Hi again, hope I'm not boring you ;) In don't claim that GNU tar is worse than other TAR implementations how to handle USTAR, but the state of testing USTAR compatibility between TAR implementations is a bit sad.
I wrote a quick Perl snippet to run my tests again, now also including "CMake" built in TAR. I also ran "tar -tf" for each package produced (if return code was 0) and grepped for the original path in the output. A bit simplistic test, I know, but the randomness in the result is a bit shocking, at least to me. For CMake TAR I note that they happily return the code 0 even if things go wrong. Also a bit "funny" is that they sometimes cut the path by few characters (even when not needed) and store the file anyway. As the path is split between the "prefix" and "name" fields, both parts might be cut by a few characters. Result and test script is attached, kent
TARERR.gz
Description: Binary data
ustartest
Description: Binary data
-- Kent Boortz, Senior Production Engineer Sun Microsystems Inc., the MySQL team Office: +46 863 11 363 Mobile: +46 70 279 11 71
