Inconsisten handling of charset.aliases breaking some packages

[Iain Morgan <Iain.Morgan%nasa.gov@localhost>]

Hello,

A few months back, I reported a problem with the nandling of
lib/charset.alias in math/octave on this mailing list. I eventually
suggested an admittedly kludgy patch to address that issue and a second
issue with octave on Linux. Subsequently, I've found another package,
www/elinks, which appears to have a similar issue with
lib/charset.alias.

In both cases, the PLIST file for the package unconditionally includes
charset.alias. However, the build process does not install this file
on all platforms; which causes pkg_create to fail.

I noticed in another package, net/lftp, that the reference to
charset.alias only appears as a comment in the PLIST file;
indicating that the file belongs to libiconv. So, could we take a
similar approach for other packages -- or just remove the reference to
charset.alias in packages other than libiconv?
release and current wip is as follows:

Looking over the PLIST* files, I find the following:

        converters/libiconv/PLIST:share/examples/libiconv/charset.alias
        math/octave/PLIST:lib/charset.alias
        sysutils/libvirt/PLIST:lib/charset.alias
        emulators/mednafen/PLIST:@comment belongs to libiconv: lib/charset.alias
        www/elinks/PLIST:share/elinks/locale/charset.alias
        wip/elinks-snapshot/PLIST:@comment    belongs to libiconv: 
lib/charset.alias
        wip/nios2-gdb/PLIST:lib/charset.alias
        wip/elinks-libmm/PLIST:share/elinks/locale/charset.alias
        wip/ccrypt/PLIST:lib/charset.alias
        wip/nightfall/PLIST:lib/charset.alias
        wip/gdb/PLIST.common:lib/charset.alias
        net/lftp/PLIST:@comment belongs to libiconv: lib/charset.alias

As you can see, we don't handle the file in a consisten manner.

Thanks,

-- 
Iain Morgan