ports and NO_STAGE: portmaster leaves port-system in corrupt state
ohartman at zedat.fu-berlin.de
Thu Oct 3 12:17:47 UTC 2013
When NO_STAGE=yes is missing in the port's Makefile, as it is for a
couple of ports like lang/perl5.16, then portmaster compiles, installs
and - corrupt the port-system, because it tries to "lstat" files it can
not find and ends up at the end of an unfinished installation. This
leaves the entry in the installed-port database corrupted. The port got
installed, but the database hasn't an entry anymore.
I found that the port system is in a very bad shape when one is using
CURRENT (10.0), where several things happened the same time like
converters/libiconv ahs gone as required port, libstdc++ has gone in
favour of libc++ and now ports that do not have this NO_STAGE= tag in
the toplevel Makefile. About the last piece - I miss a
warning/hint/information for those who has not the time following every
second informations on the mailing lists!
I regret that I forgot about three other ports I stumbled in where the
missing NO_STAGE=yes obviously solved the problem after I put it into
the Makefile - but that was simply a hunch - without knowing exactly
what I do. Again, I miss some informations about that and "googling"
didn't brought up deeper insight into that.
If someone would be so kind an d delegate me to a proper official
website where this NO_STAGE for the ports is explained a bit and
further if someone tells me what to do when I stumble into the next
port out of the 1200 I have to recompile, I would really appreciate
-------------- next part --------------
A non-text attachment was scrubbed...
Size: 488 bytes
Desc: not available
More information about the freebsd-ports