On Sun, May 26, 2013 at 01:34:47PM +0200, Marc Espie wrote:
> I definitely vote against it.
> I really need those errors.
>
> You're only thinking of the cases where the error lies in the port, I also
> have to live with the cases where there is something fishy with the
> infrastructure.
>
> So that's a big NO, NO WAY from me.
I still think it's worth to get this in. Looking at my patched
bsd.port.mk file, the patch modifies line 1839 in below snippet and as
far as I understand that make file, code jumps from line 1839 to line
1821. I agreed that my patch can hide problem in line 1822 if mkdir
fails. However pkg_add execution in line 1824 has already stdout and
stderr redirected to /dev/null, so my change hides that mkdir execution.
Taking all above into account, I'm not sure to what infrastructure
problems are you referring to Marc. I'm not convinced my patch is
bad idea.
1819 .for _S in ${MULTI_PACKAGES}
1820
1821 ${_CACHE_REPO}/${_PKGFILE${_S}}:
1822 @mkdir -p ${@D}
1823 @${ECHO_MSG} -n "===> Looking for ${_PKGFILE${_S}} in
\$$PKG_PATH - "
1824 @if ${SETENV} ${_TERM_ENV} PKG_CACHE=${_CACHE_REPO}
PKG_PATH=${_CACHE_REPO}:${_PKG_REPO}:${PACKAGE_REPOSITORY}/${NO_ARCH}/:${PKG_PATH}
${_PKG_ADD} -n -q ${_PKG_ADD_FORCE} -D installed -D downgrade ${_PKGFILE${_S}}
>/dev/null 2>&1; then \
1825 ${ECHO_MSG} "found"; \
1826 exit 0; \
1827 else \
1828 ${ECHO_MSG} "not found"; \
1829 exit 1; \
1830 fi
1831
1832
1833 # The real package
1834
1835 ${_PACKAGE_COOKIE${_S}}:
1836 @mkdir -p ${@D} ${_TMP_REPO}
1837 . if ${FETCH_PACKAGES:L} == "yes" &&
!defined(_TRIED_FETCHING_${_PACKAGE_COOKIE${_S}})
1838 @f=${_CACHE_REPO}/${_PKGFILE${_S}}; \
1839 cd ${.CURDIR} && ${MAKE} $$f 2>/dev/null && \
1840 { ln $$f $@ 2>/dev/null || cp -p $$f $@ ; } || \
1841 cd ${.CURDIR} && ${MAKE}
_TRIED_FETCHING_${_PACKAGE_COOKIE${_S}}=Yes _internal-package-only
1842 . else
1843 @${_MAKE} ${_PACKAGE_COOKIE_DEPS}
1844 # What PACKAGE normally does:
--
best regards
q#
