Bug 281391

Congratulations! You are only the 12th person to ever report this bug. That qualifies you as a bona
fide rpm hacker!

Meanwhile, don't do that is the answer. Ther better answer is ripping support for rpmbuild -ta foo.tar.gz,
but noone wants to hear that answer.

Jeff: If that sentence above the brilliant workaround is your usual response to
bug reports I finally understand the poor reputation you have.

hunh ... and who fed rpmbuild garbage, ignored the return code, and then complained?

Here I thought Unix like OS's provided power to users to push limits with new
combinations of approaches, rather than holding their hands, having guard rails,
and rolling over and dying.  I guess saying that an 'error' was detected **
seven times **, and returning a non-zero return code was not enough.

[herrold@couch ~]$ rpmbuild -ta /dev/null || echo "return code was not zero"
error: File /dev/null is smaller than 4 bytes
error: Name field must be present in package: (main package)
error: Version field must be present in package: (main package)
error: Release field must be present in package: (main package)
error: Summary field must be present in package: (main package)
error: Group field must be present in package: (main package)
error: License field must be present in package: (main package)
return code was not zero
[herrold@couch ~]$

I wonder what will happen if I feed a chicken to rpmbuild ...

An error was detected alright, but creating files with weird names ('tar:
*.spec: Not found in archive') as a side-effect is ugly if nothing else.

Fixed upstream (finally).