The stable ebuild for simh is 3.3.0 and the latest unstable ebuild is 3.8.1. There are some things that needs 3.9-0 and that's available from http://simh.trailing-edge.com/sources/simhv390.zip. Is it possible for a new unstable ebuild for 3.9.0 to be put into portage? Thanks!
CORRECTION: http://simh.trailing-edge.com/sources/simhv39-0.zip
Created attachment 311919 [details] simh 3.9.0 ebuild
Hello Alex, can you please attach the mentioned ${P}-nolto.patch file, we can establish a proxy maintainer thingy for some packages, if you like. Michael
Created attachment 312241 [details] Removes -flto option, not used with GCC until 4.6.x
(In reply to comment #3) > can you please attach the mentioned ${P}-nolto.patch file, > > we can establish a proxy maintainer thingy for some packages, if you like. That would be nice, I have several emulator packages in my overlay. I did try to bec ome a developer last year but found that I had no time to be one as it seems to be a full time thing :-(
(In reply to comment #4) > Created attachment 312241 [details] > Removes -flto option, not used with GCC until 4.6.x I've added an toolchain-funcs.eclass:gcc-{major,minor}-version test to decide whether or not to set NO_LTO=1. I'd to fix an race condition with MAKEOPTS="-j2" (mkdir BIN failed). https://github.com/simh/simh/issues/3 CFLAGS/LDFLAGS should work, too. https://github.com/simh/simh/issues/4 +*simh-3.9.0 (19 May 2012) + + 19 May 2012; Michael Weber <xmw@gentoo.org> +simh-3.9.0.ebuild, + +files/simh-3.9.0-fix-mkdir-race.patch, + +files/simh-3.9.0-respect-FLAGS.patch, metadata.xml: + Version bump and proxy-maint (bug 416173, thanks Alex Buell), EAPI-4, respect + LDFLAGS, ebuild cleanups. +
Reasons I changed the ebuild file (i.e. did not use the exact version atached) 1) The old one was EAPI-0 w/o src_prepare, so the src_unpack had the mkdir/unpack ${A} lines. This is no longer needed using EAPI-1 or -2, which have src_prepare for that. 1b) I decided to use EAPI-4 do spare some `|| die` expressions. 2) The build did not respect CFLAGS and LDFLAGS, which is bad for documented reasons, so I wrote/applied/sent-upstream an corresponding patch. 3) Parallel make failed with a certain probability (mkdir race condition) -> patch. 4) gcc-4.6 and -4.7 are in the tree and will appear on the horizon, so I added an conitional NO_LTO=1 to be only applied for gcc < 4.6. 5) for sake of readability and less lines in src_install, I dropped the `cd "${S}"/BIN` and `cd ${S}` lines around `for BINFILE ... done`, and - mea cupla - I forgot the $(basename . ). So, I accept my responsibility for the missing binaries (bug 423069), but I do not accept any accusations for "not just committing the attached ebuild". There is an difference between "works OK" and passing qa standards (FLAGS, parallel make) and being future proof (testing gcc, maintainability). Regards, Michael Weber
It was just a quick and dirty ebuild put together from the older versions. I think that newer versions will follow your ebuild next time. Thanks for putting it into the tree.
