[Date Prev][Date Next][Thread Prev][Thread Next][Date Index][Thread Index]

Re: Features.



On Tue, 22 Aug 2000, Todd T. Fries wrote:
> In x11/kde/games/Makefile:
> 
>     MULTI_PACKAGES=-kabalone -kasteroids -kblackbox -kmahjongg -kmines\
>     -konquest -kpat -kpoker -kreversi -ksame -kshisen -ksirtet -ksmiletris \
>     -ksnake -ksokoban
> 
> In infrastructure/mk/bsd.port.mk:
> 
>     .if defined(MULTI_PACKAGES) && empty(SUBPACKAGE)
>     .  for _sub in ${MULTI_PACKAGES}
> 
>     _SUBPACKAGE_COOKIES+= ${_PACKAGE_COOKIE}${_sub}
>     .    if ${FAKE:U} == "YES"
>     ${_PACKAGE_COOKIE}${_sub}: ${_FAKE_COOKIE}
>     .    else
>     ${_PACKAGE_COOKIE}${_sub}: ${_INSTALL_COOKIE}
>     .    endif
>             @cd ${.CURDIR} && exec ${MAKE} package SUBPACKAGE='${_sub}' \
> 		FLAVOR='${FLAVOR}'
> 
>     .  endfor
>     .endif
> 
> I'm sure this is not all there is to it but it is the obvious pieces that
> are the basic magic behind the scenes.  Other pieces from other parts of
> the referenced files might be used as well.  In general, I'd look for
> SUBPACKAGE/MULTI_PACKAGE/FAKE in bsd.port.mk and in Makefile's ..

Ah cool - thanks for pointing me at the right bits! (I was really misled
by the MAKE_xxx variables in egcs X-)


> etc .. I know there are other things in the queue that would simlify some
> of the pkg_* tools .. like libpax allowing for no temporary extraction area
> and such .. again, not sure of the status.

Won't this prevent rollback functionality if it's found that the pkg
cannot be installed at a late stage?


> As I was writing this I thought about another feature OpenBSD has recently
> introduced as of the 2.7 release .. which is /usr/local for all x based
> packages.  This requires a modification to the config/cf files similar
> to NetBSD .. we just make it a part of x itself instead of a package to
> add to x .. since x is not a package itself for us .. but similar concepts,
> different implementations.  At the same time we made some modifications to
> the cf files so that if a -Dports were passed to xmkmf, it would propogate
> correctly (also requires an enhancement to xmkmf) to allow for doing a fake
> install of an x app that is targeted at a different DESTDIR than what the
> package is finally installed to.  This allowed most x ports to be fake'ed
> quite easily.

IC - couldn't the same be achived a bit easier by not hardcoding
/usr/local in patch-xfree-4.0, but using $DESTDIR/$LOCALBASE instead?


 - Hubert

-- 
NetBSD - because Unix isn't just #include <linux.h>, i386, ILP32, ELF, ...!


To unsubscribe: send mail to <majordomo@unixathome.org>
with "unsubscribe bsdports" in the body of the message