of the previous definition
   48 | #define __ll_highpart(t) ((UWtype) (t) >> (W_TYPE_SIZE / 2))
      | 
In file included from ../../../../Macaulay2/e/aring-translate.hpp:24:
../../../../Macaulay2/e/aring-zzp-flint.hpp: In member function ‘void M2::ARingZZpFlint::negate(ElementType&, ElementType) const’:
../../../../Macaulay2/e/aring-zzp-flint.hpp:121:14: error: ‘nmod_neg’ was not declared in this scope; did you mean ‘nmod_t’?
  121 |     result = nmod_neg(a, mModulus);
      |              ^~~~~~~~
      |              nmod_t
../../../../Macaulay2/e/aring-zzp-flint.hpp: In member function ‘void M2::ARingZZpFlint::invert(ElementType&, ElementType) const’:
../../../../Macaulay2/e/aring-zzp-flint.hpp:128:14: error: ‘n_invmod’ was not declared in this scope; did you mean ‘fmpz_invmod’?
  128 |     result = n_invmod(a, mCharac);
      |              ^~~~~~~~
      |              fmpz_invmod
../../../../Macaulay2/e/aring-zzp-flint.hpp: In member function ‘void M2::ARingZZpFlint::add(ElementType&, ElementType, ElementType) const’:
../../../../Macaulay2/e/aring-zzp-flint.hpp:133:14: error: ‘nmod_add’ was not declared in this scope; did you mean ‘nmod_t’?
  133 |     result = nmod_add(a, b, mModulus);
      |              ^~~~~~~~
      |              nmod_t
../../../../Macaulay2/e/aring-zzp-flint.hpp: In member function ‘void M2::ARingZZpFlint::subtract(ElementType&, ElementType, ElementType) const’:
../../../../Macaulay2/e/aring-zzp-flint.hpp:138:14: error: ‘nmod_sub’ was not declared in this scope; did you mean ‘nmod_t’?
  138 |     result = nmod_sub(a, b, mModulus);
      |              ^~~~~~~~
      |              nmod_t
../../../../Macaulay2/e/aring-zzp-flint.hpp: In member function ‘void M2::ARingZZpFlint::subtract_multiple(ElementType&, ElementType, ElementType) const’:
../../../../Macaulay2/e/aring-zzp-flint.hpp:145:22: error: ‘nmod_neg’ was not declared in this scope; did you mean ‘nmod_t’?
  145 |     ElementType a1 = nmod_neg(a, mModulus);
      |                      ^~~~~~~~
      |                      nmod_t
../../../../Macaulay2/e/aring-zzp-flint.hpp:160:5: error: ‘NMOD_ADDMUL’ was not declared in this scope
  160 |     NMOD_ADDMUL(result, a1, b, mModulus);
      |     ^~~~~~~~~~~
../../../../Macaulay2/e/aring-zzp-flint.hpp: In member function ‘void M2::ARingZZpFlint::mult(ElementType&, ElementType, ElementType) const’:
../../../../Macaulay2/e/aring-zzp-flint.hpp:165:14: error: ‘nmod_mul’ was not declared in this scope; did you mean ‘nmod_t’?
  165 |     result = nmod_mul(a, b, mModulus);
      |              ^~~~~~~~
      |              nmod_t
../../../../Macaulay2/e/aring-zzp-flint.hpp: In member function ‘void M2::ARingZZpFlint::divide(ElementType&, ElementType, ElementType) const’:
../../../../Macaulay2/e/aring-zzp-flint.hpp:172:14: error: ‘nmod_div’ was not declared in this scope; did you mean ‘nmod_t’?
  172 |     result = nmod_div(a, b, mModulus);
      |              ^~~~~~~~
      |              nmod_t
../../../../Macaulay2/e/aring-zzp-flint.hpp: In member function ‘void M2::ARingZZpFlint::power(ElementType&, ElementType, long int) const’:
../../../../Macaulay2/e/aring-zzp-flint.hpp:179:18: error: ‘n_powmod2_preinv’ was not declared in this scope
  179 |         result = n_powmod2_preinv(a, n, mModulus.n, mModulus.ninv);
      |                  ^~~~~~~~~~~~~~~~
../../../../Macaulay2/e/aring-zzp-flint.hpp: In member function ‘void M2::ARingZZpFlint::power_mpz(ElementType&, ElementType, mpz_srcptr) const’:
../../../../Macaulay2/e/aring-zzp-flint.hpp:195:18: error: ‘n_powmod2_ui_preinv’ was not declared in this scope
  195 |         result = n_powmod2_ui_preinv(a, nbar, mModulus.n, mModulus.ninv);
      |                  ^~~~~~~~~~~~~~~~~~~
In file included from ../../../../Macaulay2/e/mutablemat-defs.hpp:86,
                 from ../../../../Macaulay2/e/mutablemat.hpp:6,
                 from ../../../../Macaulay2/e/aring-glue.hpp:10:
../../../../Macaulay2/e/mat-linalg.hpp: In function ‘size_t MatrixOps::rowReducedEchelonForm(const DMatGFFlint&, DMatGFFlint&)’:
../../../../Macaulay2/e/mat-linalg.hpp:932:71: error: cannot convert ‘const fq_zech_ctx_struct*’ to ‘const fq_zech_mat_struct*’
  932 | ong rank = fq_zech_mat_rref(A1.fq_zech_mat(), A.ring().flintContext());
      |                                               ~~~~~~~~~~~~~~~~~~~~~^~
      |                                                                    |
      |                                                                    const fq_zech_ctx_struct*

In file included from /home/profzoom/src/macaulay2/M2/M2/BUILD/doug/usr-host/include/flint/fq_zech_mat.h:36,
                 from ../../../../Macaulay2/e/dmat-gf-flint.hpp:15,
                 from ../../../../Macaulay2/e/dmat.hpp:69,
                 from ../../../../Macaulay2/e/mutablemat-defs.hpp:82:
/home/profzoom/src/macaulay2/M2/M2/BUILD/doug/usr-host/include/flint/fq_mat_templates.h:247:76: note:   initializing argument 2 of ‘mp_limb_signed_t fq_zech_mat_rref(fq_zech_mat_struct*, const fq_zech_mat_struct*, const fq_zech_ctx_struct*)’
  247 | T, mat_rref)(TEMPLATE(T, mat_t) B, const TEMPLATE(T, mat_t) A, const TEMPLATE(T, ctx_t) ctx);
      |                                    ~~~~~~~~~~~~~~~~~~~~~~~~~^

At global scope:
cc1plus: note: unrecognized command-line option ‘-Wno-macro-redefined’ may have been intended to silence earlier diagnostics
make[2]: *** [Makefile.common:33: NCAlgebras/NCF4.o] Error 1
make[2]: Leaving directory '/home/profzoom/src/macaulay2/M2/M2/BUILD/doug/Macaulay2/e'
make[1]: *** [Makefile:15: all-in-e] Error 2
make[1]: Leaving directory '/home/profzoom/src/macaulay2/M2/M2/BUILD/doug/Macaulay2'
make: *** [GNUmakefile:251: all-in-Macaulay2] Error 2
make: Leaving directory '/home/profzoom/src/macaulay2/M2/M2/BUILD/doug'

[mahrud]

Yea I realized it's not enough. I'm evaluating how twisted of a solution is warranted here (e.g. a macro that repeats a variable depending on version of flint, or just a bunch of if/elses)

[mikestillman]

If this compiles under flint 3.1.0, it looks good (and a simple fix too).

[mikestillman]

I was getting ready to look at this, but it looks like you have it done! Is this all set possibly? I'll look at the changes you made, actually, just did, looks good. It looks like it is not running builds though. @DanGrayson, should I just accept it?

[mahrud]

This branch isn't set up to have tests. I don't really understand why there's a separate branch for the release. Everything should be happening on development and occasionally there should be a new release tag.

[mikestillman]

@mahrud If I update brew now, it will install flint 3.1.0, and remove the old version? So if I want to keep doing development on other branches, I'll need to cherry pick these fixes in to these branches?

[mahrud]

Not to mention that this commit is will be missing from development, so you'd need to merge this branch into development, adding an entirely unnecessary merge commit.

[d-torrance]

I'm still getting an error:

make -C e all 
make[2]: Entering directory '/home/profzoom/src/macaulay2/M2/M2/BUILD/doug/Macaulay2/e'
CXX NCAlgebras/NCF4.cpp
In file included from /home/profzoom/src/macaulay2/M2/M2/BUILD/doug/usr-host/include/flint/flint.h:197,
                 from ../../../../Macaulay2/e/aring-qq-flint.hpp:13,
                 from ../../../../Macaulay2/e/aring-qq.hpp:6,
                 from ../../../../Macaulay2/e/aring-translate.hpp:18,
                 from ../../../../Macaulay2/e/aring-glue.hpp:7,
                 from ../../../../Macaulay2/e/VectorArithmetic.hpp:42,
                 from ../../../../Macaulay2/e/NCAlgebras/NCF4.hpp:14,
                 from ../../../../Macaulay2/e/NCAlgebras/NCF4.cpp:1:
/home/profzoom/src/macaulay2/M2/M2/BUILD/doug/usr-host/include/flint/longlong.h:74: warning: "__ll_B" redefined
   74 | #define __ll_B ((ulong) 1 << (FLINT_BITS / 2))
      | 
In file included from /usr/include/recint/recdefine.h:145,
                 from /usr/include/recint/ruruint.h:43,
                 from /usr/include/recint/rumanip.h:42,
                 from /usr/include/recint/ruconvert.h:21,
                 from /usr/include/gmp++/gmp++_int.h:26,
                 from /usr/include/gmp++/gmp++.h:57,
                 from /usr/include/givaro/givinteger.h:19,
                 from /usr/include/givaro/givpoly1dense.h:22,
                 from /usr/include/givaro/givpoly1.h:38,
                 from /usr/include/fflas-ffpack/ffpack/ffpack.h:38,
                 from ../../../../Macaulay2/e/aring-zzp-ffpack.hpp:17,
                 from ../../../../Macaulay2/e/aring-translate.hpp:17:
/usr/include/recint/reclonglong.h:46: note: this is the location of the previous definition
   46 | #define __ll_B ((UWtype) 1 << (W_TYPE_SIZE / 2))
      | 
/home/profzoom/src/macaulay2/M2/M2/BUILD/doug/usr-host/include/flint/longlong.h:75: warning: "__ll_lowpart" redefined
   75 | #define __ll_lowpart(t) ((ulong) (t) & (__ll_B - 1))
      | 
/usr/include/recint/reclonglong.h:47: note: this is the location of the previous definition
   47 | #define __ll_lowpart(t) ((UWtype) (t) & (__ll_B - 1))
      | 
/home/profzoom/src/macaulay2/M2/M2/BUILD/doug/usr-host/include/flint/longlong.h:76: warning: "__ll_highpart" redefined
   76 | #define __ll_highpart(t) ((ulong) (t) >> (FLINT_BITS / 2))
      | 
/usr/include/recint/reclonglong.h:48: note: this is the location of the previous definition
   48 | #define __ll_highpart(t) ((UWtype) (t) >> (W_TYPE_SIZE / 2))
      | 
In file included from ../../../../Macaulay2/e/aring-translate.hpp:24:
../../../../Macaulay2/e/aring-zzp-flint.hpp: In member function ‘void M2::ARingZZpFlint::negate(ElementType&, ElementType) const’:
../../../../Macaulay2/e/aring-zzp-flint.hpp:121:14: error: ‘nmod_neg’ was not declared in this scope; did you mean ‘nmod_t’?
  121 |     result = nmod_neg(a, mModulus);
      |              ^~~~~~~~
      |              nmod_t
../../../../Macaulay2/e/aring-zzp-flint.hpp: In member function ‘void M2::ARingZZpFlint::invert(ElementType&, ElementType) const’:
../../../../Macaulay2/e/aring-zzp-flint.hpp:128:14: error: ‘n_invmod’ was not declared in this scope; did you mean ‘fmpz_invmod’?
  128 |     result = n_invmod(a, mCharac);
      |              ^~~~~~~~
      |              fmpz_invmod
../../../../Macaulay2/e/aring-zzp-flint.hpp: In member function ‘void M2::ARingZZpFlint::add(ElementType&, ElementType, ElementType) const’:
../../../../Macaulay2/e/aring-zzp-flint.hpp:133:14: error: ‘nmod_add’ was not declared in this scope; did you mean ‘nmod_t’?
  133 |     result = nmod_add(a, b, mModulus);
      |              ^~~~~~~~
      |              nmod_t
../../../../Macaulay2/e/aring-zzp-flint.hpp: In member function ‘void M2::ARingZZpFlint::subtract(ElementType&, ElementType, ElementType) const’:
../../../../Macaulay2/e/aring-zzp-flint.hpp:138:14: error: ‘nmod_sub’ was not declared in this scope; did you mean ‘nmod_t’?
  138 |     result = nmod_sub(a, b, mModulus);
      |              ^~~~~~~~
      |              nmod_t
../../../../Macaulay2/e/aring-zzp-flint.hpp: In member function ‘void M2::ARingZZpFlint::subtract_multiple(ElementType&, ElementType, ElementType) const’:
../../../../Macaulay2/e/aring-zzp-flint.hpp:145:22: error: ‘nmod_neg’ was not declared in this scope; did you mean ‘nmod_t’?
  145 |     ElementType a1 = nmod_neg(a, mModulus);
      |                      ^~~~~~~~
      |                      nmod_t
../../../../Macaulay2/e/aring-zzp-flint.hpp:160:5: error: ‘NMOD_ADDMUL’ was not declared in this scope
  160 |     NMOD_ADDMUL(result, a1, b, mModulus);
      |     ^~~~~~~~~~~
../../../../Macaulay2/e/aring-zzp-flint.hpp: In member function ‘void M2::ARingZZpFlint::mult(ElementType&, ElementType, ElementType) const’:
../../../../Macaulay2/e/aring-zzp-flint.hpp:165:14: error: ‘nmod_mul’ was not declared in this scope; did you mean ‘nmod_t’?
  165 |     result = nmod_mul(a, b, mModulus);
      |              ^~~~~~~~
      |              nmod_t
../../../../Macaulay2/e/aring-zzp-flint.hpp: In member function ‘void M2::ARingZZpFlint::divide(ElementType&, ElementType, ElementType) const’:
../../../../Macaulay2/e/aring-zzp-flint.hpp:172:14: error: ‘nmod_div’ was not declared in this scope; did you mean ‘nmod_t’?
  172 |     result = nmod_div(a, b, mModulus);
      |              ^~~~~~~~
      |              nmod_t
../../../../Macaulay2/e/aring-zzp-flint.hpp: In member function ‘void M2::ARingZZpFlint::power(ElementType&, ElementType, long int) const’:
../../../../Macaulay2/e/aring-zzp-flint.hpp:179:18: error: ‘n_powmod2_preinv’ was not declared in this scope
  179 |         result = n_powmod2_preinv(a, n, mModulus.n, mModulus.ninv);
      |                  ^~~~~~~~~~~~~~~~
../../../../Macaulay2/e/aring-zzp-flint.hpp: In member function ‘void M2::ARingZZpFlint::power_mpz(ElementType&, ElementType, mpz_srcptr) const’:
../../../../Macaulay2/e/aring-zzp-flint.hpp:195:18: error: ‘n_powmod2_ui_preinv’ was not declared in this scope
  195 |         result = n_powmod2_ui_preinv(a, nbar, mModulus.n, mModulus.ninv);
      |                  ^~~~~~~~~~~~~~~~~~~
At global scope:
cc1plus: note: unrecognized command-line option ‘-Wno-macro-redefined’ may have been intended to silence earlier diagnostics
make[2]: *** [Makefile.common:33: NCAlgebras/NCF4.o] Error 1
make[2]: Leaving directory '/home/profzoom/src/macaulay2/M2/M2/BUILD/doug/Macaulay2/e'
make[1]: *** [Makefile:15: all-in-e] Error 2
make[1]: Leaving directory '/home/profzoom/src/macaulay2/M2/M2/BUILD/doug/Macaulay2'
make: *** [GNUmakefile:251: all-in-Macaulay2] Error 2
make: Leaving directory '/home/profzoom/src/macaulay2/M2/M2/BUILD/doug'

[mahrud]

The builds worked locally for me and also here on ubuntu and macOS: Macaulay2/homebrew-tap#217

[d-torrance]

The recint stuff makes me think it might be another givaro/flint header order thing...

[mikestillman]

@d-torrance I'll take a look. Is this the only commit so far tying to fix flint 3.1.0 use?

@mahrud @DanGrayson @d-torrance Maybe we could streamline build release branches/tags. We should discuss what we can do to streamline the build process, especially if the two of you (or a subset, or others too) will be doing releases going forward.

[mikestillman]

I think I'll plan on localizing all use of ffpack and givaro to one file (that doesn't have flint either). This will entail removing ARing types from FFPACK and Givaro (or rather, using them in one file only, for computing rank of a dense matrix over a prime finite field. This will not be for this release.

[mikestillman]

@d-torrance This appears in aring-zzp-flint.hpp. I don't see the define anywhere. Do you know how this is being set?

 #ifdef HAVE_FLINT_NMOD_H
  #include <flint/nmod.h>   // for nmod_neg, nmod_add, nmod_div, nmod_mul
#endif

edit: nevermind, I didn't grep far enough up the food chain, it os being set during configure time.

It certainly seems that on the machine where the failure is occuring, it is either not including this file, or if so, it isn't any longer referring to these functions?

[d-torrance]

Yes -- it's set during configuration. See the following commits:

• 2777eb2 (autotools)
• d068a38 (cmake)

[d-torrance]

Mike, I think you're right! The autotools build only sets that #define if it detects a newer version of FLINT during configuration. But I'm building FLINT after that.

[d-torrance]

Yup, that did the trick! I'm still getting those recint warnings, but no more errors.

[d-torrance]

The recint compiler warnings go away if the FLINT headers appear before the Givaro ones, but swapping the order leads to errors on various architectures, so I guess we'll just have to deal with them. I've submitted an issue to Givaro.

[mahrud]

Maybe we could streamline build release branches/tags. We should discuss what we can do to streamline the build process, especially if the two of you (or a subset, or others too) will be doing releases going forward.

I didn't realize that was the plan. Is this the expectation for this release?

[d-torrance]

Should we update the FLINT submodule to point to 3.1.0? (Not necessarily for the release, but maybe for the development branch?)

On a related note, I haven't forgotten that I volunteered to update the autotools build to use the submodule a few months ago. :)

[mikestillman]

@mahrud I think maybe going forward, but not for this release (I only said you are part of the group doing releases, because you are creating the brew Macaulay2 release each time already...)