Date: Sun, 4 Nov 2012 11:31:46 +0000 From: Chris Rees <crees@FreeBSD.org> To: Thomas Mueller <mueller23@insightbb.com> Cc: David Naylor <naylor.b.david@gmail.com>, FreeBSD Mailing List <freebsd-ports@freebsd.org> Subject: Re: wine/i386 for FreeBSD/amd64 port (aka wine-fbsd64) Message-ID: <CADLo83-RLh2e8-Q__QTJLDc2nu_XP=qSS%2BmKFaK3p-wS3a4Y=g@mail.gmail.com> In-Reply-To: <5D.CB.23131.5AE46905@smtp02.insight.synacor.com> References: <5D.CB.23131.5AE46905@smtp02.insight.synacor.com>
next in thread | previous in thread | raw e-mail | index | archive | help
On 4 November 2012 11:16, Thomas Mueller <mueller23@insightbb.com> wrote: > from David Naylor <naylor.b.david@gmail.com>: > >> Hi List, > >> # Executive Summary > >> Over the past years I have been maintaining the wine-fbsd64 port (see >> http://mediafire.com/wine_fbsd64 for more). The port itself effectively does >> static linking (it bundles all the libraries wine needs) with scripts to >> bootstrap the environment to easily use wine from FreeBSD/amd64. There is >> also a script to install the i386 nVidia graphic drivers so that wine has >> access to nVidia accelerated graphics from FreeBSD/amd64. > >> I would like to propose this port gets included in the port's collection and >> would like to get feedback, your comments please :-). > >> P.S. I'm not subscribed to the list, so please ensure I'm cc'ed in the >> discussion. > >> # Details of the Port > >> Please see attached for the actual port. > >> ## Port Preamble > >> This port is a slave port to emulators/wine(-devel). The master port needed >> to be modified (already done): >> - to conditionally set USE_LDCONFIG (if USE_LDCONFIG32 was not set) >> - to allow the library directory to be changed (see WINELIBDIR) >> - to allow configure arguments to be appended > >> ## Port Targets > >> The port itself does the following in the preamble: >> - specifies the pkg(de)install script to handle nVidia driver patching >> - overrides ACTUAL-PACKAGE-DEPENDS (all depends are bundled with the port) >> - defined the library directory to ${PREFIX}/lib32 >> - defined the binary directory to ${PREFIX}/bin32 >> - patches the PLIST to refer to lib32 (not lib) >> - defined USE_LDCONFIG32 appropriately > >> The post-install-script target: >> - Installs the files/binbounce file in ${PREFIX}/bin for each ${PREFIX}/bin32 >> file (hard linked) >> - Finds all linked library, copies them to ${PREFIX}/lib32, and added them to >> the plist >> - Finds all dlopen'ed libraries, copies them to ${PREFIX}/lib32, and added >> them to the plist >> - Installs the nVidia patch file >> - Run the (PRE-|POST-)INSTALL script > >> The post-package-script (run only if WITH_PKGNG is defined): >> - Amends the package so the arch label to 64bit > >> ## Port scripts (in files/) > >> The binbounce file does the following to transparently fix the environment to >> allow seamless running of the wine programs: >> - determines the location of the TARGET (follows symbolic links to itself) >> - fixes LD_LIBRARY_PATH if in an i386 environment (so lib32, lib32/wine is >> found) >> - fixes LD_32_LIBRARY_PATH if in an amd64 environment (so lib32, lib32/wine, >> /usr/lib32) >> - fixes PATH (so bin32 is found) >> - passes execution to the counterpart in bin32 > >> The patch-nvidia.sh file does the following: >> - Downloads the nVidia distfile for i386 (iff nVidia amd64 driver is >> installed) >> - Installs the required libraries into ${PREFIX}/lib32 >> - When run from the install script it does _not_ download the distfile, only >> installs the libraries iff the distfiles are already downloaded. >> # Shortcomings of the port > >> The following are shortcomings that I am aware of: >> - Can only be compiled in an i386 environment, but the resulting package is >> *intended* for amd64 (although works fine in an i386 environment) >> - If, somehow, there is a recursive calling of wine programs then >> LD_(32_)LIBRARY_PATH and PATH will continue to grow with every iteration. >> - The pkgng ports cannot be installed in an i386 environment as they are >> labelled for amd64. > >> # Testing > >> The ports published on mediafire have been tested by many users. The port >> itself works flawlessly however there have been some reports about some flaws >> in the 32-bit compatibility layer of the kernel (although I cannot remember >> the specifics now). > >> To produce the package on an amd64 system do the following: >> # (cd /usr/ports/emulators/; patch -p0 < /path/to/diff) >> # make -C /usr/src world DESTDIR=/i386 TARGET=i386 >> # mount -t devfs devfs /i386/dev >> # mkdir /i386/usr/ports >> # mount -t nullfs /usr/ports /i386/usr/ports >> # chroot make -C /usr/ports/emulators/wine-fbsd64 package WITH_PKGNG=yes > >> The package wine-fbsd64-1.5.16,1.txz (in pkgng format) will be available from >> /usr/ports/packages/All/ > >> # Conclusion > >> "It is based completely off the main port and uses the hack to, >> effectively, use static linking (or bundling of libraries). In a >> sense it is a complete, yet quite stable and encompassing, hack. " >> - David ;-) > > It would be nice to have wine-fbsd64 as a port, but that might unfortunately > deprive the user of certain flexibility. > > Also, nVidia support should be an option, since users with other graphics > cards might have no use for it. > > I would really prefer to build the i386 FreeBSD system as a separate part, including kernel, > since some users, myself included, might want to run an actual FreeBSD i386, > especially on an older computer. So one could build this FreeBSD i386 on a > USB stick or USB hard drive, and then be able to run wine on an i386 system. > > Would wine-fbsd64 be a separate port, or would it be wine built on i386, as > the page http://wiki.freebsd.org/Wine suggests? It would be nice to be able > to run Wine on i386 as well as amd64. It would be a separate port, as he has explained :) I think this is very interesting... but I'm not 100% convinced the best place for this is in the ports tree. However, it would improve visibility for it, with a good IGNORE message. We have a problem however; we can't include bsd.port.pre.mk in a slave port. The solution I can think of is; post-package-script: if [ "${PKG_BIN:T}" = "pkg" ]; then \ ${XZ_CMD} -dc ${PKGFILE} | \ ${SED} -e "s/^\(arch: freebsd:.*:x86\):32/\1:64/" | \ ${XZ_CMD} > ${WRKDIR}/${PKGNAME}.txz; \ ${MV} ${WRKDIR}/${PKGNAME}.txz ${PKGFILE}; \ fi Chris
Want to link to this message? Use this URL: <https://mail-archive.FreeBSD.org/cgi/mid.cgi?CADLo83-RLh2e8-Q__QTJLDc2nu_XP=qSS%2BmKFaK3p-wS3a4Y=g>