Notes taken at the Binutils BoF Attendees: 92 Topic: Future of Gold * no longer maintained by Google * still popular? * Ben Woodwards noted that the National Labs are disappointed it is going away. Looking at LLD * Nathan Sidwell noted that Facebook is in the same boat * questions over whether LLD has the needed functionality * Peter Zijlstra note that gold isblacklisted by kernel * issues with linker script incompatibility * could unblacklist if the bugs were fixed * not enough people active to maintain it * problem is parallel linking * lots of cores today, but relatively slow * can we add parallelism to GNU ld * can elfutils add a linker * used to have one for x86 but rotted * do we need another linker project * Ian Lance-Taylor's blog series on gold noted that ld architecture was inside out * Nick Desaulniers noted Google has switched to LLD Topic: autogenerate ChangeLogs * GCC does this automatically, but you still have to write a ChangeLog in the commit message and checks you have all the right content * Nick Alcock noted the discipline that ChangeLog imposed was important, annoyance is in the administration * Nick Clifton noted that probably would need to coordinate with GDB * Jakub Jelinek noted you can make an exception for certain directories Topic: Closer ties with LLD * run each other's testsuite * feature and command line compatibility * Catherine Moore noted value of compatibility * Jakub Jelinek noted LLD is compatible, but ignores some options silently * Frank Ch Eigler noted that going overboard with compatibiltiy could stagnate both projects * shared mailing list? * supported, particularly for kernel * agreement to make it happen Other: Integration of CGEN in binutils * Frank Ch Eigler, only issue is copyright assignment. Some of the code predates the RedHat copyright assignment. * It's currently licensed GPLv3+, multiple (C) holders * Action: Nick will ask the FSF Other: Jose E Marchesi noted problem of duplicate gnulib tree within binutils-gdb * Pedro Alves noted was moved out of GDB with just the bits it needs * ideal if all the tools used the same gnulib tree * opportunity to move gnulib to the top level and move some compatibility