Hi, clucene's API makes heavy use of the type float_t. On s390, float_t has historically been defined as double for no good reason. For getting rid of performance overhead in some cases and contradictions with the C standard in others, we discuss plans to clean up that definition - float_t should become float on s390.
As a result of that change, all these places in clucene's ABI would flip from double to float. Existing shared libs of clucene would become incompatible with binaries built with new versions of glibc/gcc and vice versa -- potentially causing very bad update experiences. To avoid that ABI breakage, I propose to stabilize the use of float_t to always use double on s390x. Please review my patch posted in the ticket https://sourceforge.net/p/clucene/bugs/233/ where I also posted more background on float_t and its status quo on s390. What do you think of this approach? What alternative may I have missed? (If you are also subscribed to the tickets and received this twice, please excuse the duplication.) Regards, Marius -- Marius Hillenbrand Linux on Z development IBM Deutschland Research & Development GmbH Vors. des Aufsichtsrats: Gregor Pillen / Geschäftsführung: Dirk Wittkopp Sitz der Gesellschaft: Böblingen / Registergericht: Amtsgericht Stuttgart, HRB 243294 _______________________________________________ CLucene-developers mailing list CLucene-developers@lists.sourceforge.net https://lists.sourceforge.net/lists/listinfo/clucene-developers
