[Bug 226841] sysutils/dupd: update to version 1.6

[bugzilla-noreply at freebsd.org]

https://bugs.freebsd.org/bugzilla/show_bug.cgi?id=226841

--- Comment #3 from Thomas Hurst <tom at hur.st> ---
> What is 'bfd' in this context?

GNU ld, basically.  It uses GNU's Binary File Descriptor library, so I guess
ld.bdf makes more sense than ld.gnu.  *shrug*.

> The GITHASH definition is harmless even if git not present. The value
> is only ever used in development builds, so because this port build is
> meant for release builds, it it irrelevant. It is safe to patch it out
> as before, but see above on minimizing patches.

OK.  I mainly patched it out because I don't like seeing spurious "command not
found" errors in my build logs - it's one more thing to remember to ignore when
testing a port and one more thing for users to get worried over ("why is this
giving me fatal git errors during build?!").

If the Makefile set it conditionally it'd be easy to disable just by passing in
a definition to override it.

> The -m64 flag is unnecessary. However, I don't know if it'll work
> correctly as a 32bit build. I have not tried that in many
> releases/years. Is that important for FreeBSD's needs?

If it only supports certain architectures it should be set ONLY_FOR_ARCHS=, but
I'd rather only do that if it's specifically known- broken.  Luckily there is a
test suite :)

Either way, it's blatting ports-provided CFLAGS, which is to be avoided:

  https://www.freebsd.org/doc/en/books/porters-handbook/book.html#dads-cflags

> As to the objcopy/ld line, I'll look into a way to make this more
> flexible in a way that meets the needs on all platforms. That'd be for
> next release, so I guess you can retain that patch for this time.

How about:

    system("man dupd");

;)

-- 
You are receiving this mail because:
You are the assignee for the bug.