Before this patch, building talloc on darwin (OS X 10.8.5) results in:
gcc -dynamiclib -Wl,-search_paths_first -undefined error -o libtalloc.dylib.2.0.1 ./talloc.o ./libreplace/replace.o ./libreplace/snprintf.o ./libreplace/getpass.o ./libreplace/strptime.o -install_namelibtalloc.dylib.2
gcc: error: unrecognized command line option '-install_namelibtalloc.dylib.2'
make: *** [libtalloc.dylib.2.0.1] Error 1
Setting SONAMEFLAG to either "-install_name " (note trailing space), or
"-Wl,-install_name," results in a successful build. The latter seems
more clear.
...instead of just for i686-linux.
Without a --with-pcap= configure flag, the build system tries to
auto-detect the backend by poking in /dev, /usr/include etc... In the
pure nix world, this auto-detection fails. (Good, we don't want host
details to leak into the build.)
The end result of this (failed) auto-detection; no packet capture
possible:
$ sudo tcpdump -i wlp2s0
tcpdump: live packet capture not supported on this system
This fixes the above issue on 64-bit linux systems, and probably other
architectures supported by linux.
This likely exacerbates the non-determinism in ghc package ids, so until
that is fixed let's live with the slow builds.
This reverts commit 817c0e4144.
retroarchBare was too old for any practical use at this point, so it was removed in favor
of fetching releases from git. RetroArch often recommends checkout out the latest git master
if you are on a desktop platform, so we are conforming with their recommendation. I plan on
bumping the git of retroarch and each core every month, and I plan on adding more libretro
cores in the future.
The libretro cores that were added are nestopia, quicknes, snes9x, and vba-m
planned next are : bsnes-accuracy, bsnes-balanced, bsnes-performance, a few libretro mame ports
Closes#4448
nox-review should review against master, it's travis that should check out a hydra-available master commit to set the PR on. Perhaps we could rebase the current HEAD on the hydra-built commitish instead?
