Some heimdal libs can't be linked if I run configure with --builtin-libraries=ALL This is reproducable with: ./configure --builtin-libraries=ALL && make HEAD is on 8972c92342211fefb3d76bc306f7fbc467089010 [3213/3514] Linking default/source4/heimdal_build/libhcrypto-samba4.so default/source4/heimdal/lib/hcrypto/rand-egd_86.o: In function `connect_egd': rand-egd.c:(.text+0x68): undefined reference to `rep_strlcpy' collect2: ld returned 1 exit status Waf: Leaving directory `/home/bbaumba/src/git/samba/bin' Build failed: -> task failed (err #1): {task: cc_link bncore_85.o,bn_mp_init_85.o,bn_mp_clear_85.o,bn_mp_exch_85.o,bn_mp_grow_85.o,bn_mp_shrink_85.o,bn_mp_clamp_85.o,bn_mp_zero_85.o,bn_mp_zero_multi_85.o,bn_mp_set_85.o,bn_mp_set_int_85.o,bn_mp_init_size_85.o,bn_mp_copy_85.o,bn_mp_init_copy_85.o,bn_mp_abs_85.o,bn_mp_neg_85.o,bn_mp_cmp_mag_85.o,bn_mp_cmp_85.o,bn_mp_cmp_d_85.o,bn_mp_rshd_85.o,bn_mp_lshd_85.o,bn_mp_mod_2d_85.o,bn_mp_div_2d_85.o,bn_mp_mul_2d_85.o,bn_mp_div_2_85.o,bn_mp_mul_2_85.o,bn_s_mp_add_85.o,bn_s_mp_sub_85.o,bn_fast_s_mp_mul_digs_85.o,bn_s_mp_mul_digs_85.o,bn_fast_s_mp_mul_high_digs_85.o,bn_s_mp_mul_high_digs_85.o,bn_fast_s_mp_sqr_85.o,bn_s_mp_sqr_85.o,bn_mp_add_85.o,bn_mp_sub_85.o,bn_mp_karatsuba_mul_85.o,bn_mp_mul_85.o,bn_mp_karatsuba_sqr_85.o,bn_mp_sqr_85.o,bn_mp_div_85.o,bn_mp_mod_85.o,bn_mp_add_d_85.o,bn_mp_sub_d_85.o,bn_mp_mul_d_85.o,bn_mp_div_d_85.o,bn_mp_mod_d_85.o,bn_mp_expt_d_85.o,bn_mp_addmod_85.o,bn_mp_submod_85.o,bn_mp_mulmod_85.o,bn_mp_sqrmod_85.o,bn_mp_gcd_85.o,bn_mp_lcm_85.o,bn_fast_mp_invmod_85.o,bn_mp_invmod_85.o,bn_mp_reduce_85.o,bn_mp_montgomery_setup_85.o,bn_fast_mp_montgomery_reduce_85.o,bn_mp_montgomery_reduce_85.o,bn_mp_exptmod_fast_85.o,bn_mp_exptmod_85.o,bn_mp_2expt_85.o,bn_mp_n_root_85.o,bn_mp_jacobi_85.o,bn_reverse_85.o,bn_mp_count_bits_85.o,bn_mp_read_unsigned_bin_85.o,bn_mp_read_signed_bin_85.o,bn_mp_to_unsigned_bin_85.o,bn_mp_to_signed_bin_85.o,bn_mp_unsigned_bin_size_85.o,bn_mp_signed_bin_size_85.o,bn_mp_xor_85.o,bn_mp_and_85.o,bn_mp_or_85.o,bn_mp_rand_85.o,bn_mp_montgomery_calc_normalization_85.o,bn_mp_prime_is_divisible_85.o,bn_prime_tab_85.o,bn_mp_prime_fermat_85.o,bn_mp_prime_miller_rabin_85.o,bn_mp_prime_is_prime_85.o,bn_mp_prime_next_prime_85.o,bn_mp_find_prime_85.o,bn_mp_isprime_85.o,bn_mp_dr_reduce_85.o,bn_mp_dr_is_modulus_85.o,bn_mp_dr_setup_85.o,bn_mp_reduce_setup_85.o,bn_mp_toom_mul_85.o,bn_mp_toom_sqr_85.o,bn_mp_div_3_85.o,bn_s_mp_exptmod_85.o,bn_mp_reduce_2k_85.o,bn_mp_reduce_is_2k_85.o,bn_mp_reduce_2k_setup_85.o,bn_mp_reduce_2k_l_85.o,bn_mp_reduce_is_2k_l_85.o,bn_mp_reduce_2k_setup_l_85.o,bn_mp_radix_smap_85.o,bn_mp_read_radix_85.o,bn_mp_toradix_85.o,bn_mp_radix_size_85.o,bn_mp_fread_85.o,bn_mp_fwrite_85.o,bn_mp_cnt_lsb_85.o,bn_error_85.o,bn_mp_init_multi_85.o,bn_mp_clear_multi_85.o,bn_mp_exteuclid_85.o,bn_mp_toradix_n_85.o,bn_mp_prime_random_ex_85.o,bn_mp_get_int_85.o,bn_mp_sqrt_85.o,bn_mp_is_square_85.o,bn_mp_init_set_85.o,bn_mp_init_set_int_85.o,bn_mp_invmod_slow_85.o,bn_mp_prime_rabin_miller_trials_85.o,bn_mp_to_signed_bin_n_85.o,bn_mp_to_unsigned_bin_n_85.o,aes_86.o,bn_86.o,dh_86.o,dh-ltm_86.o,des_86.o,dsa_86.o,engine_86.o,md2_86.o,md4_86.o,md5_86.o,rsa_86.o,rsa-ltm_86.o,rc2_86.o,rc4_86.o,rijndael-alg-fst_86.o,rnd_keys_86.o,sha_86.o,sha256_86.o,sha512_86.o,ui_86.o,evp_86.o,evp-hcrypto_86.o,pkcs5_86.o,pkcs12_86.o,rand_86.o,rand-egd_86.o,rand-unix_86.o,rand-fortuna_86.o,rand-timer_86.o,hmac_86.o,camellia_86.o,camellia-ntt_86.o,common_86.o,validate_86.o -> libhcrypto-samba4.so} make: *** [all] Error 1
Abartlet, could you please look into this? It seems to be heimdal-specific.
Created attachment 7342 [details] Remove builtin-libraries option I think the reporter is confusing --builtin-libraries with --bundled-libraries. With the experience of hindsight, I think --builtin-libraries is a bad idea, so I've prepared a patch to remove it. Tridge will need to look at this.
Comment on attachment 7342 [details] Remove builtin-libraries option Oops, this is more complex than I imagined, and my patch was untested and wrong.
Please don't remove --builtin-libraries; it's very useful for e.g. building in libreplace or libccan when packaging. If it's really confusing, perhaps we can rename it to something else?
This bug was referenced in samba master: 14e7112734bbb31db99e394323ef2cb31385ebf7
Created attachment 17254 [details] patch for samba 4.16 Older versions have conflicts, so just 4.16 for now.
Pushed to autobuild-v4-16-test.
This bug was referenced in samba v4-16-test: 337301252b6c01e7cac10ca40164700514687645
Thanks! Reassigning to Andrew for older versions.
Closing. This comes up when folks package new releases so covering master and 4.16 will do.
The usage of f'''string''' breaks the build with PYTHON=python2.7 ./configure --without-ad-dc --disable-python for no good reason....
While I understand the frustration, particularly if this was working accidentally (but untested) until now, but Samba dropped Python 2.7 support for our build tooling in Sep 2020. This was discussed at: https://lists.samba.org/archive/samba-technical/2020-September/135751.html This actually is exactly why we have the rule that the environments we support are those we test, after may frustrations keeping Python 2.6 support working. So while this might be an easy fix, the principle matters, so I really don't want us walking back into supporting Python 2.7 again.
(In reply to Andrew Bartlett from comment #12) Yes, even without this 4.16 doesn't support python2 for the build. I just wanted to post the warnings, so that we don't backport something like this to older stable branches.
This bug was referenced in samba v4-16-stable (Release samba-4.16.1): 337301252b6c01e7cac10ca40164700514687645