On Monday 22 September 2003 09:23 pm, Jamie Heilman wrote:
I didn't say anything about standards, but since you bring it up http://www.gnu.org/manual/tar/html_node/tar_117.html#SEC112
Okay. Touche. But I still find it hard to accept the idea that a program that can handle an arbitrarily long pathname should be considered the "wrong" implementation. That's obviously what you would intuitively expect out of an archiver program.
There is a note about tar pecularites in the FAQ document inside the tarball... which is fine I guess, but its evidently not enough. I think the real problem is that the symptoms of the problem tend vary, sometimes the breakage doesn't manifest as a checksum error, or it scrolls off the console too fast to see or something, so people invariably end up asking the list. Anyway, I sort of doubt the source will change, but it would avoid the problem, and it would make it easier on the newbies. Somebody told me this community was into that.
Hmm. Well, modifying it so that it would break more spectacularly would probably be good: One could imagine having a check for a file which happens to have a long pathname and raising an exception with a "Corrupted archive extraction -- please be sure you are using a Gnu compatible tar program" message. But the directory structure of a program is of practical documentary use to those who work on it, and even to people like me who simply use it for reference. I would find Zope unusable without the ability to find stuff in the source code, and if "AccessControl" suddenly became "AC" or something, purely for brevity, the code would become much less readable for me. So I'd be really unhappy if the code was actually modified to accomodate tar programs with path limits. Cheers, Terry -- Terry Hancock ( hancock at anansispaceworks.com ) Anansi Spaceworks http://www.anansispaceworks.com