Bug#887744: redeclipse FTBFS with glibc 2.26
Source: redeclipse Version: 1.5.8-1 Severity: serious Tags: buster sid https://tests.reproducible-builds.org/debian/rb-pkg/unstable/amd64/redeclipse.html ... In file included from /usr/include/math.h:724:0, from /usr/include/c++/7/cmath:45, from /usr/include/c++/7/math.h:36, from shared/cube.h:13, from shared/crypto.cpp:1: /usr/include/x86_64-linux-gnu/bits/math-finite.h: In function 'double tgamma(double)': /usr/include/x86_64-linux-gnu/bits/math-finite.h:203:21: error: 'gamma_r_finite' was not declared in this scope _Mdouble_ __res = __REDIRTO (gamma, _r, _MSUF_) (__d, &__local_signgam); ^ /usr/include/x86_64-linux-gnu/bits/math-finite.h:203:21: note: suggested alternative: '__gamma_r_finite'
Bug#887744: redeclipse FTBFS with glibc 2.26
On Fri, 19 Jan 2018 17:58:06 +0200 Adrian Bunk wrote: > Source: redeclipse > ... > In file included from /usr/include/math.h:724:0, > from /usr/include/c++/7/cmath:45, > from /usr/include/c++/7/math.h:36, > from shared/cube.h:13, > from shared/crypto.cpp:1: > /usr/include/x86_64-linux-gnu/bits/math-finite.h: In function 'double > tgamma(double)': > /usr/include/x86_64-linux-gnu/bits/math-finite.h:203:21: error: > 'gamma_r_finite' was not declared in this scope >_Mdouble_ __res = __REDIRTO (gamma, _r, _MSUF_) (__d, &__local_signgam); > ^ > /usr/include/x86_64-linux-gnu/bits/math-finite.h:203:21: note: suggested > alternative: '__gamma_r_finite' The commit "remove gamma name hack" might be related, apparently there was some preprocessor hackery going on. https://github.com/red-eclipse/base/commit/b16b4963c1ad81bb9ef784bc4913a4c8ab5f1bb4 Regards, Juhani
Bug#887744: redeclipse FTBFS with glibc 2.26
I had a similar error compiling lamiae [1] and I solved it by removing the -ffast-math flag from src/makefile I don't know if it would work for you, but the code base is the same, so I assume it might be that. Greetings, Miry [1] https://github.com/Hirato/lamiae
Bug#887744: redeclipse FTBFS with glibc 2.26
On Tue, 2018-01-23 at 13:46 +0200, Juhani Numminen wrote: > On Fri, 19 Jan 2018 17:58:06 +0200 Adrian Bunk > wrote: > > Source: redeclipse > > ... > > In file included from /usr/include/math.h:724:0, > > from /usr/include/c++/7/cmath:45, > > from /usr/include/c++/7/math.h:36, > > from shared/cube.h:13, > > from shared/crypto.cpp:1: > > /usr/include/x86_64-linux-gnu/bits/math-finite.h: In function > > 'double tgamma(double)': > > /usr/include/x86_64-linux-gnu/bits/math-finite.h:203:21: error: > > 'gamma_r_finite' was not declared in this scope > > _Mdouble_ __res = __REDIRTO (gamma, _r, _MSUF_) (__d, > > &__local_signgam); > > ^ > > /usr/include/x86_64-linux-gnu/bits/math-finite.h:203:21: note: > > suggested alternative: '__gamma_r_finite' > > The commit "remove gamma name hack" might be related, apparently > there was some > preprocessor hackery going on. > https://github.com/red-eclipse/base/commit/b16b4963c1ad81bb9ef784bc49 > 13a4c8ab5f1bb4 Yep, applying this commit solves the issue. Previously 'gamma' from math.h was masked in order to make it available to be defined by cubescript, and this masking was dependent on implementation-specifics which had changed. With this fix cubescript uses a different associated global name 'reqgamma' and thus does not need the masking. -- Martin Erik Werner
