Prototype fixes #2452

albinahlback · 2025-10-30T14:24:06Z

Background

With #2438, I enabled warnings when global functions were defined without a prior prototype given. This intends to warns users that

A global function is defined without it being defined in a header;
A global function should be declared static.

Moreover, declaring prototypes inside source files should be considered a bad habit. Typically in FLINT, prototypes are given inside source files for:

When an internal function is defined in another source file;
Forward declarations (which is usually a good use case).

However, when giving the prototype inside a source file, no check is done for ABI compatibilities. Hence, things may break.

This PR

In this PR, I declare a lot of functions static. Moreover, I implement */impl.h headers, intended for declaring internal functions and use them where necessary.

However, this PR is not complete, and I will continue working on doing all modules (now I have only done it up until GR) and all tests (so that I can push -Wmissing-prototypes to the tests).

albinahlback · 2025-10-30T14:24:50Z

I used dev/check_prototypes to list all prototypes given in the source files. Some exceptions exists, which I have listed in there.

src/acb/zeta.c

albinahlback · 2025-10-30T14:26:51Z

src/acb_dirichlet/vec_mellin_arb.c

This file is completely unused in FLINT.

This is potentially a useful algorithm. Maybe @pascalmolin can document it?

albinahlback · 2025-10-30T14:27:57Z

src/arb_mat/entrywise_is_zero.c

I removed this file and entrywise_not_is_zero.c. Not sure why they exist.

They were added at some point because the author needed them. It's possible that they are used externally. We can perhaps deprecate them and remove them in the next next release?

albinahlback · 2025-10-30T14:29:25Z

src/arb/log_precompute_reductions.c

This was also unused? @fredrik-johansson can you just verify that this has no use case for FLINT? I recall it existing in some comment somewhere.

This function is useful.

albinahlback · 2025-10-30T14:30:34Z

src/fmpz_mpoly_factor/add.c

I think at some point Dan wanted to do arithmetic on fmpz_mpoly in factored form, but this function is not very interesting on its own. OK to delete.

fredrik-johansson · 2025-10-30T15:21:08Z

Could you please do this as several independent PRs, where each PR does one kind of change?

fredrik-johansson · 2025-10-30T15:25:36Z

I'm really not convinced that the impl.h headers are an improvement. In the cases I looked at, we could just decide whether the functions should be static or not.

fredrik-johansson · 2025-10-30T15:49:11Z

src/nmod_mpoly/mpolyun.c

@@ -404,301 +404,6 @@ int nmod_mpolyun_equal(
    return 1;
 }

-


I'm a bit uneasy about outright deleting this kind of mpoly code. It sometimes helps understand the (poorly documented) code that is actually used. Could you ifdef it out instead, with a comment like "This code is unused. TODO: delete it when the surrounding code is adequately tested and commented."? I think I've already done this elsewhere.

albinahlback · 2025-10-31T09:00:36Z

src/qsieve/large_prime_variant.c

-/*
-    Display a relation for debugging purposes
-*/
-void qsieve_display_relation(qs_t qs_inf, relation_t a)
-{
-    slong i;
-
-    flint_printf("%wu ", a.lp);
-
-    for (i = 0; i < qs_inf->small_primes; i++)


albinahlback · 2025-10-31T09:01:04Z

src/nmod_mpoly/mpolyu.c

-
-void nmod_mpolyu_msub(nmod_mpolyu_t R, nmod_mpolyu_t A, nmod_mpolyu_t B,
-                           nmod_mpoly_t c, slong e, const nmod_mpoly_ctx_t ctx)
-{
-    slong i, j, k;
-    nmod_mpoly_t T;
-
-    nmod_mpolyu_fit_length(R, A->length + B->length, ctx);
-


albinahlback · 2025-10-31T09:01:43Z

src/n_poly/n_fq_poly_divrem.c

-void _n_fq_poly_rem_basecase_(
-    ulong * Q,
-    ulong * A,
-    const ulong * AA, slong Alen,
-    const ulong * B, slong Blen,
-    const ulong * invB,
-    const fq_nmod_ctx_t ctx,
-    n_poly_stack_t St)
-{
-    slong i;
-    slong d = fq_nmod_ctx_degree(ctx);
-    nmod_t mod = fq_nmod_ctx_mod(ctx);
-    ulong * tmp = n_poly_stack_vec_init(St, d*(3 + N_FQ_POLY_DIVREM_BASECASE_ITCH));
-    ulong * u = tmp + d*N_FQ_POLY_DIVREM_BASECASE_ITCH;
-    ulong * q0 = u + d;
-    ulong * q1 = q0 + d;
-
-    if (A != AA)
-        _nmod_vec_set(A, AA, d*Alen);


albinahlback · 2025-10-31T09:02:51Z

src/fq_zech_mpoly/repack_bits.c

-int fq_zech_mpoly_repack_bits(fq_zech_mpoly_t A, const fq_zech_mpoly_t B,
-                              flint_bitcnt_t Abits, const fq_zech_mpoly_ctx_t ctx)
-{
-    slong i;
-    int success;
-    fq_zech_mpoly_t T;
-
-    Abits = mpoly_fix_bits(Abits, ctx->minfo);


albinahlback · 2025-10-31T09:03:02Z

src/fq_zech_mpoly/mpolyu.c

-
-/* if the coefficient doesn't exist, a new one is created (and set to zero) */
-fq_zech_mpoly_struct * _fq_zech_mpolyu_get_coeff(fq_zech_mpolyu_t A,


albinahlback · 2025-10-31T09:03:20Z

src/fq_zech_mpoly/evaluate_all.c

-void fq_zech_mpoly_evaluate_all_ui(fq_zech_t eval, const fq_zech_mpoly_t A,
-                  fq_zech_struct * const * vals, const fq_zech_mpoly_ctx_t ctx)
-{


albinahlback · 2025-10-31T09:03:37Z

src/fq_nmod_mpoly/mpolyu.c

-void fq_nmod_mpoly_to_mpolyuu_perm_deflate(
-    fq_nmod_mpolyu_t A,
-    const fq_nmod_mpoly_ctx_t uctx,
-    const fq_nmod_mpoly_t B,


albinahlback · 2025-10-31T09:05:07Z

src/fmpzi/gcd_binary.c

-fmpzi_norm_approx_d_2exp(slong * exp, const fmpzi_t x)
-{
-    double a, b;
-    slong aexp, bexp;
-    int e;
-
-    a = fmpz_get_d_2exp(&aexp, fmpzi_realref(x));
-    b = fmpz_get_d_2exp(&bexp, fmpzi_imagref(x));


albinahlback · 2025-10-31T09:07:23Z

src/fmpz_mpoly/to_from_fmpz_poly.c

-/*
-    assuming that the conversion can be done, set poly1 and poly1_shift
-    so that poly2 is poly1 * X^poly1_shift
-    TODO: handle multiprecision exponents
-*/


move to header

albinahlback · 2025-10-31T09:08:11Z

src/fmpz_mod_mpoly_factor/mpoly_hlift_zippel.c

-void fmpz_mod_mpolyu_repack_bits_inplace(
-    fmpz_mod_mpolyu_t A,
-    flint_bitcnt_t bits,
-    const fmpz_mod_mpoly_ctx_t ctx)


albinahlback · 2025-10-31T09:08:17Z

src/fmpz_mod_mpoly_factor/mpoly_hlift_zippel.c

-void fmpz_mod_mpolyu_one(fmpz_mod_mpolyu_t A, const fmpz_mod_mpoly_ctx_t uctx)
-{
-    fmpz_mod_mpolyu_fit_length(A, WORD(1), uctx);


albinahlback · 2025-10-31T09:09:39Z

src/fmpz_mod_mpoly_factor/interp.c

-/*
-    E = A(v = alpha)
-    A is in R[v][x]
-    E is in R[x]


albinahlback · 2025-10-31T09:10:01Z

src/fmpz_mod_mpoly/univar.c

-void fmpz_mod_mpoly_univar_set(
-    fmpz_mod_mpoly_univar_t A,
-    const fmpz_mod_mpoly_univar_t B,
-    const fmpz_mod_mpoly_ctx_t ctx)
-{
-    slong i;


albinahlback · 2025-10-31T09:17:59Z

src/acb_dirichlet/theta_arb.c

-void
-_acb_dirichlet_theta_arb_series(acb_t res, const dirichlet_group_t G, const dirichlet_char_t chi, const arb_t xt, slong len, slong prec)
-{


albinahlback · 2025-10-31T09:18:11Z

src/acb_dirichlet/platt_local_hardy_z_zeros.c

-acb_dirichlet_platt_isolate_local_hardy_z_zeros(
-        arf_interval_ptr res, const fmpz_t n, slong len, slong prec)
-{
-    if (len <= 0 || fmpz_sizeinbase(n, 10) < 5)
-    {


albinahlback · 2025-10-31T09:18:29Z

src/acb_dirichlet/platt_local_hardy_z_zeros.c

-slong
-_acb_dirichlet_platt_isolate_local_hardy_z_zeros(
-        arf_interval_ptr res, const fmpz_t n, slong len,
-        const fmpz_t T, slong A, slong B,
-        const arb_t h, const fmpz_t J, slong K, slong sigma_grid,


albinahlback · 2025-10-31T09:19:21Z

src/acb_dft/cyc.c

-void
-acb_dft_inverse_cyc(acb_ptr w, acb_srcptr v, slong len, slong prec)
-{


albinahlback · 2025-10-31T09:19:47Z

src/acb/rsqrt.c

-void
-mag_rsqrt_re_quadrant1_lower(mag_t res, const mag_t x, const mag_t y)
-{
-    if (mag_is_zero(x))
-    {
-        mag_one(res);
-        mag_mul_2exp_si(res, res, -1);
-    }
-    else
-    {


albinahlback · 2025-10-31T10:42:51Z

@fredrik-johansson everything should be fixed accordingly now.

Add dev/check_prototypes for listing prototypes

33b0d24

albinahlback commented Oct 30, 2025

View reviewed changes

src/acb/zeta.c Show resolved Hide resolved

albinahlback commented Oct 30, 2025

View reviewed changes

albinahlback force-pushed the prototype_fixes branch 2 times, most recently from 17ac6c3 to 5fd7ecb Compare October 30, 2025 15:20

fredrik-johansson reviewed Oct 30, 2025

View reviewed changes

albinahlback commented Oct 31, 2025

View reviewed changes

albinahlback and others added 3 commits October 31, 2025 11:42

Fix prototypes by adding */impl.h headers

f5a2352

Fix warning in test

40e6a96

Fix function attributes in exception.c

501aa7a

albinahlback force-pushed the prototype_fixes branch from 5fd7ecb to 501aa7a Compare October 31, 2025 10:42

albinahlback merged commit 88eb14c into flintlib:main Nov 1, 2025
14 checks passed

albinahlback deleted the prototype_fixes branch November 1, 2025 11:57


		/* if the coefficient doesn't exist, a new one is created (and set to zero) */
		fq_zech_mpoly_struct * _fq_zech_mpolyu_get_coeff(fq_zech_mpolyu_t A,

Prototype fixes #2452

Prototype fixes #2452

Uh oh!

Conversation

albinahlback commented Oct 30, 2025

Background

This PR

Uh oh!

albinahlback commented Oct 30, 2025

Uh oh!

Uh oh!

Choose a reason for hiding this comment

Uh oh!

Choose a reason for hiding this comment

Uh oh!

Choose a reason for hiding this comment

Uh oh!

Choose a reason for hiding this comment

Uh oh!

Choose a reason for hiding this comment

Uh oh!

Choose a reason for hiding this comment

Uh oh!

Choose a reason for hiding this comment

Uh oh!

Choose a reason for hiding this comment

Uh oh!

fredrik-johansson commented Oct 30, 2025

Uh oh!

fredrik-johansson commented Oct 30, 2025

Uh oh!

Choose a reason for hiding this comment

Uh oh!

Choose a reason for hiding this comment

Uh oh!

Choose a reason for hiding this comment

Uh oh!

Choose a reason for hiding this comment

Uh oh!

Choose a reason for hiding this comment

Uh oh!

Choose a reason for hiding this comment

Uh oh!

Choose a reason for hiding this comment

Uh oh!

Choose a reason for hiding this comment

Uh oh!

Choose a reason for hiding this comment

Uh oh!

Choose a reason for hiding this comment

Uh oh!

Choose a reason for hiding this comment

Uh oh!

Choose a reason for hiding this comment

Uh oh!

Choose a reason for hiding this comment

Uh oh!

Choose a reason for hiding this comment

Uh oh!

Choose a reason for hiding this comment

Uh oh!

Choose a reason for hiding this comment

Uh oh!

Choose a reason for hiding this comment

Uh oh!

Choose a reason for hiding this comment

Uh oh!

Choose a reason for hiding this comment

Uh oh!

Choose a reason for hiding this comment

Uh oh!

albinahlback commented Oct 31, 2025

Uh oh!

Uh oh!

Reviewers

Assignees

Labels

Projects

Milestone

Development