On Tue, Oct 21, 2014 at 08:15:39AM -0400, Greg Troxel wrote:
I don't understand the charset.alias issue. It seems that there's
something going on with iconv or similar where packages choose to
install a charset.alias depending on something else.
The something else turned out to be interesting. From:
http://git.savannah.gnu.org/cgit/gnulib.git/tree/modules/localcharset#n45
if test $(GLIBC21) = no; then \
case '$(host_os)' in \
darwin[56]*) \
need_charset_alias=true ;; \
darwin* | cygwin* | mingw* | pw32* | cegcc*) \
need_charset_alias=false ;; \
*) \
need_charset_alias=true ;; \
esac ; \
else \
need_charset_alias=false ; \
We seem to have our old friend the hand coded list of operating system names.
There are two
approaches to deal with this. One is to make conditional PLIST entries
to register the file on platforms that create it. The other is to
understand and fix the root cause of the variance. I think we should
figure out and fix the issue. This is partly because I think it will be
easier in the long run, and partly becuase installation of files that
aren't about the package itself will lead to odd conflicts.
It appears that the current solution is to not install it by removing
it from the PLIST, as /usr/pkg/lib/charset.alias is listed in
VARNAME=CHECK_FILES_SKIP.
Cheers,
Patrick