[Bug-tar] Re^2: Verify failures and "obsolescent base-64 headers"

[Andreas Grosche]

# On Thu, 29 Jun 2006, Sergey Poznyakoff wrote:

> > no errors are detected when using tar-1.15.* -diff on the same archives.
>
> I don't see any problem, given that it has been fixed in 1.15 branch.

Neither do I NOW anymore, knowing that I have to use the latest branch. It did
take several people quite some time, though, to realise that there was nothing
broken on the first (affected) machine at all, but that the earlier tar itself
had an internal problem indeed.

> > This seems to indicate some severe bug in tar-1.[1]3.25, which is still
> > widely in use

That was referring to "tar (GNU tar) 1.13.25" of course (had a missing "1"
in the first post - sorry, this was copied from a photographed screenshot!).

> It should not be, given that the newer version is available.

> large file support [...] is enabled automatically starting at least from 1.14

Anyway you are right of course that the new version should be used - just the
worry was whether this had an explicit fix for a "surrounded" bug... or simply
other code behaving differently - which might show similar problems elsewhere:

> starting from which version, if any, can this issue be considered fixed?

If you know what caused the "obsolescent base-64 headers" and related errors
shown above in this thread, of course I'll know tar-1.15.* should not cause any
of the headaches tar-1.13.25 did. On the other hand, just in case it has not
been determined why they stopped occurring, is it safe to say they won't come
back? Given how widespread tar is, I was stunned that the problem did not seem
to have been found and diagnosed by many others as well, over all these years.